From 236f9462921e29dfd64afe7f004518932d816dbb Mon Sep 17 00:00:00 2001
From: Chao Liu <chao.liu2@amd.com>
Date: Fri, 29 Jul 2022 18:19:25 -0500
Subject: [PATCH] Clean up conv example, Instances, profiler and test (#324)

* convnd_fwd fp16 example

* update example

* update example

* update instance

* updating refernce conv

* update reference conv

* update conv fwd profiler

* update conv 1d and 3d instance

* update include path

* clean

* update profiler for conv bwd data and weight

* update conv bwd weight

* clean

* update conv example

* update profiler for conv bwd weight

* update ckprofiler for conv bwd data

* fix reference conv bwd data bug; update conv bwd data test

* update examples

* fix initialization issue

* update test for conv fwd

* clean

* clean

* remove test case too sensitive to error threshhold

* fix test

* clean

* fix build

* adding conv multiple d

* adding conv multiple D

* add matrix padder

* add gemm padding to convnd

* adding group conv

* update gemm multi-d

* refactor

* refactor

* refactor

* clean

* clean

* refactor

* refactor

* reorg

* add ds

* add bias

* clean

* add G

* adding group

* adding group

* adding group

* update Tensor

* clean

* update example

* update DeviceGemmMultipleD_Xdl_CShuffle

* update conv bwd-data and bwd-weight

* upate contraction example

* update gemm and batch gemm with e permute

* fix example build

* instance for grouped conv1d

* update example

* adding group conv instance

* update gemm bilinear instance

* update gemm+add+add+fastgelu instance

* update profiler

* update profiler

* update test

* update test and client example

* clean

* add grouped conv into profiler

* update profiler

* clean

* add test grouped conv, update all conv test to gtest

* update test

[ROCm/composable_kernel commit: 500fa9951297c033a9c4c1d300b03895a46528d2]
---
 README.md                                     |    3 +-
 .../gemm_add_add_fastgelu.cpp                 |   20 +-
 example/01_gemm/gemm_dl_fp16.cpp              |   12 +-
 example/01_gemm/gemm_dl_fp32.cpp              |   12 +-
 example/01_gemm/gemm_dl_int8.cpp              |   12 +-
 example/01_gemm/gemm_xdl_bf16.cpp             |   12 +-
 example/01_gemm/gemm_xdl_fp16.cpp             |   12 +-
 example/01_gemm/gemm_xdl_fp64.cpp             |   12 +-
 example/01_gemm/gemm_xdl_int8.cpp             |   12 +-
 .../gemm_bilinear_xdl_fp16.cpp                |   37 +-
 .../gemm_bias_relu_xdl_fp16.cpp               |   23 +-
 .../gemm_add_add_fastgelu_xdl_fp16.cpp        |   27 +-
 .../06_conv2d_fwd_bias_relu/CMakeLists.txt    |    2 -
 example/06_conv2d_fwd_bias_relu/README.md     |   22 -
 .../conv2d_fwd_xdl_bias_relu.cpp              |  313 ---
 .../CMakeLists.txt                            |    3 -
 example/07_conv2d_fwd_bias_relu_add/README.md |   24 -
 .../conv2d_fwd_xdl_bias_relu_add.cpp          |  328 ---
 example/09_convnd_fwd/CMakeLists.txt          |    7 +-
 example/09_convnd_fwd/convnd_fwd_common.hpp   |  173 ++
 example/09_convnd_fwd/convnd_fwd_xdl_bf16.cpp |  227 +++
 example/09_convnd_fwd/convnd_fwd_xdl_fp16.cpp |  481 ++---
 example/09_convnd_fwd/convnd_fwd_xdl_fp32.cpp |  483 ++---
 example/09_convnd_fwd/convnd_fwd_xdl_fp64.cpp |  490 ++---
 example/09_convnd_fwd/convnd_fwd_xdl_int8.cpp |  481 ++---
 example/10_conv2d_bwd_data/CMakeLists.txt     |    2 -
 example/10_conv2d_bwd_data/README.md          |   47 -
 .../conv2d_bwd_data_xdl.cpp                   |  259 ---
 example/11_conv2d_bwd_weight/CMakeLists.txt   |    2 -
 example/11_conv2d_bwd_weight/README.md        |   25 -
 .../conv2d_bwd_weight_xdl.cpp                 |  300 ---
 example/12_reduce/reduce_blockwise.cpp        |   16 +-
 .../12_reduce/reduce_blockwise_two_call.cpp   |   18 +-
 example/13_pool2d_fwd/pool2d_fwd_common.hpp   |   13 +-
 .../gemm_xdl_requant_relu_requant_int8.cpp    |   12 +-
 .../15_grouped_gemm/grouped_gemm_xdl_fp16.cpp |   35 +-
 .../gemm_reduce_xdl_max_fp16.cpp              |   14 +-
 .../gemm_reduce_xdl_mean_squaremean_fp16.cpp  |   16 +-
 example/17_convnd_bwd_data/CMakeLists.txt     |    2 +
 .../README.md                                 |    0
 .../convnd_bwd_data_common.hpp                |  149 ++
 .../convnd_bwd_data_xdl_fp16.cpp              |  207 ++
 example/17_convnd_bwd_data_xdl/CMakeLists.txt |    2 -
 .../convnd_bwd_data_xdl.cpp                   |  352 ----
 .../batched_gemm_reduce_xdl_fp16.cpp          |   16 +-
 .../broadcast_add_2d_amn_bn.cpp               |   12 +-
 .../broadcast_add_3d_am_bmnk.cpp              |   12 +-
 .../elementwise_add_1d.cpp                    |   12 +-
 .../elementwise_add_4d.cpp                    |   12 +-
 example/20_convnd_bwd_weight/CMakeLists.txt   |    5 +
 .../convnd_bwd_weight_common.hpp              |  152 ++
 .../convnd_bwd_weight_xdl_bf16.cpp            |  219 ++
 .../convnd_bwd_weight_xdl_fp16.cpp            |  216 ++
 .../20_convnd_bwd_weight_xdl/CMakeLists.txt   |    4 -
 .../convnd_bwd_weight_xdl.cpp                 |  385 ----
 .../convnd_bwd_weight_xdl_bf16_splitk.cpp     |  427 ----
 .../gemm_bias_relu_add_layernorm_xdl_fp16.cpp |   27 +-
 .../gemm_layernorm_xdl_fp16.cpp               |   23 +-
 .../gemm_xdl_layernorm_single_kernel_fp16.cpp |   20 +-
 example/22_cgemm/cgemm_xdl_fp16.cpp           |   18 +-
 example/23_softmax/softmax_blockwise.cpp      |   10 +-
 .../24_batched_gemm_c_permute/CMakeLists.txt  |    2 -
 .../24_batched_gemm_e_permute/CMakeLists.txt  |    2 +
 .../batched_gemm_e_permute_xdl_fp16.cpp}      |   61 +-
 example/25_gemm_bias_c_permute/CMakeLists.txt |    1 -
 example/25_gemm_bias_e_permute/CMakeLists.txt |    1 +
 .../gemm_bias_e_permute_xdl_fp16.cpp}         |   20 +-
 .../contraction_bilinear_xdl_fp32.cpp         |   14 +-
 .../contraction_scale_xdl_fp32.cpp            |   18 +-
 example/27_layernorm/layernorm_blockwise.cpp  |   16 +-
 .../grouped_gemm_bias_xdl_fp16.cpp            |   66 +-
 .../batched_gemm_bias_xdl_fp16.cpp            |   38 +-
 .../batched_gemm_xdl_fp16.cpp                 |   33 +-
 .../CMakeLists.txt                            |    2 +
 .../30_grouped_convnd_fwd_bias_relu/README.md |   28 +
 .../grouped_convnd_fwd_bias_common.hpp        |  192 ++
 .../grouped_convnd_fwd_bias_relu_xdl_fp16.cpp |  370 ++++
 example/CMakeLists.txt                        |   19 +-
 include/ck/ck.hpp                             |    2 +-
 .../device_prop.hpp                           |    0
 .../hip_check_error.hpp                       |    0
 include/ck/host_utility/io.hpp                |   41 +
 .../kernel_launch.hpp                         |    2 +-
 ...nvolution_backward_data_specialization.hpp |   16 +-
 ...olution_backward_weight_specialization.hpp |   13 +
 .../convolution_forward_specialization.hpp    |    4 +-
 .../gpu/device/device_5ary_elementwise.hpp    |    4 +-
 .../device_batched_gemm_c_permute_xdl.hpp     |    4 +-
 ....hpp => device_batched_gemm_e_permute.hpp} |   30 +-
 .../device_batched_gemm_e_permute_xdl.hpp     |  682 +++++++
 .../device/device_batched_gemm_multi_d.hpp    |   33 +-
 .../device_batched_gemm_multi_d_xdl.hpp       |  553 ++---
 ...evice_batched_gemm_reduce_xdl_cshuffle.hpp |    4 +-
 .../gpu/device/device_batched_gemm_xdl.hpp    |    4 +-
 .../gpu/device/device_binary_elementwise.hpp  |    4 +-
 .../device_cgemm_4gemm_xdl_cshuffle.hpp       |    4 +-
 .../device/device_contraction_multiple_d.hpp  |   16 +-
 ...ce_contraction_multiple_d_xdl_cshuffle.hpp |  452 ++--
 ...rd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp |   15 +-
 ...ice_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp |   13 +-
 ...fle_bias_activation_add_nhwc_kyxc_nhwk.hpp |    4 +-
 ...shuffle_bias_activation_nhwc_kyxc_nhwk.hpp |    4 +-
 ...onv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp |   15 +-
 .../device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp  |   15 +-
 .../gpu/device/device_conv_bwd_data.hpp       |   17 +-
 ..._weight.hpp => device_conv_bwd_weight.hpp} |   16 +-
 .../gpu/device/device_conv_fwd.hpp            |   16 +-
 ...evice_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp} |   71 +-
 ...d_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp} |  493 +----
 .../device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp  | 1046 ----------
 .../gpu/device/device_gemm.hpp                |   19 -
 ...vice_gemm_bias_add_reduce_xdl_cshuffle.hpp |    4 +-
 ...ute.hpp => device_gemm_bias_e_permute.hpp} |    6 -
 ...hpp => device_gemm_bias_e_permute_xdl.hpp} |  357 +---
 .../gpu/device/device_gemm_dl.hpp             |    4 +-
 .../gpu/device/device_gemm_multiple_d.hpp     |   24 +-
 .../device_gemm_multiple_d_xdl_cshuffle.hpp   |  399 ++--
 .../device_gemm_reduce_xdl_cshuffle.hpp       |    4 +-
 .../gpu/device/device_gemm_xdl.hpp            |    4 +-
 .../gpu/device/device_gemm_xdl_cshuffle.hpp   |    4 +-
 .../device_gemm_xdl_layernorm_cshuffle.hpp    |    4 +-
 .../gpu/device/device_gemm_xdl_splitk.hpp     |    4 +-
 .../device_gemm_xdl_splitk_c_shuffle.hpp      |    4 +-
 .../device_grouped_conv_fwd_multiple_d.hpp    |   63 +
 ...ouped_conv_fwd_multiple_d_xdl_cshuffle.hpp | 1813 +++++++++++++++++
 .../gpu/device/device_grouped_gemm.hpp        |   26 +-
 .../gpu/device/device_grouped_gemm_xdl.hpp    |  513 ++---
 .../gpu/device/device_layernorm.hpp           |    4 +-
 .../device/device_pool2d_fwd_nhwc_nhwc.hpp    |    4 +-
 .../gpu/device/device_reduce_multiblock.hpp   |    4 +-
 .../gpu/device/device_reduce_threadwise.hpp   |    4 +-
 .../gpu/device/device_softmax.hpp             |    4 +-
 .../gpu/device/device_unary_elementwise.hpp   |    4 +-
 .../gpu/device/matrix_padder.hpp              |  184 ++
 .../gpu/device/tensor_layout.hpp              |  319 ++-
 .../element/binary_element_wise_operation.hpp |   25 +-
 .../element/unary_element_wise_operation.hpp  |   65 +-
 .../gridwise_gemm_multiple_d_xdl_cshuffle.hpp |  286 ++-
 include/ck/utility/tuple.hpp                  |   28 +-
 library/CMakeLists.txt                        |    1 -
 .../cpu/reference_batched_gemm.hpp            |    2 +-
 .../cpu/reference_cgemm.hpp                   |    2 +-
 .../cpu/reference_conv_bwd_data.hpp           |  235 ++-
 ...ight.hpp => reference_conv_bwd_weight.hpp} |  182 +-
 .../cpu/reference_conv_fwd.hpp                |  188 +-
 .../reference_conv_fwd_bias_activation.hpp    |    2 +-
 ...reference_conv_fwd_bias_activation_add.hpp |    2 +-
 .../cpu/reference_gemm.hpp                    |    2 +-
 .../cpu/reference_gemm_bias_2d.hpp            |    2 +-
 .../cpu/reference_gemm_bias_activation.hpp    |    2 +-
 .../reference_gemm_bias_activation_add.hpp    |    2 +-
 .../cpu/reference_layernorm.hpp               |    4 +-
 .../cpu/reference_softmax.hpp                 |    4 +-
 .../device_operation_instance_factory.hpp     |   55 +-
 .../gpu/contraction_bilinear.hpp              |    8 +-
 .../gpu/contraction_scale.hpp                 |    8 +-
 .../gpu/convolution_backward_data.hpp         |  270 +++
 .../gpu/convolution_backward_weight.hpp       |  230 +++
 .../gpu/convolution_forward.hpp               |  128 ++
 .../gpu/gemm_add_add_fastgelu.hpp             |   50 +-
 .../gpu/gemm_bilinear.hpp                     |   49 +-
 .../gpu/grouped_convolution_forward.hpp       |  352 ++++
 .../gpu/grouped_gemm.hpp                      |   35 +-
 .../include/ck/library/utility/check_err.hpp  |    9 +-
 .../{host_tensor => utility}/conv_common.hpp  |    0
 .../include/ck/library/utility/conv_util.hpp  |  574 ------
 ...volution_host_tensor_descriptor_helper.hpp |  354 ++++
 .../library/utility/convolution_parameter.hpp |   86 +
 .../device_memory.hpp                         |    0
 .../host_common_util.hpp                      |    0
 .../{host_tensor => utility}/host_conv.hpp    |    0
 .../{host_tensor => utility}/host_gemm.hpp    |    0
 .../host_reduction.hpp                        |    4 +-
 .../{host_tensor => utility}/host_tensor.hpp  |  104 +-
 .../host_tensor_generator.hpp                 |    0
 .../ck/library/utility/op_instance_engine.hpp |   10 +-
 library/src/host_tensor/CMakeLists.txt        |   32 -
 .../gpu/CMakeLists.txt                        |   33 +-
 ..._shuffle_f32_f32_f32_f32_kknn_instance.cpp |   30 +-
 ..._shuffle_f32_f32_f32_f32_knnn_instance.cpp |   36 +-
 ..._shuffle_f32_f32_f32_f32_mknn_instance.cpp |   36 +-
 ..._shuffle_f32_f32_f32_f32_mnnn_instance.cpp |   36 +-
 ...xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp |   30 +-
 ...xdl_c_shuffle_f32_f32_f32_knn_instance.cpp |   36 +-
 ...xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp |   36 +-
 ...xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp |   36 +-
 .../gpu/conv1d_bwd_data/CMakeLists.txt        |   14 +
 ...bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp |  102 +
 ..._bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp |   95 +
 ..._bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp |   94 +
 ...bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp |   99 +
 .../gpu/conv1d_bwd_weight/CMakeLists.txt      |   13 +
 ...d_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp |  102 +
 ...wd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp |  102 +
 ...wd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp |  101 +
 .../gpu/conv1d_fwd/CMakeLists.txt             |   14 -
 ...nv1d_fwd_xdl_nwc_kxc_nwk_bf16_instance.cpp |  115 --
 ...onv1d_fwd_xdl_nwc_kxc_nwk_f16_instance.cpp |  115 --
 ...onv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp |  118 --
 ...nv1d_fwd_xdl_nwc_kxc_nwk_int8_instance.cpp |  117 --
 ..._data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp |   75 +-
 ...d_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp |   72 +-
 ...d_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp |   72 +-
 ..._data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp |  143 +-
 .../gpu/conv2d_bwd_weight/CMakeLists.txt      |   12 +-
 ...eight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp |  102 +
 ...weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp |  114 +-
 ...weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp |  113 +-
 .../gpu/conv2d_fwd/CMakeLists.txt             |   11 -
 ..._c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp |    8 +-
 ...d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp |   15 +-
 ...2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp |    8 +-
 ...2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp |    8 +-
 ...d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp |   17 +-
 ...d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp |  118 --
 ...2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp |  117 --
 ...2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp |  116 --
 ...d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp |  117 --
 .../gpu/conv3d_bwd_data/CMakeLists.txt        |   14 +
 ...ta_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp |  102 +
 ...ata_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp |  102 +
 ...ata_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp |  101 +
 ...ta_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp |   99 +
 .../gpu/conv3d_bwd_weight/CMakeLists.txt      |   13 +
 ...ht_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp |  104 +
 ...ght_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp |  103 +
 ...ght_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp |  102 +
 .../gpu/conv3d_fwd/CMakeLists.txt             |   12 -
 ...wd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp |  115 --
 ...fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp |  115 --
 ...fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp |  114 --
 ...wd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp |  117 --
 .../gpu/convnd_bwd_data/CMakeLists.txt        |   22 -
 ...bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp |   89 -
 ..._bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp |   91 -
 ..._bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp |   88 -
 ...bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp |   91 -
 ..._data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp |   89 -
 ...d_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp |   89 -
 ...d_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp |   88 -
 ..._data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp |   89 -
 ...ta_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp |   89 -
 ...ata_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp |   89 -
 ...ata_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp |   88 -
 ...ta_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp |   89 -
 .../gpu/convnd_bwd_weight/CMakeLists.txt      |   19 -
 ...d_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp |   87 -
 ...wd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp |   87 -
 ...wd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp |   86 -
 ...eight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp |   87 -
 ...weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp |   87 -
 ...weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp |   86 -
 ...ht_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp |   87 -
 ...ght_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp |   88 -
 ...ght_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp |   87 -
 .../gpu/gemm_add_add_fastgelu/CMakeLists.txt  |    8 +-
 ...16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp |   85 +
 ...16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp |   85 +
 ...16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp |   85 +
 ...16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp |   82 +
 ..._shuffle_f16_f16_f16_km_kn_mn_instance.cpp |   81 -
 ..._shuffle_f16_f16_f16_km_nk_mn_instance.cpp |   81 -
 ..._shuffle_f16_f16_f16_mk_kn_mn_instance.cpp |   81 -
 ..._shuffle_f16_f16_f16_mk_nk_mn_instance.cpp |   78 -
 .../gpu/gemm_bilinear/CMakeLists.txt          |    8 +-
 ...e_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp |  105 +
 ...e_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp |  105 +
 ...e_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp |  105 +
 ...e_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp |   99 +
 ..._shuffle_f16_f16_f16_km_kn_mn_instance.cpp |  103 -
 ..._shuffle_f16_f16_f16_km_nk_mn_instance.cpp |  103 -
 ..._shuffle_f16_f16_f16_mk_kn_mn_instance.cpp |  104 -
 ..._shuffle_f16_f16_f16_mk_nk_mn_instance.cpp |   97 -
 .../gpu/grouped_conv1d_fwd/CMakeLists.txt     |   12 +
 ...d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp |  129 ++
 ...1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp |  129 ++
 ...1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp |  128 ++
 ...d_fwd_xdl_gnwc_gkxc_gnwk_int8_instance.cpp |  125 ++
 .../gpu/grouped_conv2d_fwd/CMakeLists.txt     |   15 +
 ...wd_xdl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp |  129 ++
 ...fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instance.cpp |  129 ++
 ...fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instance.cpp |  128 ++
 ...wd_xdl_gnhwc_gkyxc_gnhwk_int8_instance.cpp |  125 ++
 ...fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instance.cpp |  129 ++
 .../gpu/grouped_conv3d_fwd/CMakeLists.txt     |   12 +
 ...xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp |  129 ++
 ..._xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp |  129 ++
 ..._xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp |  128 ++
 ...xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp |  125 ++
 ...gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp |   48 +-
 ...gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp |   47 +-
 ...gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp |   48 +-
 ...gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp |   42 +-
 library/src/utility/CMakeLists.txt            |   33 +-
 library/src/utility/conv_util.cpp             |  242 ---
 library/src/utility/convolution_parameter.cpp |  175 ++
 .../device_memory.cpp                         |    5 +-
 .../{host_tensor => utility}/host_tensor.cpp  |    4 +-
 profiler/CMakeLists.txt                       |   23 +-
 .../include/profile_batched_gemm_impl.hpp     |   12 +-
 .../profile_batched_gemm_reduce_impl.hpp      |   24 +-
 .../include/profile_conv_bwd_data_impl.hpp    |  249 +++
 .../include/profile_conv_bwd_weight_impl.hpp  |  381 ++--
 .../profile_conv_fwd_bias_relu_add_impl.hpp   |   16 +-
 .../profile_conv_fwd_bias_relu_impl.hpp       |   14 +-
 profiler/include/profile_conv_fwd_impl.hpp    |  221 ++
 .../profile_gemm_add_add_fastgelu_impl.hpp    |   31 +-
 .../profile_gemm_bias_add_reduce_impl.hpp     |   22 +-
 .../include/profile_gemm_bilinear_impl.hpp    |   26 +-
 profiler/include/profile_gemm_impl.hpp        |   41 +-
 profiler/include/profile_gemm_reduce_impl.hpp |   18 +-
 profiler/include/profile_gemm_splitk_impl.hpp |   12 +-
 .../include/profile_grouped_conv_fwd_impl.hpp |  250 +++
 .../include/profile_grouped_gemm_impl.hpp     |   31 +-
 .../include/profile_normalization_impl.hpp    |   12 +-
 profiler/include/profile_reduce_impl.hpp      |   14 +-
 profiler/src/profile_batched_gemm_reduce.cpp  |    5 +-
 profiler/src/profile_conv_bwd_data.cpp        |  184 ++
 profiler/src/profile_conv_bwd_weight.cpp      |  267 +--
 profiler/src/profile_conv_fwd.cpp             |  186 ++
 profiler/src/profile_convnd_bwd_data.cpp      |  229 ---
 profiler/src/profile_convnd_bwd_weight.cpp    |  226 --
 profiler/src/profile_convnd_fwd.cpp           |  359 ----
 profiler/src/profile_gemm.cpp                 |   60 +-
 .../src/profile_gemm_add_add_fastgelu.cpp     |   26 +-
 profiler/src/profile_gemm_bilinear.cpp        |   25 +-
 profiler/src/profile_grouped_conv_fwd.cpp     |  258 +++
 profiler/src/profile_grouped_gemm.cpp         |    8 +-
 profiler/src/profile_reduce.cpp               |    2 +-
 profiler/src/profiler.cpp                     |   45 +-
 test/CMakeLists.txt                           |    4 +-
 test/batched_gemm/CMakeLists.txt              |    2 +-
 test/batched_gemm_reduce/CMakeLists.txt       |    2 +-
 test/conv2d_bwd_data/CMakeLists.txt           |    3 -
 test/conv2d_bwd_data/conv2d_bwd_data.cpp      |  330 ---
 test/conv2d_bwd_weight/CMakeLists.txt         |    2 -
 test/conv2d_bwd_weight/conv2d_bwd_weight.cpp  |  217 --
 test/conv_util/CMakeLists.txt                 |    2 +-
 test/conv_util/conv_util.cpp                  |  219 +-
 test/convnd_bwd_data/CMakeLists.txt           |    4 +-
 test/convnd_bwd_data/convnd_bwd_data.cpp      |  532 ++---
 test/convnd_bwd_weight/CMakeLists.txt         |    4 +-
 test/convnd_bwd_weight/convnd_bwd_weight.cpp  |  460 ++---
 test/convnd_fwd/CMakeLists.txt                |   15 +-
 test/convnd_fwd/conv1d_fwd.cpp                |  192 --
 test/convnd_fwd/conv2d_fwd.cpp                |  266 ---
 test/convnd_fwd/conv3d_fwd.cpp                |  317 ---
 test/convnd_fwd/conv_util.hpp                 |  174 --
 test/convnd_fwd/convnd_fwd.cpp                |  241 +++
 test/gemm/CMakeLists.txt                      |    8 +-
 test/gemm/gemm_bf16.cpp                       |    6 +-
 test/gemm/gemm_fp16.cpp                       |    6 +-
 test/gemm/gemm_fp32.cpp                       |    6 +-
 test/gemm/gemm_fp64.cpp                       |    6 +-
 test/gemm/gemm_int8.cpp                       |    6 +-
 test/gemm/gemm_util.hpp                       |   12 +-
 test/gemm_reduce/CMakeLists.txt               |    2 +-
 test/gemm_split_k/CMakeLists.txt              |    2 +-
 test/gemm_split_k/gemm_split_k.cpp            |   14 +-
 test/grouped_convnd_fwd/CMakeLists.txt        |    3 +
 .../grouped_convnd_fwd/grouped_convnd_fwd.cpp |  270 +++
 test/grouped_gemm/CMakeLists.txt              |    2 +-
 test/layernorm/CMakeLists.txt                 |    8 +-
 test/layernorm/test_layernorm_util.hpp        |   12 +-
 test/magic_number_division/CMakeLists.txt     |    2 +-
 .../magic_number_division.cpp                 |    6 +-
 test/reduce/CMakeLists.txt                    |    4 +-
 test/reduce/reduce_no_index.cpp               |    2 +-
 test/reduce/reduce_with_index.cpp             |    2 +-
 test/reference_conv_fwd/CMakeLists.txt        |    2 +-
 .../reference_conv_fwd/reference_conv_fwd.cpp |  288 +--
 test/softmax/CMakeLists.txt                   |    8 +-
 test/softmax/test_softmax_util.hpp            |    8 +-
 373 files changed, 17544 insertions(+), 17013 deletions(-)
 delete mode 100644 example/06_conv2d_fwd_bias_relu/CMakeLists.txt
 delete mode 100644 example/06_conv2d_fwd_bias_relu/README.md
 delete mode 100644 example/06_conv2d_fwd_bias_relu/conv2d_fwd_xdl_bias_relu.cpp
 delete mode 100644 example/07_conv2d_fwd_bias_relu_add/CMakeLists.txt
 delete mode 100644 example/07_conv2d_fwd_bias_relu_add/README.md
 delete mode 100644 example/07_conv2d_fwd_bias_relu_add/conv2d_fwd_xdl_bias_relu_add.cpp
 create mode 100644 example/09_convnd_fwd/convnd_fwd_common.hpp
 create mode 100644 example/09_convnd_fwd/convnd_fwd_xdl_bf16.cpp
 delete mode 100644 example/10_conv2d_bwd_data/CMakeLists.txt
 delete mode 100644 example/10_conv2d_bwd_data/README.md
 delete mode 100644 example/10_conv2d_bwd_data/conv2d_bwd_data_xdl.cpp
 delete mode 100644 example/11_conv2d_bwd_weight/CMakeLists.txt
 delete mode 100644 example/11_conv2d_bwd_weight/README.md
 delete mode 100644 example/11_conv2d_bwd_weight/conv2d_bwd_weight_xdl.cpp
 create mode 100644 example/17_convnd_bwd_data/CMakeLists.txt
 rename example/{17_convnd_bwd_data_xdl => 17_convnd_bwd_data}/README.md (100%)
 create mode 100644 example/17_convnd_bwd_data/convnd_bwd_data_common.hpp
 create mode 100644 example/17_convnd_bwd_data/convnd_bwd_data_xdl_fp16.cpp
 delete mode 100644 example/17_convnd_bwd_data_xdl/CMakeLists.txt
 delete mode 100644 example/17_convnd_bwd_data_xdl/convnd_bwd_data_xdl.cpp
 create mode 100644 example/20_convnd_bwd_weight/CMakeLists.txt
 create mode 100644 example/20_convnd_bwd_weight/convnd_bwd_weight_common.hpp
 create mode 100644 example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_bf16.cpp
 create mode 100644 example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_fp16.cpp
 delete mode 100644 example/20_convnd_bwd_weight_xdl/CMakeLists.txt
 delete mode 100644 example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl.cpp
 delete mode 100644 example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl_bf16_splitk.cpp
 delete mode 100644 example/24_batched_gemm_c_permute/CMakeLists.txt
 create mode 100644 example/24_batched_gemm_e_permute/CMakeLists.txt
 rename example/{24_batched_gemm_c_permute/batched_gemm_c_permute_xdl_fp16.cpp => 24_batched_gemm_e_permute/batched_gemm_e_permute_xdl_fp16.cpp} (67%)
 delete mode 100644 example/25_gemm_bias_c_permute/CMakeLists.txt
 create mode 100644 example/25_gemm_bias_e_permute/CMakeLists.txt
 rename example/{25_gemm_bias_c_permute/gemm_bias_c_permute_xdl_fp16.cpp => 25_gemm_bias_e_permute/gemm_bias_e_permute_xdl_fp16.cpp} (96%)
 create mode 100644 example/30_grouped_convnd_fwd_bias_relu/CMakeLists.txt
 create mode 100644 example/30_grouped_convnd_fwd_bias_relu/README.md
 create mode 100644 example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_common.hpp
 create mode 100644 example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_relu_xdl_fp16.cpp
 rename include/ck/{device_utility => host_utility}/device_prop.hpp (100%)
 rename include/ck/{device_utility => host_utility}/hip_check_error.hpp (100%)
 create mode 100644 include/ck/host_utility/io.hpp
 rename include/ck/{device_utility => host_utility}/kernel_launch.hpp (97%)
 rename include/ck/tensor_operation/gpu/device/{device_batched_gemm_c_permute.hpp => device_batched_gemm_e_permute.hpp} (52%)
 create mode 100644 include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute_xdl.hpp
 rename include/ck/tensor_operation/gpu/device/{device_conv_backward_weight.hpp => device_conv_bwd_weight.hpp} (82%)
 rename include/ck/tensor_operation/gpu/device/{device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp => device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp} (97%)
 rename include/ck/tensor_operation/gpu/device/{device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp => device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp} (71%)
 delete mode 100644 include/ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp
 rename include/ck/tensor_operation/gpu/device/{device_gemm_bias_c_permute.hpp => device_gemm_bias_e_permute.hpp} (84%)
 rename include/ck/tensor_operation/gpu/device/{device_gemm_bias_c_permute_xdl.hpp => device_gemm_bias_e_permute_xdl.hpp} (59%)
 create mode 100644 include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp
 create mode 100644 include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp
 create mode 100644 include/ck/tensor_operation/gpu/device/matrix_padder.hpp
 rename library/include/ck/library/reference_tensor_operation/cpu/{reference_conv_backward_weight.hpp => reference_conv_bwd_weight.hpp} (57%)
 create mode 100644 library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp
 create mode 100644 library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_weight.hpp
 create mode 100644 library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp
 create mode 100644 library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp
 rename library/include/ck/library/{host_tensor => utility}/conv_common.hpp (100%)
 delete mode 100644 library/include/ck/library/utility/conv_util.hpp
 create mode 100644 library/include/ck/library/utility/convolution_host_tensor_descriptor_helper.hpp
 create mode 100644 library/include/ck/library/utility/convolution_parameter.hpp
 rename library/include/ck/library/{host_tensor => utility}/device_memory.hpp (100%)
 rename library/include/ck/library/{host_tensor => utility}/host_common_util.hpp (100%)
 rename library/include/ck/library/{host_tensor => utility}/host_conv.hpp (100%)
 rename library/include/ck/library/{host_tensor => utility}/host_gemm.hpp (100%)
 rename library/include/ck/library/{host_tensor => utility}/host_reduction.hpp (99%)
 rename library/include/ck/library/{host_tensor => utility}/host_tensor.hpp (79%)
 rename library/include/ck/library/{host_tensor => utility}/host_tensor_generator.hpp (100%)
 delete mode 100644 library/src/host_tensor/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_data/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv1d_fwd/CMakeLists.txt
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_data/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv3d_fwd/CMakeLists.txt
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/CMakeLists.txt
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/CMakeLists.txt
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
 delete mode 100644 library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp
 create mode 100644 library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp
 delete mode 100644 library/src/utility/conv_util.cpp
 create mode 100644 library/src/utility/convolution_parameter.cpp
 rename library/src/{host_tensor => utility}/device_memory.cpp (88%)
 rename library/src/{host_tensor => utility}/host_tensor.cpp (92%)
 create mode 100644 profiler/include/profile_conv_bwd_data_impl.hpp
 create mode 100644 profiler/include/profile_conv_fwd_impl.hpp
 create mode 100644 profiler/include/profile_grouped_conv_fwd_impl.hpp
 create mode 100644 profiler/src/profile_conv_bwd_data.cpp
 create mode 100644 profiler/src/profile_conv_fwd.cpp
 delete mode 100644 profiler/src/profile_convnd_bwd_data.cpp
 delete mode 100644 profiler/src/profile_convnd_bwd_weight.cpp
 delete mode 100644 profiler/src/profile_convnd_fwd.cpp
 create mode 100644 profiler/src/profile_grouped_conv_fwd.cpp
 delete mode 100644 test/conv2d_bwd_data/CMakeLists.txt
 delete mode 100644 test/conv2d_bwd_data/conv2d_bwd_data.cpp
 delete mode 100644 test/conv2d_bwd_weight/CMakeLists.txt
 delete mode 100644 test/conv2d_bwd_weight/conv2d_bwd_weight.cpp
 delete mode 100644 test/convnd_fwd/conv1d_fwd.cpp
 delete mode 100644 test/convnd_fwd/conv2d_fwd.cpp
 delete mode 100644 test/convnd_fwd/conv3d_fwd.cpp
 delete mode 100644 test/convnd_fwd/conv_util.hpp
 create mode 100644 test/convnd_fwd/convnd_fwd.cpp
 create mode 100644 test/grouped_convnd_fwd/CMakeLists.txt
 create mode 100644 test/grouped_convnd_fwd/grouped_convnd_fwd.cpp

diff --git a/README.md b/README.md
index aa1100dd13..bbc4d2bc30 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@ rocm/tensorflow:rocm5.1-tf2.6-dev              \
 /bin/bash
 ```
 
-# Install the new rocm-cmake version
+# Install newer version of rocm-cmake
 https://github.com/RadeonOpenCompute/rocm-cmake
 
 ## Build
@@ -54,6 +54,7 @@ make install
 ```
 
 ## Using CK as pre-built kernel library
+Instructions for using CK as a pre-built kernel library are under ```client_example/```
 
 ## Caveat
 ### Kernel Timing and Verification
diff --git a/client_example/02_gemm_add_add_fastgelu/gemm_add_add_fastgelu.cpp b/client_example/02_gemm_add_add_fastgelu/gemm_add_add_fastgelu.cpp
index dbf2e634f0..f88e72b62e 100644
--- a/client_example/02_gemm_add_add_fastgelu/gemm_add_add_fastgelu.cpp
+++ b/client_example/02_gemm_add_add_fastgelu/gemm_add_add_fastgelu.cpp
@@ -31,11 +31,11 @@ using D0DataType = F16;
 using D1DataType = F16;
 using EDataType  = F16;
 
-using ALayout   = Row;
-using BLayout   = Col;
-using DDELayout = Row;
-using DDELayout = Row;
-using DELayout  = Row;
+using ALayout  = Row;
+using BLayout  = Col;
+using D0Layout = Row;
+using D1Layout = Row;
+using ELayout  = Row;
 
 struct SimpleDeviceMem
 {
@@ -105,16 +105,16 @@ int main(int argc, char* argv[])
     SimpleDeviceMem a_device_buf(sizeof(ADataType) * f_matrix_space_size(M, K, StrideA, ALayout{}));
     SimpleDeviceMem b_device_buf(sizeof(BDataType) * f_matrix_space_size(K, N, StrideB, BLayout{}));
     SimpleDeviceMem d0_m_n_device_buf(sizeof(D0DataType) *
-                                      f_matrix_space_size(M, N, StrideD0, DDELayout{}));
+                                      f_matrix_space_size(M, N, StrideD0, D0Layout{}));
     SimpleDeviceMem d1_m_n_device_buf(sizeof(D1DataType) *
-                                      f_matrix_space_size(M, N, StrideD1, DDELayout{}));
-    SimpleDeviceMem e_device_buf(sizeof(EDataType) *
-                                 f_matrix_space_size(M, N, StrideE, DELayout{}));
+                                      f_matrix_space_size(M, N, StrideD1, D1Layout{}));
+    SimpleDeviceMem e_device_buf(sizeof(EDataType) * f_matrix_space_size(M, N, StrideE, ELayout{}));
 
     using DeviceOp = ck::tensor_operation::device::DeviceGemmMultipleD<
         ALayout,
         BLayout,
-        DDELayout,
+        ck::Tuple<D0Layout, D1Layout>,
+        ELayout,
         ADataType,
         BDataType,
         ck::Tuple<D0DataType, D1DataType>,
diff --git a/example/01_gemm/gemm_dl_fp16.cpp b/example/01_gemm/gemm_dl_fp16.cpp
index 0a3060fdc7..e4bd3906c2 100644
--- a/example/01_gemm/gemm_dl_fp16.cpp
+++ b/example/01_gemm/gemm_dl_fp16.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -142,9 +142,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_Sequential<1>{});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/01_gemm/gemm_dl_fp32.cpp b/example/01_gemm/gemm_dl_fp32.cpp
index d9677da9b9..0b5d5b6de1 100644
--- a/example/01_gemm/gemm_dl_fp32.cpp
+++ b/example/01_gemm/gemm_dl_fp32.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -141,9 +141,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_Sequential<1>{});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/01_gemm/gemm_dl_int8.cpp b/example/01_gemm/gemm_dl_int8.cpp
index 65206d602f..7787110580 100644
--- a/example/01_gemm/gemm_dl_int8.cpp
+++ b/example/01_gemm/gemm_dl_int8.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -139,9 +139,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_Sequential<1>{});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/01_gemm/gemm_xdl_bf16.cpp b/example/01_gemm/gemm_xdl_bf16.cpp
index 0575c0bd9e..f1a2448025 100644
--- a/example/01_gemm/gemm_xdl_bf16.cpp
+++ b/example/01_gemm/gemm_xdl_bf16.cpp
@@ -11,9 +11,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -170,9 +170,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/01_gemm/gemm_xdl_fp16.cpp b/example/01_gemm/gemm_xdl_fp16.cpp
index 0d19440377..17a067a94c 100644
--- a/example/01_gemm/gemm_xdl_fp16.cpp
+++ b/example/01_gemm/gemm_xdl_fp16.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -155,9 +155,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_Sequential<1>{});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/01_gemm/gemm_xdl_fp64.cpp b/example/01_gemm/gemm_xdl_fp64.cpp
index 1b222c9712..82e2f99b98 100644
--- a/example/01_gemm/gemm_xdl_fp64.cpp
+++ b/example/01_gemm/gemm_xdl_fp64.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_xdl.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -165,9 +165,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_1<BDataType>{1});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/01_gemm/gemm_xdl_int8.cpp b/example/01_gemm/gemm_xdl_int8.cpp
index 4ed1f177db..ca5c66f8af 100644
--- a/example/01_gemm/gemm_xdl_int8.cpp
+++ b/example/01_gemm/gemm_xdl_int8.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -167,9 +167,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/02_gemm_bilinear/gemm_bilinear_xdl_fp16.cpp b/example/02_gemm_bilinear/gemm_bilinear_xdl_fp16.cpp
index 9b340807ba..081f2b5142 100644
--- a/example/02_gemm_bilinear/gemm_bilinear_xdl_fp16.cpp
+++ b/example/02_gemm_bilinear/gemm_bilinear_xdl_fp16.cpp
@@ -11,9 +11,9 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -51,33 +51,34 @@ using BDataType        = F16;
 using AccDataType      = F32;
 using CShuffleDataType = F32;
 using DDataType        = F16;
-using DsDataType       = ck::Tuple<DDataType>;
 using EDataType        = F16;
 
-using ALayout  = Row;
-using BLayout  = Col;
-using DELayout = Row;
+using ALayout = Row;
+using BLayout = Col;
+using DLayout = Row;
+using ELayout = Row;
 
 using AElementOp   = PassThrough;
 using BElementOp   = PassThrough;
 using CDEElementOp = AlphaBetaAdd;
 
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
 using DeviceOpInstance =
     ck::tensor_operation::device::DeviceGemmMultipleD_Xdl_CShuffle<ALayout,
                                                                    BLayout,
-                                                                   DELayout,
+                                                                   ck::Tuple<DLayout>,
+                                                                   ELayout,
                                                                    ADataType,
                                                                    BDataType,
                                                                    AccDataType,
                                                                    CShuffleDataType,
-                                                                   DsDataType,
+                                                                   ck::Tuple<DDataType>,
                                                                    EDataType,
                                                                    AElementOp,
                                                                    BElementOp,
                                                                    CDEElementOp,
-                                                                   GemmDefault,
+                                                                   GemmSpec,
                                                                    1,
                                                                    256,
                                                                    256,
@@ -190,9 +191,9 @@ int main(int argc, char* argv[])
 
     Tensor<ADataType> a_m_k(f_host_tensor_descriptor(M, K, StrideA, ALayout{}));
     Tensor<BDataType> b_k_n(f_host_tensor_descriptor(K, N, StrideB, BLayout{}));
-    Tensor<DDataType> d_m_n(f_host_tensor_descriptor(M, N, StrideD, DELayout{}));
-    Tensor<EDataType> e_m_n_host_result(f_host_tensor_descriptor(M, N, StrideE, DELayout{}));
-    Tensor<EDataType> e_m_n_device_result(f_host_tensor_descriptor(M, N, StrideE, DELayout{}));
+    Tensor<DDataType> d_m_n(f_host_tensor_descriptor(M, N, StrideD, DLayout{}));
+    Tensor<EDataType> e_m_n_host_result(f_host_tensor_descriptor(M, N, StrideE, ELayout{}));
+    Tensor<EDataType> e_m_n_device_result(f_host_tensor_descriptor(M, N, StrideE, ELayout{}));
 
     std::cout << "a_m_k: " << a_m_k.mDesc << std::endl;
     std::cout << "b_k_n: " << b_k_n.mDesc << std::endl;
@@ -213,10 +214,10 @@ int main(int argc, char* argv[])
         d_m_n.GenerateTensorValue(GeneratorTensor_3<DDataType>{-0.5, 0.5});
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem d_device_buf(sizeof(DDataType) * d_m_n.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem d_device_buf(sizeof(DDataType) * d_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/03_gemm_bias_relu/gemm_bias_relu_xdl_fp16.cpp b/example/03_gemm_bias_relu/gemm_bias_relu_xdl_fp16.cpp
index e36280f42d..ae5e323410 100644
--- a/example/03_gemm_bias_relu/gemm_bias_relu_xdl_fp16.cpp
+++ b/example/03_gemm_bias_relu/gemm_bias_relu_xdl_fp16.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -47,33 +47,34 @@ using BDataType        = F16;
 using AccDataType      = F32;
 using CShuffleDataType = F16;
 using DDataType        = F16;
-using DsDataType       = ck::Tuple<DDataType>;
 using EDataType        = F16;
 
 using ALayout = Row;
 using BLayout = Col;
+using DLayout = Row;
 using ELayout = Row;
 
 using AElementOp   = PassThrough;
 using BElementOp   = PassThrough;
 using CDEElementOp = AddRelu;
 
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
 using DeviceOpInstance =
     ck::tensor_operation::device::DeviceGemmMultipleD_Xdl_CShuffle<ALayout,
                                                                    BLayout,
+                                                                   ck::Tuple<DLayout>,
                                                                    ELayout,
                                                                    ADataType,
                                                                    BDataType,
                                                                    AccDataType,
                                                                    CShuffleDataType,
-                                                                   DsDataType,
+                                                                   ck::Tuple<DDataType>,
                                                                    EDataType,
                                                                    AElementOp,
                                                                    BElementOp,
                                                                    CDEElementOp,
-                                                                   GemmDefault,
+                                                                   GemmSpec,
                                                                    1,
                                                                    256,
                                                                    256,
@@ -191,10 +192,10 @@ int main(int argc, char* argv[])
         d_m_n.GenerateTensorValue(GeneratorTensor_3<DDataType>{0.0, 1.0});
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem d_device_buf(sizeof(DDataType) * d_m_n.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem d_device_buf(sizeof(DDataType) * d_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/04_gemm_add_add_fastgelu/gemm_add_add_fastgelu_xdl_fp16.cpp b/example/04_gemm_add_add_fastgelu/gemm_add_add_fastgelu_xdl_fp16.cpp
index 4bfbbbadf8..c440297ec6 100644
--- a/example/04_gemm_add_add_fastgelu/gemm_add_add_fastgelu_xdl_fp16.cpp
+++ b/example/04_gemm_add_add_fastgelu/gemm_add_add_fastgelu_xdl_fp16.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -43,6 +43,7 @@ using ALayout  = Row;
 using BLayout  = Col;
 using D0Layout = Row;
 using D1Layout = Row;
+using DsLayout = ck::Tuple<D0Layout, D1Layout>;
 using ELayout  = Row;
 
 using AElementOp   = PassThrough;
@@ -53,11 +54,11 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 
 // clang-format off
 using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMultipleD_Xdl_CShuffle
-//######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-//######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-//######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-//######|        |        |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
+//######| ALayout| BLayout| DsLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+//######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+//######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 int main(int argc, char* argv[])
@@ -156,11 +157,11 @@ int main(int argc, char* argv[])
         d1_m_n.GenerateTensorValue(GeneratorTensor_3<D1DataType>{0.0, 1.0});
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem d0_device_buf(sizeof(D0DataType) * d0_m_n.mDesc.GetElementSpace());
-    DeviceMem d1_device_buf(sizeof(D1DataType) * d1_m_n.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem d0_device_buf(sizeof(D0DataType) * d0_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem d1_device_buf(sizeof(D1DataType) * d1_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/06_conv2d_fwd_bias_relu/CMakeLists.txt b/example/06_conv2d_fwd_bias_relu/CMakeLists.txt
deleted file mode 100644
index 4e1dd1f3e6..0000000000
--- a/example/06_conv2d_fwd_bias_relu/CMakeLists.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-add_example_executable(example_conv2d_fwd_xdl_bias_relu conv2d_fwd_xdl_bias_relu.cpp)
-target_link_libraries(example_conv2d_fwd_xdl_bias_relu PRIVATE conv_util)
diff --git a/example/06_conv2d_fwd_bias_relu/README.md b/example/06_conv2d_fwd_bias_relu/README.md
deleted file mode 100644
index 4c30563ef0..0000000000
--- a/example/06_conv2d_fwd_bias_relu/README.md
+++ /dev/null
@@ -1,22 +0,0 @@
-# Instructions for ```example_conv_xdl_bias_relu```
-
-## Run ```example_conv_xdl_bias_relu```
-```bash
-#arg1: verification (0=no, 1=yes)
-#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
-#arg3: run kernel # of times (>1)
-#arg4 to 18: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, RightPx
-./bin/example_conv_xdl_bias_relu 0 1 5
-```
-
-Result (MI100 @ 1087Mhz, 133.5TFlops peak FP16)
-```
-in_n_c_hi_wi: dim 4, lengths {128, 192, 71, 71}, strides {967872, 1, 13632, 192}
-wei_k_c_y_x: dim 4, lengths {256, 192, 3, 3}, strides {1728, 1, 576, 192}
-out_n_k_ho_wo: dim 4, lengths {128, 256, 36, 36}, strides {331776, 1, 9216, 256}
-bias_k: dim 1, lengths {256}, strides {1}
-launch_and_time_kernel: grid_dim {1296, 1, 1}, block_dim {256, 1, 1}
-Warm up
-Start running 5 times...
-Perf: 1.39009 ms, 105.581 TFlops, 239.981 GB/s
-```
diff --git a/example/06_conv2d_fwd_bias_relu/conv2d_fwd_xdl_bias_relu.cpp b/example/06_conv2d_fwd_bias_relu/conv2d_fwd_xdl_bias_relu.cpp
deleted file mode 100644
index b3c492fd23..0000000000
--- a/example/06_conv2d_fwd_bias_relu/conv2d_fwd_xdl_bias_relu.cpp
+++ /dev/null
@@ -1,313 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp"
-
-namespace {
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InLayout  = ck::tensor_layout::convolution::NHWC;
-using WeiLayout = ck::tensor_layout::convolution::KYXC;
-using OutLayout = ck::tensor_layout::convolution::NHWK;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::AddRelu;
-
-static constexpr auto MemorySet = ck::InMemoryDataOperationEnum::Set;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-// clang-format off
-using DeviceConvFwdInstance = ck::tensor_operation::device::
-    DeviceConv2dFwdXdl_C_Shuffle_Bias_Activation_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        InDataType,                   // InDataType
-        WeiDataType,                  // WeiDataType
-        OutDataType,                  // OutDataType
-        AccDataType,                  // AccDataType
-        InElementOp,                  // InElementwiseOperation
-        WeiElementOp,                 // WeiElementwiseOperation
-        OutElementOp,                 // OutElementwiseOperation
-        MemorySet,                    // OutGlobalMemoryDataOperation
-        ConvFwdDefault,               // ConvForwardSpecialization
-        256,                          // BlockSize
-        128,                          // MPerBlock
-        256,                          // NPerBlock
-        4,                            // K0PerBlock
-        8,                            // K1
-        32,                           // MPerXdl
-        32,                           // NPerXdl
-        2,                            // MXdlPerWave
-        4,                            // NXdlPerWave
-        S<4, 64, 1>,                  // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,                   // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,                   // ABlockTransferSrcAccessOrder
-        2,                            // ABlockTransferSrcVectorDim
-        8,                            // ABlockTransferSrcScalarPerVector
-        8,                            // ABlockTransferDstScalarPerVector_K1
-        true,                         // ABlockLdsAddExtraM
-        S<4, 64, 1>,                  // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,                   // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,                   // BBlockTransferSrcAccessOrder
-        2,                            // BBlockTransferSrcVectorDim
-        8,                            // BBlockTransferSrcScalarPerVector
-        8,                            // BBlockTransferDstScalarPerVector_K1
-        true,                         // BBlockLdsAddExtraN
-        1,                            // CShuffleMXdlPerWavePerShuffle
-        1,                            // CShuffleNXdlPerWavePerShuffle
-        S<1, 1, 32, 1, 1, 8>,         // CBlockTransferClusterLengths_MBlock_MXdlPerWave_MWaveMPerXdl_NBlock_NXdlPerWave_NWaveNPerXdl
-        8>;                           // CBlockTransferScalarPerVector_NWaveNPerXdl
-// clang-format on
-
-using ReferenceConvFwdInstance =
-    ck::tensor_operation::host::ReferenceConvFwd_Bias_Activation<InDataType,
-                                                                 WeiDataType,
-                                                                 OutDataType,
-                                                                 InElementOp,
-                                                                 WeiElementOp,
-                                                                 OutElementOp>;
-
-void PrintUseMsg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "Following arguments:\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams ParseConvParams(int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int num_dim_spatial = 2;
-    int conv_args       = 3 + num_dim_spatial * 6;
-    int cmdline_nargs   = conv_args + 4;
-    if(cmdline_nargs != argc)
-    {
-        PrintUseMsg();
-        exit(0);
-    }
-
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 4;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // anonymous namespace
-
-int main(int argc, char* argv[])
-{
-    using namespace ck::utils::conv;
-
-    bool do_verification      = true;
-    int init_method           = 1;
-    bool time_kernel          = false;
-    const int num_dim_spatial = 2;
-
-    ck::utils::conv::ConvParams params;
-
-    if(argc >= 4)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-    }
-
-    if(argc >= 5)
-    {
-        params = ParseConvParams(argc, argv);
-    }
-
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
-
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> input(get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> weights(get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> host_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    Tensor<OutDataType> device_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    // bias: assume contiguous 1d vector
-    Tensor<OutDataType> bias(
-        HostTensorDescriptor(std::vector<std::size_t>({static_cast<std::size_t>(params.K_)})));
-
-    std::cout << "input: " << input.mDesc << std::endl;
-    std::cout << "weights: " << weights.mDesc << std::endl;
-    std::cout << "output: " << host_output.mDesc << std::endl;
-    std::cout << "bias: " << bias.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        weights.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-        bias.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
-        break;
-    default:
-        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        weights.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
-        bias.GenerateTensorValue(GeneratorTensor_3<OutDataType>{0.0, 1.0});
-    }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * weights.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpace());
-    DeviceMem bias_device_buf(sizeof(OutDataType) * bias.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(input.mData.data());
-    wei_device_buf.ToDevice(weights.mData.data());
-    bias_device_buf.ToDevice(bias.mData.data());
-
-    auto conv    = DeviceConvFwdInstance{};
-    auto invoker = conv.MakeInvoker();
-    auto argument =
-        conv.MakeArgument(static_cast<const InDataType*>(in_device_buf.GetDeviceBuffer()),
-                          static_cast<const WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                          static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                          static_cast<const OutDataType*>(bias_device_buf.GetDeviceBuffer()),
-                          params.N_,
-                          params.K_,
-                          params.C_,
-                          params.input_spatial_lengths_,
-                          params.filter_spatial_lengths_,
-                          output_spatial_lengths,
-                          params.conv_filter_strides_,
-                          params.conv_filter_dilations_,
-                          params.input_left_pads_,
-                          params.input_right_pads_,
-                          InElementOp{},
-                          WeiElementOp{},
-                          OutElementOp{});
-
-    if(!conv.IsSupportedArgument(argument))
-    {
-        throw std::runtime_error(
-            "wrong! device operator with the specified compilation parameters does "
-            "not support this problem");
-    }
-
-    float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype =
-        get_btype<InDataType, WeiDataType, OutDataType>(params.N_,
-                                                        params.C_,
-                                                        params.K_,
-                                                        params.input_spatial_lengths_,
-                                                        params.filter_spatial_lengths_,
-                                                        output_spatial_lengths) +
-        sizeof(OutDataType) * (params.K_);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
-    {
-        auto ref_conv    = ReferenceConvFwdInstance{};
-        auto ref_invoker = ref_conv.MakeInvoker();
-
-        auto ref_argument = ref_conv.MakeArgument(input,
-                                                  weights,
-                                                  host_output,
-                                                  bias,
-                                                  params.conv_filter_strides_,
-                                                  params.conv_filter_dilations_,
-                                                  params.input_left_pads_,
-                                                  params.input_right_pads_,
-                                                  InElementOp{},
-                                                  WeiElementOp{},
-                                                  OutElementOp{});
-        ref_invoker.Run(ref_argument);
-        out_device_buf.FromDevice(device_output.mData.data());
-        return ck::utils::check_err(device_output.mData, host_output.mData) ? 0 : 1;
-    }
-
-    return 0;
-}
diff --git a/example/07_conv2d_fwd_bias_relu_add/CMakeLists.txt b/example/07_conv2d_fwd_bias_relu_add/CMakeLists.txt
deleted file mode 100644
index b4dd39d83a..0000000000
--- a/example/07_conv2d_fwd_bias_relu_add/CMakeLists.txt
+++ /dev/null
@@ -1,3 +0,0 @@
-# FIXME: should fix validation failure
-add_example_executable_no_testing(example_conv2d_fwd_xdl_bias_relu_add conv2d_fwd_xdl_bias_relu_add.cpp)
-target_link_libraries(example_conv2d_fwd_xdl_bias_relu_add PRIVATE conv_util)
diff --git a/example/07_conv2d_fwd_bias_relu_add/README.md b/example/07_conv2d_fwd_bias_relu_add/README.md
deleted file mode 100644
index 99afcae9c8..0000000000
--- a/example/07_conv2d_fwd_bias_relu_add/README.md
+++ /dev/null
@@ -1,24 +0,0 @@
-# Instructions for ```example_conv_xdl_bias_relu_add```
-
-
-## Run ```example_conv_xdl_bias_relu_add```
-```bash
-#arg1: verification (0=no, 1=yes)
-#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
-#arg3: run kernel # of times (>1)
-#arg4 to 18: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, RightPx
-./bin/example_conv_xdl_bias_relu_add 0 1 5
-```
-
-Result (MI100 @ 1087Mhz, 133.5TFlops peak FP16)
-```
-in_n_c_hi_wi: dim 4, lengths {128, 192, 71, 71}, strides {967872, 1, 13632, 192}
-wei_k_c_y_x: dim 4, lengths {256, 192, 3, 3}, strides {1728, 1, 576, 192}
-out_n_k_ho_wo: dim 4, lengths {128, 256, 36, 36}, strides {331776, 1, 9216, 256}
-bias_k: dim 1, lengths {256}, strides {1}
-resi_n_k_ho_wo: dim 4, lengths {128, 256, 36, 36}, strides {331776, 1, 9216, 256}
-launch_and_time_kernel: grid_dim {1296, 1, 1}, block_dim {256, 1, 1}
-Warm up
-Start running 5 times...
-Perf: 1.44711 ms, 101.421 TFlops, 289.218 GB/s
-```
diff --git a/example/07_conv2d_fwd_bias_relu_add/conv2d_fwd_xdl_bias_relu_add.cpp b/example/07_conv2d_fwd_bias_relu_add/conv2d_fwd_xdl_bias_relu_add.cpp
deleted file mode 100644
index 7950630adb..0000000000
--- a/example/07_conv2d_fwd_bias_relu_add/conv2d_fwd_xdl_bias_relu_add.cpp
+++ /dev/null
@@ -1,328 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp"
-
-namespace {
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InLayout  = ck::tensor_layout::convolution::NHWC;
-using WeiLayout = ck::tensor_layout::convolution::KYXC;
-using OutLayout = ck::tensor_layout::convolution::NHWK;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::AddReluAdd;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-// clang-format off
-using DeviceConvFwdInstance = ck::tensor_operation::device::
-    DeviceConv2dFwdXdl_C_Shuffle_Bias_Activation_Add_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        InDataType,              // InDataType
-        WeiDataType,             // WeiDataType
-        OutDataType,             // OutDataType
-        AccDataType,             // AccDataType
-        InElementOp,             // InElementwiseOperation
-        WeiElementOp,            // WeiElementwiseOperation
-        OutElementOp,            // OutElementwiseOperation
-        ConvFwdDefault,          // ConvForwardSpecialization
-        256,                     // BlockSize
-        128,                     // MPerBlock
-        256,                     // NPerBlock
-        4,                       // K0PerBlock
-        8,                       // K1
-        32,                      // MPerXdl
-        32,                      // NPerXdl
-        2,                       // MXdlPerWave
-        4,                       // NXdlPerWave
-        S<4, 64, 1>,             // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,              // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,              // ABlockTransferSrcAccessOrder
-        2,                       // ABlockTransferSrcVectorDim
-        8,                       // ABlockTransferSrcScalarPerVector
-        8,                       // ABlockTransferDstScalarPerVector_K1
-        true,                    // ABlockLdsAddExtraM
-        S<4, 64, 1>,             // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,              // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,              // BBlockTransferSrcAccessOrder
-        2,                       // BBlockTransferSrcVectorDim
-        8,                       // BBlockTransferSrcScalarPerVector
-        8,                       // BBlockTransferDstScalarPerVector_K1
-        true,                    // BBlockLdsAddExtraN
-        1,                       // CShuffleMXdlPerWavePerShuffle
-        1,                       // CShuffleNXdlPerWavePerShuffle
-        S<1, 1, 32, 1, 1, 8>,    // CBlockTransferClusterLengths_MBlock_MXdlPerWave_MWaveMPerXdl_NBlock_NXdlPerWave_NWaveNPerXdl
-        8>;                      // CBlockTransferScalarPerVector_NWaveNPerXdl
-// clang-format on
-
-using ReferenceConvFwdInstance =
-    ck::tensor_operation::host::ReferenceConvFwd_Bias_Activation_Add<InDataType,
-                                                                     WeiDataType,
-                                                                     OutDataType,
-                                                                     InElementOp,
-                                                                     WeiElementOp,
-                                                                     OutElementOp>;
-
-void PrintUseMsg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "Following arguments:\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams ParseConvParams(int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int num_dim_spatial = 2;
-    int conv_args       = 3 + num_dim_spatial * 6;
-    int cmdline_nargs   = conv_args + 4;
-    if(cmdline_nargs != argc)
-    {
-        PrintUseMsg();
-        exit(0);
-    }
-
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 4;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // anonymous namespace
-
-int main(int argc, char* argv[])
-{
-    using namespace ck::utils::conv;
-
-    bool do_verification      = true;
-    int init_method           = 1;
-    bool time_kernel          = false;
-    const int num_dim_spatial = 2;
-
-    ck::utils::conv::ConvParams params;
-
-    if(argc >= 4)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-    }
-
-    if(argc >= 5)
-    {
-        params = ParseConvParams(argc, argv);
-    }
-
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
-
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> input(get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> weights(get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> host_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    Tensor<OutDataType> device_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    // bias: assume contiguous 1d vector
-    Tensor<OutDataType> bias(
-        HostTensorDescriptor(std::vector<std::size_t>({static_cast<std::size_t>(params.K_)})));
-
-    // residual: assume same layout as output tensor
-    Tensor<OutDataType> residual(get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "input: " << input.mDesc << std::endl;
-    std::cout << "weights: " << weights.mDesc << std::endl;
-    std::cout << "output: " << host_output.mDesc << std::endl;
-    std::cout << "bias: " << bias.mDesc << std::endl;
-    std::cout << "residual: " << residual.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-2, 2});
-        weights.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-2, 2});
-        bias.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-2, 2});
-        residual.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-2, 2});
-        break;
-    default:
-        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        weights.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
-        bias.GenerateTensorValue(GeneratorTensor_3<OutDataType>{0.0, 1.0});
-        residual.GenerateTensorValue(GeneratorTensor_3<OutDataType>{0.0, 1.0});
-    }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * weights.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpace());
-    DeviceMem bias_device_buf(sizeof(OutDataType) * bias.mDesc.GetElementSpace());
-    DeviceMem resi_device_buf(sizeof(OutDataType) * residual.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(input.mData.data());
-    wei_device_buf.ToDevice(weights.mData.data());
-    bias_device_buf.ToDevice(bias.mData.data());
-    resi_device_buf.ToDevice(residual.mData.data());
-
-    const auto in_element_op  = InElementOp{};
-    const auto wei_element_op = WeiElementOp{};
-    const auto out_element_op = OutElementOp{};
-
-    auto conv    = DeviceConvFwdInstance{};
-    auto invoker = conv.MakeInvoker();
-    auto argument =
-        conv.MakeArgument(static_cast<const InDataType*>(in_device_buf.GetDeviceBuffer()),
-                          static_cast<const WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                          static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                          static_cast<const OutDataType*>(bias_device_buf.GetDeviceBuffer()),
-                          static_cast<const OutDataType*>(resi_device_buf.GetDeviceBuffer()),
-                          params.N_,
-                          params.K_,
-                          params.C_,
-                          params.input_spatial_lengths_,
-                          params.filter_spatial_lengths_,
-                          output_spatial_lengths,
-                          params.conv_filter_strides_,
-                          params.conv_filter_dilations_,
-                          params.input_left_pads_,
-                          params.input_right_pads_,
-                          in_element_op,
-                          wei_element_op,
-                          out_element_op);
-
-    if(!conv.IsSupportedArgument(argument))
-    {
-        throw std::runtime_error(
-            "wrong! device operator with the specified compilation parameters does "
-            "not support this problem");
-    }
-
-    float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype =
-        get_btype<InDataType, WeiDataType, OutDataType>(params.N_,
-                                                        params.C_,
-                                                        params.K_,
-                                                        params.input_spatial_lengths_,
-                                                        params.filter_spatial_lengths_,
-                                                        output_spatial_lengths) +
-        sizeof(OutDataType) * (params.K_) +
-        sizeof(OutDataType) *
-            (params.N_ * params.K_ * output_spatial_lengths[0] * output_spatial_lengths[1]);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
-    {
-        auto ref_conv    = ReferenceConvFwdInstance{};
-        auto ref_invoker = ref_conv.MakeInvoker();
-
-        auto ref_argument = ref_conv.MakeArgument(input,
-                                                  weights,
-                                                  host_output,
-                                                  bias,
-                                                  residual,
-                                                  params.conv_filter_strides_,
-                                                  params.conv_filter_dilations_,
-                                                  params.input_left_pads_,
-                                                  params.input_right_pads_,
-                                                  in_element_op,
-                                                  wei_element_op,
-                                                  out_element_op);
-
-        ref_invoker.Run(ref_argument);
-        out_device_buf.FromDevice(device_output.mData.data());
-        return ck::utils::check_err(device_output.mData, host_output.mData) ? 0 : 1;
-    }
-
-    return 0;
-}
diff --git a/example/09_convnd_fwd/CMakeLists.txt b/example/09_convnd_fwd/CMakeLists.txt
index 1724e51f3f..b373d1d6c0 100644
--- a/example/09_convnd_fwd/CMakeLists.txt
+++ b/example/09_convnd_fwd/CMakeLists.txt
@@ -1,9 +1,6 @@
 add_example_executable(example_convnd_fwd_xdl_fp32 convnd_fwd_xdl_fp32.cpp)
-add_example_executable(example_convnd_fwd_xdl_int8 convnd_fwd_xdl_int8.cpp)
 add_example_executable(example_convnd_fwd_xdl_fp16 convnd_fwd_xdl_fp16.cpp)
+add_example_executable(example_convnd_fwd_xdl_bf16 convnd_fwd_xdl_bf16.cpp)
+add_example_executable(example_convnd_fwd_xdl_int8 convnd_fwd_xdl_int8.cpp)
 # FIXME: re-enable this exampe as test when SWDEV-335738 is fixed
 add_example_executable_no_testing(example_convnd_fwd_xdl_fp64 convnd_fwd_xdl_fp64.cpp)
-target_link_libraries(example_convnd_fwd_xdl_fp64 PRIVATE conv_util)
-target_link_libraries(example_convnd_fwd_xdl_fp32 PRIVATE conv_util)
-target_link_libraries(example_convnd_fwd_xdl_int8 PRIVATE conv_util)
-target_link_libraries(example_convnd_fwd_xdl_fp16 PRIVATE conv_util)
diff --git a/example/09_convnd_fwd/convnd_fwd_common.hpp b/example/09_convnd_fwd/convnd_fwd_common.hpp
new file mode 100644
index 0000000000..c05ab86f60
--- /dev/null
+++ b/example/09_convnd_fwd/convnd_fwd_common.hpp
@@ -0,0 +1,173 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <numeric>
+#include <type_traits>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+
+void print_helper_msg()
+{
+    std::cout << "arg1: verification (0=no, 1=yes)\n"
+              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
+              << "arg3: time kernel (0=no, 1=yes)\n"
+              << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+}
+
+template <ck::index_t NDimSpatial,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementOp,
+          typename WeiElementOp,
+          typename OutElementOp,
+          typename DeviceConvNDFwdInstance>
+int run_grouped_conv_fwd(bool do_verification,
+                         int init_method,
+                         bool time_kernel,
+                         const ck::utils::conv::ConvParam& conv_param,
+                         const HostTensorDescriptor& in_g_n_c_wis_desc,
+                         const HostTensorDescriptor& wei_g_k_c_xs_desc,
+                         const HostTensorDescriptor& out_g_n_k_wos_desc,
+                         const InElementOp& in_element_op,
+                         const WeiElementOp& wei_element_op,
+                         const OutElementOp& out_element_op)
+{
+    Tensor<InDataType> in(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> wei(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> out_host(out_g_n_k_wos_desc);
+    Tensor<OutDataType> out_device(out_g_n_k_wos_desc);
+
+    std::cout << "in: " << in.mDesc << std::endl;
+    std::cout << "wei: " << wei.mDesc << std::endl;
+    std::cout << "out: " << out_host.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        in.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
+        wei.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        break;
+    default:
+        in.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
+        wei.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * out_device.mDesc.GetElementSpaceSize());
+
+    in_device_buf.ToDevice(in.mData.data());
+    wei_device_buf.ToDevice(wei.mData.data());
+
+    std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_strides{};
+    std::array<ck::index_t, NDimSpatial> conv_filter_strides{};
+    std::array<ck::index_t, NDimSpatial> conv_filter_dilations{};
+    std::array<ck::index_t, NDimSpatial> input_left_pads{};
+    std::array<ck::index_t, NDimSpatial> input_right_pads{};
+
+    auto copy = [](auto& x, auto& y) { std::copy(x.begin(), x.end(), y.begin()); };
+
+    copy(in_g_n_c_wis_desc.GetLengths(), a_g_n_c_wis_lengths);
+    copy(in_g_n_c_wis_desc.GetStrides(), a_g_n_c_wis_strides);
+    copy(wei_g_k_c_xs_desc.GetLengths(), b_g_k_c_xs_lengths);
+    copy(wei_g_k_c_xs_desc.GetStrides(), b_g_k_c_xs_strides);
+    copy(out_g_n_k_wos_desc.GetLengths(), e_g_n_k_wos_lengths);
+    copy(out_g_n_k_wos_desc.GetStrides(), e_g_n_k_wos_strides);
+    copy(conv_param.conv_filter_strides_, conv_filter_strides);
+    copy(conv_param.conv_filter_dilations_, conv_filter_dilations);
+    copy(conv_param.input_left_pads_, input_left_pads);
+    copy(conv_param.input_right_pads_, input_right_pads);
+
+    // do Conv
+    auto conv     = DeviceConvNDFwdInstance{};
+    auto invoker  = conv.MakeInvoker();
+    auto argument = conv.MakeArgument(in_device_buf.GetDeviceBuffer(),
+                                      wei_device_buf.GetDeviceBuffer(),
+                                      std::array<const void*, 0>{},
+                                      out_device_buf.GetDeviceBuffer(),
+                                      a_g_n_c_wis_lengths,
+                                      a_g_n_c_wis_strides,
+                                      b_g_k_c_xs_lengths,
+                                      b_g_k_c_xs_strides,
+                                      std::array<std::array<ck::index_t, NDimSpatial + 3>, 0>{{}},
+                                      std::array<std::array<ck::index_t, NDimSpatial + 3>, 0>{{}},
+                                      e_g_n_k_wos_lengths,
+                                      e_g_n_k_wos_strides,
+                                      conv_filter_strides,
+                                      conv_filter_dilations,
+                                      input_left_pads,
+                                      input_right_pads,
+                                      in_element_op,
+                                      wei_element_op,
+                                      out_element_op);
+
+    if(!conv.IsSupportedArgument(argument))
+    {
+        throw std::runtime_error(
+            "wrong! device_conv with the specified compilation parameters does "
+            "not support this Conv problem");
+    }
+
+    float avg_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
+
+    std::size_t flop      = conv_param.GetFlops();
+    std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+    float tflops     = static_cast<float>(flop) / 1.E9 / avg_time;
+    float gb_per_sec = num_btype / 1.E6 / avg_time;
+    std::cout << "Perf: " << avg_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s, "
+              << conv.GetTypeString() << std::endl;
+
+    if(do_verification)
+    {
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvFwd<NDimSpatial,
+                                                                     InDataType,
+                                                                     WeiDataType,
+                                                                     OutDataType,
+                                                                     InElementOp,
+                                                                     WeiElementOp,
+                                                                     OutElementOp>();
+
+        auto ref_invoker  = ref_conv.MakeInvoker();
+        auto ref_argument = ref_conv.MakeArgument(in,
+                                                  wei,
+                                                  out_host,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  in_element_op,
+                                                  wei_element_op,
+                                                  out_element_op);
+
+        ref_invoker.Run(ref_argument);
+
+        out_device_buf.FromDevice(out_device.mData.data());
+
+        return ck::utils::check_err(
+                   out_device.mData, out_host.mData, "Error: incorrect results!", 1e-5f, 1e-4f)
+                   ? 0
+                   : 1;
+    }
+
+    return 0;
+}
diff --git a/example/09_convnd_fwd/convnd_fwd_xdl_bf16.cpp b/example/09_convnd_fwd/convnd_fwd_xdl_bf16.cpp
new file mode 100644
index 0000000000..016704ea04
--- /dev/null
+++ b/example/09_convnd_fwd/convnd_fwd_xdl_bf16.cpp
@@ -0,0 +1,227 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "convnd_fwd_common.hpp"
+
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+
+using InDataType       = ck::bhalf_t;
+using WeiDataType      = ck::bhalf_t;
+using AccDataType      = float;
+using CShuffleDataType = float;
+using OutDataType      = ck::bhalf_t;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvSpec =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+template <ck::index_t NDimSpatial, typename InLayout, typename WeiLayout, typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        256,         // NPerBlock
+        32,          // KPerBlock
+        8,           // AK1
+        8,           // BK1
+        32,          // MPerXdl
+        32,          // NPerXdl
+        2,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        8,           // ABlockTransferSrcScalarPerVector
+        8,           // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        8,           // BBlockTransferSrcScalarPerVector
+        8,           // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 32, 1, 8>,
+        8>;
+
+int main(int argc, char* argv[])
+{
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
+
+    bool do_verification = true;
+    int init_method      = 1;
+    bool time_kernel     = false;
+
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
+
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
+    {
+        do_verification = std::stoi(argv[1]);
+        init_method     = std::stoi(argv[2]);
+        time_kernel     = std::stoi(argv[3]);
+    }
+    else
+    {
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
+    }
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    if(conv_param.num_dim_spatial_ == 1)
+    {
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            1,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<1, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
+    }
+    else if(conv_param.num_dim_spatial_ == 2)
+    {
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            2,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<2, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
+    }
+    else if(conv_param.num_dim_spatial_ == 3)
+    {
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            3,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<3, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
+    }
+
+    return 0;
+}
diff --git a/example/09_convnd_fwd/convnd_fwd_xdl_fp16.cpp b/example/09_convnd_fwd/convnd_fwd_xdl_fp16.cpp
index 5866956105..c4df64abe4 100644
--- a/example/09_convnd_fwd/convnd_fwd_xdl_fp16.cpp
+++ b/example/09_convnd_fwd/convnd_fwd_xdl_fp16.cpp
@@ -1,342 +1,227 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include <cstdlib>
-#include <iostream>
-#include <numeric>
-#include <type_traits>
+#include "convnd_fwd_common.hpp"
 
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
 
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
 
-namespace {
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
+using InDataType       = ck::half_t;
+using WeiDataType      = ck::half_t;
+using AccDataType      = float;
+using CShuffleDataType = ck::half_t;
+using OutDataType      = ck::half_t;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
-using InLayout  = ck::tensor_layout::convolution::NHWC;
-using WeiLayout = ck::tensor_layout::convolution::KYXC;
-using OutLayout = ck::tensor_layout::convolution::NHWK;
-
 using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
 using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::UnaryConvert;
 
-static constexpr auto ConvFwdDefault =
+static constexpr auto ConvSpec =
     ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
 
-using DeviceConvFwdBasePtr =
-    ck::tensor_operation::device::DeviceConvFwdPtr<InElementOp, WeiElementOp, OutElementOp>;
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
-template <ck::index_t NumDimSpatial>
-using DeviceConvNDFwdInstance = ck::tensor_operation::device::
-    DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        // clang-format off
-        InDataType,         //
-        WeiDataType,        //
-        OutDataType,        //
-        AccDataType,        //
-        InElementOp,        // Input Elementwise Operation
-        WeiElementOp,       // Weights Elementwise Operation
-        OutElementOp,       // Output Elementwise Operation
-        ConvFwdDefault,     // ConvForwardSpecialization
-        NumDimSpatial,      // NumDimSpatial
-        256,                // BlockSize
-        128,                // MPerBlock
-        256,                // NPerBlock
-        4,                  // K0PerBlock
-        8,                  // K1
-        32,                 // MPerXdl
-        32,                 // NPerXdl
-        2,                  // MXdlPerWave
-        4,                  // NXdlPerWave
-        S<4, 64, 1>,        // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,         // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // ABlockTransferSrcAccessOrder
-        2,                  // ABlockTransferSrcVectorDim
-        8,                  // ABlockTransferSrcScalarPerVector
-        8,                  // ABlockTransferDstScalarPerVector_K1
-        true,               // ABlockLdsAddExtraM
-        S<4, 64, 1>,        // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,         // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // BBlockTransferSrcAccessOrder
-        2,                  // BBlockTransferSrcVectorDim
-        8,                  // BBlockTransferSrcScalarPerVector
-        8,                  // BBlockTransferDstScalarPerVector_K1
-        true,               // BBlockLdsAddExtraN
-        7,                  // CThreadTransferSrcDstVectorDim
-        1>;                 // CThreadTransferDstScalarPerVector
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvNDFwdInstance = ck::tensor_operation::host::ReferenceConvFwd<InDataType,
-                                                                                WeiDataType,
-                                                                                OutDataType,
-                                                                                InElementOp,
-                                                                                WeiElementOp,
-                                                                                OutElementOp,
-                                                                                NumDimSpatial>;
-
-DeviceConvFwdBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvNDFwdInstance<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvNDFwdInstance<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvNDFwdInstance<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "arg4: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int conv_args     = 3 + num_dim_spatial * 6;
-    int cmdline_nargs = conv_args + 5;
-    if(cmdline_nargs != argc)
-    {
-        print_use_msg();
-        exit(0);
-    }
-
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 5;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // anonymous namespace
+template <ck::index_t NDimSpatial, typename InLayout, typename WeiLayout, typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        256,         // NPerBlock
+        32,          // KPerBlock
+        8,           // AK1
+        8,           // BK1
+        32,          // MPerXdl
+        32,          // NPerXdl
+        2,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        8,           // ABlockTransferSrcScalarPerVector
+        8,           // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        8,           // BBlockTransferSrcScalarPerVector
+        8,           // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 32, 1, 8>,
+        8>;
 
 int main(int argc, char* argv[])
 {
-    using namespace ck::utils::conv;
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
 
     bool do_verification = true;
     int init_method      = 1;
     bool time_kernel     = false;
-    int num_dim_spatial  = 2;
 
-    ck::utils::conv::ConvParams params;
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
 
-    if(argc >= 5)
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
     {
         do_verification = std::stoi(argv[1]);
         init_method     = std::stoi(argv[2]);
         time_kernel     = std::stoi(argv[3]);
-        num_dim_spatial = std::stoi(argv[4]);
     }
-
-    if(argc >= 6)
+    else
     {
-        params = parse_conv_params(num_dim_spatial, argc, argv);
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
     }
 
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
 
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> input(get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> weights(get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> host_output(get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    Tensor<OutDataType> device_output(get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "input: " << input.mDesc << std::endl;
-    std::cout << "weights: " << weights.mDesc << std::endl;
-    std::cout << "output: " << host_output.mDesc << std::endl;
-
-    switch(init_method)
+    if(conv_param.num_dim_spatial_ == 1)
     {
-    case 0: break;
-    case 1:
-        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        weights.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-        break;
-    default:
-        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        weights.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            1,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<1, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * weights.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(input.mData.data());
-    wei_device_buf.ToDevice(weights.mData.data());
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{});
-
-    if(!conv->IsSupportedArgument(argument.get()))
+    else if(conv_param.num_dim_spatial_ == 2)
     {
-        throw std::runtime_error(
-            "wrong! device_conv with the specified compilation parameters does "
-            "not support this Conv problem");
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            2,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<2, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    float ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype = get_btype<InDataType, WeiDataType, OutDataType>(
-        params.N_,
-        params.C_,
-        params.K_,
-        params.input_spatial_lengths_,
-        params.filter_spatial_lengths_,
-        output_spatial_lengths);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s, "
-              << conv->GetTypeString() << std::endl;
-
-    if(do_verification)
+    else if(conv_param.num_dim_spatial_ == 3)
     {
-        auto verify_f = [&input, &weights, &host_output, &params, &out_device_buf, &device_output](
-                            const auto& ref_conv) {
-            auto ref_invoker  = ref_conv.MakeInvoker();
-            auto ref_argument = ref_conv.MakeArgument(input,
-                                                      weights,
-                                                      host_output,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
 
-            ref_invoker.Run(ref_argument);
-            out_device_buf.FromDevice(device_output.mData.data());
-            return ck::utils::check_err(
-                host_output.mData, device_output.mData, "Error: incorrect results!", 1e-5f, 1e-4f) ? 0 : 1;
-        };
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
 
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvNDFwdInstance<3>();
-            return verify_f(ref_conv);
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvNDFwdInstance<2>();
-            return verify_f(ref_conv);
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvNDFwdInstance<1>();
-            return verify_f(ref_conv);
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            3,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<3, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
+
     return 0;
 }
diff --git a/example/09_convnd_fwd/convnd_fwd_xdl_fp32.cpp b/example/09_convnd_fwd/convnd_fwd_xdl_fp32.cpp
index beb78c3e9b..bec59523e1 100644
--- a/example/09_convnd_fwd/convnd_fwd_xdl_fp32.cpp
+++ b/example/09_convnd_fwd/convnd_fwd_xdl_fp32.cpp
@@ -1,29 +1,17 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include <cstdlib>
-#include <iostream>
-#include <numeric>
-#include <type_traits>
+#include "convnd_fwd_common.hpp"
 
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
 
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
 
-namespace {
-
-using InDataType  = float;
-using WeiDataType = float;
-using OutDataType = float;
-using AccDataType = float;
+using InDataType       = float;
+using WeiDataType      = float;
+using AccDataType      = float;
+using CShuffleDataType = float;
+using OutDataType      = float;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -32,315 +20,208 @@ using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
 using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
 using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
 
-static constexpr auto ConvFwdDefault =
+static constexpr auto ConvSpec =
     ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
 
-using DeviceConvFwdBasePtr =
-    ck::tensor_operation::device::DeviceConvFwdPtr<InElementOp, WeiElementOp, OutElementOp>;
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
-template <ck::index_t NumDimSpatial>
-using DeviceConvNDFwdInstance = ck::tensor_operation::device::
-    DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        // clang-format off
-        InDataType,         //
-        WeiDataType,        //
-        OutDataType,        //
-        AccDataType,        //
-        InElementOp,        // Input Elementwise Operation
-        WeiElementOp,       // Weights Elementwise Operation
-        OutElementOp,       // Output Elementwise Operation
-        ConvFwdDefault,     // ConvForwardSpecialization
-        NumDimSpatial,      // NumDimSpatial
-        256,                // BlockSize
-        256,                // MPerBlock
-        128,                // NPerBlock
-        4,                  // K0PerBlock
-        4,                  // K1
-        32,                 // MPerXDL
-        32,                 // NPerXDL
-        4,                  // MXdlPerWave
-        2,                  // NXdlPerWave
-        S<4, 64, 1>,        // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,         // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // ABlockTransferSrcAccessOrder
-        2,                  // ABlockTransferSrcVectorDim
-        4,                  // ABlockTransferSrcScalarPerVector
-        4,                  // ABlockTransferDstScalarPerVector_K1
-        true,               // ABlockLdsAddExtraM
-        S<4, 64, 1>,        // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,         // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // BBlockTransferSrcAccessOrder
-        2,                  // BBlockTransferSrcVectorDim
-        4,                  // BBlockTransferSrcScalarPerVector
-        4,                  // BBlockTransferDstScalarPerVector_K1
-        true,               // BBlockTransferAddExtraN
-        7,                  // CThreadTransferSrcDstVectorDim
-        1>;                 // CThreadTransferDstScalarPerVector
-// clang-format on
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvNDFwdInstance = ck::tensor_operation::host::ReferenceConvFwd<InDataType,
-                                                                                WeiDataType,
-                                                                                OutDataType,
-                                                                                InElementOp,
-                                                                                WeiElementOp,
-                                                                                OutElementOp,
-                                                                                NumDimSpatial>;
-
-DeviceConvFwdBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvNDFwdInstance<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvNDFwdInstance<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvNDFwdInstance<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "arg4: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int conv_args     = 3 + num_dim_spatial * 6;
-    int cmdline_nargs = conv_args + 5;
-    if(cmdline_nargs != argc)
-    {
-        print_use_msg();
-        exit(0);
-    }
-
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 5;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // anonymous namespace
+template <ck::index_t NDimSpatial, typename InLayout, typename WeiLayout, typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        256,         // NPerBlock
+        16,          // KPerBlock
+        4,           // AK1
+        4,           // BK1
+        32,          // MPerXdl
+        32,          // NPerXdl
+        2,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        4,           // ABlockTransferSrcScalarPerVector
+        4,           // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        4,           // BBlockTransferSrcScalarPerVector
+        4,           // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 16, 1, 16>,
+        4>;
 
 int main(int argc, char* argv[])
 {
-    using namespace ck::utils::conv;
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
 
     bool do_verification = true;
     int init_method      = 1;
     bool time_kernel     = false;
-    int num_dim_spatial  = 2;
 
-    ck::utils::conv::ConvParams params;
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
 
-    if(argc >= 5)
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
     {
         do_verification = std::stoi(argv[1]);
         init_method     = std::stoi(argv[2]);
         time_kernel     = std::stoi(argv[3]);
-        num_dim_spatial = std::stoi(argv[4]);
     }
-
-    if(argc >= 6)
+    else
     {
-        params = parse_conv_params(num_dim_spatial, argc, argv);
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
     }
 
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
 
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> input(get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> weights(get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> host_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    Tensor<OutDataType> device_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "input: " << input.mDesc << std::endl;
-    std::cout << "weights: " << weights.mDesc << std::endl;
-    std::cout << "output: " << host_output.mDesc << std::endl;
-
-    switch(init_method)
+    if(conv_param.num_dim_spatial_ == 1)
     {
-    case 0: break;
-    case 1:
-        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        weights.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-        break;
-    default:
-        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        weights.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            1,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<1, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * weights.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(input.mData.data());
-    wei_device_buf.ToDevice(weights.mData.data());
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{});
-
-    if(!conv->IsSupportedArgument(argument.get()))
+    else if(conv_param.num_dim_spatial_ == 2)
     {
-        throw std::runtime_error(
-            "wrong! device_conv with the specified compilation parameters does "
-            "not support this Conv problem");
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            2,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<2, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    float ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype =
-        get_btype<InDataType, WeiDataType, OutDataType>(params.N_,
-                                                        params.C_,
-                                                        params.K_,
-                                                        params.input_spatial_lengths_,
-                                                        params.filter_spatial_lengths_,
-                                                        output_spatial_lengths);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
+    else if(conv_param.num_dim_spatial_ == 3)
     {
-        auto verify_f = [&input, &weights, &host_output, &params, &out_device_buf, &device_output](
-                            const auto& ref_conv) {
-            auto ref_invoker  = ref_conv.MakeInvoker();
-            auto ref_argument = ref_conv.MakeArgument(input,
-                                                      weights,
-                                                      host_output,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
 
-            ref_invoker.Run(ref_argument);
-            out_device_buf.FromDevice(device_output.mData.data());
-            return ck::utils::check_err(device_output.mData,
-                                        host_output.mData,
-                                        "Error: incorrect results!",
-                                        1e-5f,
-                                        1e-4f)
-                       ? 0
-                       : 1;
-        };
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
 
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvNDFwdInstance<3>();
-            return verify_f(ref_conv);
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvNDFwdInstance<2>();
-            return verify_f(ref_conv);
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvNDFwdInstance<1>();
-            return verify_f(ref_conv);
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            3,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<3, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
+
     return 0;
 }
diff --git a/example/09_convnd_fwd/convnd_fwd_xdl_fp64.cpp b/example/09_convnd_fwd/convnd_fwd_xdl_fp64.cpp
index cf1273fada..4c333f0e70 100644
--- a/example/09_convnd_fwd/convnd_fwd_xdl_fp64.cpp
+++ b/example/09_convnd_fwd/convnd_fwd_xdl_fp64.cpp
@@ -1,29 +1,17 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include <cstdlib>
-#include <iostream>
-#include <numeric>
-#include <type_traits>
+#include "convnd_fwd_common.hpp"
 
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
 
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
 
-namespace {
-
-using InDataType  = double;
-using WeiDataType = double;
-using OutDataType = double;
-using AccDataType = double;
+using InDataType       = double;
+using WeiDataType      = double;
+using AccDataType      = double;
+using CShuffleDataType = double;
+using OutDataType      = double;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -32,316 +20,208 @@ using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
 using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
 using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
 
-static constexpr auto ConvFwdDefault =
+static constexpr auto ConvSpec =
     ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
 
-using DeviceConvFwdBasePtr =
-    ck::tensor_operation::device::DeviceConvFwdPtr<InElementOp, WeiElementOp, OutElementOp>;
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
-template <ck::index_t NumDimSpatial>
-using DeviceConvNDFwdInstance = ck::tensor_operation::device::
-    DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        // clang-format off
-        InDataType,         // 
-        WeiDataType,        //
-        OutDataType,        //
-        AccDataType,        // 
-        InElementOp,        // Input Elementwise Operation
-        WeiElementOp,       // Weights Elementwise Operation
-        OutElementOp,       // Output Elementwise Operation
-        ConvFwdDefault,     // ConvForwardSpecialization
-        NumDimSpatial,      // NumDimSpatial
-        256,                // BlockSize
-        128,                // MPerBlock
-        128,                // NPerBlock
-        4,                  // K0PerBlock
-        2,                  // K1
-        16,                 // MPerXDL
-        16,                 // NPerXDL
-        4,                  // MXdlPerWave
-        4,                  // NXdlPerWave
-        S<4, 64, 1>,        // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,         // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // ABlockTransferSrcAccessOrder
-        2,                  // ABlockTransferSrcVectorDim
-        2,                  // ABlockTransferSrcScalarPerVector
-        2,                  // ABlockTransferDstScalarPerVector_K1
-        true,               // ABlockLdsAddExtraM
-        S<4, 64, 1>,        // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,         // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // BBlockTransferSrcAccessOrder
-        2,                  // BBlockTransferSrcVectorDim
-        2,                  // BBlockTransferSrcScalarPerVector
-        2,                  // BBlockTransferDstScalarPerVector_K1
-        true,               // BBlockTransferAddExtraN
-        7,                  // CThreadTransferSrcDstVectorDim
-        1>;                 // CThreadTransferDstScalarPerVector
-// clang-format on
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvNDFwdInstance = ck::tensor_operation::host::ReferenceConvFwd<InDataType,
-                                                                                WeiDataType,
-                                                                                OutDataType,
-                                                                                InElementOp,
-                                                                                WeiElementOp,
-                                                                                OutElementOp,
-                                                                                NumDimSpatial>;
-
-DeviceConvFwdBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvNDFwdInstance<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvNDFwdInstance<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvNDFwdInstance<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg3: run kernel # of times (>1)\n"
-              << "arg4: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int conv_args     = 3 + num_dim_spatial * 6;
-    int cmdline_nargs = conv_args + 5;
-    if(cmdline_nargs != argc)
-    {
-        print_use_msg();
-        exit(0);
-    }
-
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 5;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // anonymous namespace
+template <ck::index_t NDimSpatial, typename InLayout, typename WeiLayout, typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        128,         // NPerBlock
+        8,           // KPerBlock
+        2,           // AK1
+        2,           // BK1
+        16,          // MPerXdl
+        16,          // NPerXdl
+        4,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        2,           // ABlockTransferSrcScalarPerVector
+        2,           // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        2,           // BBlockTransferSrcScalarPerVector
+        2,           // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 16, 1, 16>,
+        1>;
 
 int main(int argc, char* argv[])
 {
-    using namespace ck::utils::conv;
+    namespace ctc = ck::tensor_layout::convolution;
 
-    bool do_verification = 0;
-    int init_method      = 0;
+    print_helper_msg();
+
+    bool do_verification = true;
+    int init_method      = 1;
     bool time_kernel     = false;
-    int num_dim_spatial  = 2;
 
-    ck::utils::conv::ConvParams params;
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
 
-    if(argc >= 5)
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
     {
         do_verification = std::stoi(argv[1]);
         init_method     = std::stoi(argv[2]);
         time_kernel     = std::stoi(argv[3]);
-        num_dim_spatial = std::stoi(argv[4]);
     }
-
-    if(argc >= 6)
+    else
     {
-        params = parse_conv_params(num_dim_spatial, argc, argv);
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
     }
 
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
 
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> input(get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> weights(get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> host_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    Tensor<OutDataType> device_output(
-        get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "input: " << input.mDesc << std::endl;
-    std::cout << "weights: " << weights.mDesc << std::endl;
-    std::cout << "output: " << host_output.mDesc << std::endl;
-
-    switch(init_method)
+    if(conv_param.num_dim_spatial_ == 1)
     {
-    case 0: break;
-    case 1:
-        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        weights.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-        break;
-    case 2:
-        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        weights.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
-        break;
-    default:
-        input.GenerateTensorValue(GeneratorTensor_1<InDataType>{1});
-        weights.GenerateTensorValue(GeneratorTensor_1<WeiDataType>{1});
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            1,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<1, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * weights.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(input.mData.data());
-    wei_device_buf.ToDevice(weights.mData.data());
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{});
-
-    if(!conv->IsSupportedArgument(argument.get()))
+    else if(conv_param.num_dim_spatial_ == 2)
     {
-        throw std::runtime_error(
-            "wrong! device_conv with the specified compilation parameters does "
-            "not support this Conv problem");
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            2,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<2, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    float ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype =
-        get_btype<InDataType, WeiDataType, OutDataType>(params.N_,
-                                                        params.C_,
-                                                        params.K_,
-                                                        params.input_spatial_lengths_,
-                                                        params.filter_spatial_lengths_,
-                                                        output_spatial_lengths);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
+    else if(conv_param.num_dim_spatial_ == 3)
     {
-        auto verify_f = [&input, &weights, &host_output, &params, &out_device_buf, &device_output](
-                            const auto& ref_conv) {
-            auto ref_invoker  = ref_conv.MakeInvoker();
-            auto ref_argument = ref_conv.MakeArgument(input,
-                                                      weights,
-                                                      host_output,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
 
-            ref_invoker.Run(ref_argument);
-            out_device_buf.FromDevice(device_output.mData.data());
-            ck::utils::check_err(
-                host_output.mData, device_output.mData, "Error: incorrect results!", 1e-5f, 1e-4f);
-        };
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
 
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvNDFwdInstance<3>();
-            verify_f(ref_conv);
-            break;
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvNDFwdInstance<2>();
-            verify_f(ref_conv);
-            break;
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvNDFwdInstance<1>();
-            verify_f(ref_conv);
-            break;
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            3,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<3, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
+
+    return 0;
 }
diff --git a/example/09_convnd_fwd/convnd_fwd_xdl_int8.cpp b/example/09_convnd_fwd/convnd_fwd_xdl_int8.cpp
index 3ca4b11766..18def79a5c 100644
--- a/example/09_convnd_fwd/convnd_fwd_xdl_int8.cpp
+++ b/example/09_convnd_fwd/convnd_fwd_xdl_int8.cpp
@@ -1,344 +1,227 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include <cstdlib>
-#include <iostream>
-#include <numeric>
-#include <type_traits>
+#include "convnd_fwd_common.hpp"
 
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
 
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
 
-namespace {
-
-using InDataType  = int8_t;
-using WeiDataType = int8_t;
-using OutDataType = int8_t;
-using AccDataType = int32_t;
+using InDataType       = int8_t;
+using WeiDataType      = int8_t;
+using AccDataType      = int32_t;
+using CShuffleDataType = int8_t;
+using OutDataType      = int8_t;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
-using InLayout  = ck::tensor_layout::convolution::NHWC;
-using WeiLayout = ck::tensor_layout::convolution::KYXC;
-using OutLayout = ck::tensor_layout::convolution::NHWK;
-
 using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
 using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
 using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
 
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
+static constexpr auto ConvSpec =
     ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
 
-using DeviceConvFwdBasePtr =
-    ck::tensor_operation::device::DeviceConvFwdPtr<InElementOp, WeiElementOp, OutElementOp>;
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
-template <ck::index_t NumDimSpatial>
-using DeviceConvNDFwdInstance = ck::tensor_operation::device::
-    DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        // clang-format off
-        InDataType,         //
-        WeiDataType,        //
-        OutDataType,        //
-        AccDataType,        //
-        InElementOp,        // Input Elementwise Operation
-        WeiElementOp,       // Weights Elementwise Operation
-        OutElementOp,       // Output Elementwise Operation
-        ConvFwdDefault,     // ConvForwardSpecialization
-        NumDimSpatial,      // NumDimSpatial
-        256,                // BlockSize
-        128,                // MPerBlock
-        256,                // NPerBlock
-        4,                  // K0PerBlock
-        16,                 // K1
-        32,                 // MPerXdl
-        32,                 // NPerXdl
-        2,                  // MXdlPerWave
-        4,                  // NXdlPerWave
-        S<4, 64, 1>,        // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,         // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // ABlockTransferSrcAccessOrder
-        2,                  // ABlockTransferSrcVectorDim
-        16,                 // ABlockTransferSrcScalarPerVector
-        16,                 // ABlockTransferDstScalarPerVector_K1
-        true,               // ABlockLdsAddExtraM
-        S<4, 64, 1>,        // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,         // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // BBlockTransferSrcAccessOrder
-        2,                  // BBlockTransferSrcVectorDim
-        16,                 // BBlockTransferSrcScalarPerVector
-        16,                 // BBlockTransferDstScalarPerVector_K1
-        true,               // BBlockLdsAddExtraN
-        7,                  // CThreadTransferSrcDstVectorDim
-        1>;                 // CThreadTransferDstScalarPerVector
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvNDFwdInstance = ck::tensor_operation::host::ReferenceConvFwd<InDataType,
-                                                                                WeiDataType,
-                                                                                OutDataType,
-                                                                                InElementOp,
-                                                                                WeiElementOp,
-                                                                                OutElementOp,
-                                                                                NumDimSpatial>;
-
-DeviceConvFwdBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvNDFwdInstance<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvNDFwdInstance<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvNDFwdInstance<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "arg4: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int conv_args     = 3 + num_dim_spatial * 6;
-    int cmdline_nargs = conv_args + 5;
-    if(cmdline_nargs != argc)
-    {
-        print_use_msg();
-        exit(0);
-    }
-
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 5;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // anonymous namespace
+template <ck::index_t NDimSpatial, typename InLayout, typename WeiLayout, typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        256,         // NPerBlock
+        64,          // KPerBlock
+        16,          // AK1
+        16,          // BK1
+        32,          // MPerXdl
+        32,          // NPerXdl
+        2,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        16,          // ABlockTransferSrcScalarPerVector
+        16,          // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        16,          // BBlockTransferSrcScalarPerVector
+        16,          // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 64, 1, 4>,
+        16>;
 
 int main(int argc, char* argv[])
 {
-    using namespace ck::utils::conv;
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
 
     bool do_verification = true;
     int init_method      = 1;
     bool time_kernel     = false;
-    int num_dim_spatial  = 2;
 
-    ck::utils::conv::ConvParams params;
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
 
-    if(argc >= 5)
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
     {
         do_verification = std::stoi(argv[1]);
         init_method     = std::stoi(argv[2]);
         time_kernel     = std::stoi(argv[3]);
-        num_dim_spatial = std::stoi(argv[4]);
     }
-
-    if(argc >= 6)
+    else
     {
-        params = parse_conv_params(num_dim_spatial, argc, argv);
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
     }
 
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
 
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> input(get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> weights(get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> host_output(get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-    Tensor<OutDataType> device_output(get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "input: " << input.mDesc << std::endl;
-    std::cout << "weights: " << weights.mDesc << std::endl;
-    std::cout << "output: " << host_output.mDesc << std::endl;
-
-    switch(init_method)
+    if(conv_param.num_dim_spatial_ == 1)
     {
-    case 0: break;
-    case 1:
-        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        weights.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-        break;
-    default:
-        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        weights.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            1,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<1, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * weights.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(input.mData.data());
-    wei_device_buf.ToDevice(weights.mData.data());
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{});
-
-    if(!conv->IsSupportedArgument(argument.get()))
+    else if(conv_param.num_dim_spatial_ == 2)
     {
-        throw std::runtime_error(
-            "wrong! device_conv with the specified compilation parameters does "
-            "not support this Conv problem");
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            2,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<2, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
-
-    float ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype = get_btype<InDataType, WeiDataType, OutDataType>(
-        params.N_,
-        params.C_,
-        params.K_,
-        params.input_spatial_lengths_,
-        params.filter_spatial_lengths_,
-        output_spatial_lengths);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
+    else if(conv_param.num_dim_spatial_ == 3)
     {
-        auto verify_f = [&input, &weights, &host_output, &params, &out_device_buf, &device_output](
-                            const auto& ref_conv) {
-            auto ref_invoker  = ref_conv.MakeInvoker();
-            auto ref_argument = ref_conv.MakeArgument(input,
-                                                      weights,
-                                                      host_output,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
 
-            ref_invoker.Run(ref_argument);
-            out_device_buf.FromDevice(device_output.mData.data());
-            return ck::utils::check_err(
-                host_output.mData, device_output.mData, "Error: incorrect results!", 1e-5f, 1e-4f) ? 0 : 1;
-        };
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
 
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvNDFwdInstance<3>();
-            return verify_f(ref_conv);
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvNDFwdInstance<2>();
-            return verify_f(ref_conv);
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvNDFwdInstance<1>();
-            return verify_f(ref_conv);
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_grouped_conv_fwd<
+            3,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<3, InLayout, WeiLayout, OutLayout>>(do_verification,
+                                                                               init_method,
+                                                                               time_kernel,
+                                                                               conv_param,
+                                                                               in_g_n_c_wis_desc,
+                                                                               wei_g_k_c_xs_desc,
+                                                                               out_g_n_k_wos_desc,
+                                                                               in_element_op,
+                                                                               wei_element_op,
+                                                                               out_element_op);
     }
+
     return 0;
 }
diff --git a/example/10_conv2d_bwd_data/CMakeLists.txt b/example/10_conv2d_bwd_data/CMakeLists.txt
deleted file mode 100644
index 17aca1481b..0000000000
--- a/example/10_conv2d_bwd_data/CMakeLists.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-add_example_executable(example_conv2d_bwd_data_xdl conv2d_bwd_data_xdl.cpp)
-target_link_libraries(example_conv2d_bwd_data_xdl PRIVATE conv_util)
diff --git a/example/10_conv2d_bwd_data/README.md b/example/10_conv2d_bwd_data/README.md
deleted file mode 100644
index 7503ff6d1e..0000000000
--- a/example/10_conv2d_bwd_data/README.md
+++ /dev/null
@@ -1,47 +0,0 @@
-# Instructions for ```example_conv2d_bwd_data_xdl``` Example
-
-
-## Run ```example_conv2d_bwd_data_xdl```
-```bash
-#arg1: verification (0=no, 1=yes)
-#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
-#arg3: run kernel # of times (>1)
-#arg4 to 18: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, RightPx
-./bin/example_conv2d_bwd_data_xdl 0 1 5
-```
-
-Result
-```
-in_n_c_hi_wi: dim 4, lengths {128, 256, 71, 71}, strides {1290496, 1, 18176, 256}
-wei_k_c_y_x: dim 4, lengths {256, 256, 3, 3}, strides {2304, 1, 768, 256}
-out_n_k_ho_wo: dim 4, lengths {128, 256, 36, 36}, strides {331776, 1, 9216, 256}
-arg.a_grid_desc_k0_m_k1_container_{128, 175232, 8}
-arg.b_grid_desc_k0_n_k1_container_{128, 256, 8}
-arg.c_grid_desc_m_n_container_{ 175232, 256}
-arg.c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_container_( 2738, 4, 2, 2, 4, 2 ) 
-launch_and_time_kernel: grid_dim {2738, 1, 1}, block_dim {256, 1, 1} 
-Warm up
-Start running 1 times...
-arg.a_grid_desc_k0_m_k1_container_{64, 175232, 8}
-arg.b_grid_desc_k0_n_k1_container_{64, 256, 8}
-arg.c_grid_desc_m_n_container_{ 175232, 256}
-arg.c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_container_( 2738, 4, 2, 2, 4, 2 ) 
-launch_and_time_kernel: grid_dim {2738, 1, 1}, block_dim {256, 1, 1} 
-Warm up
-Start running 1 times...
-arg.a_grid_desc_k0_m_k1_container_{64, 175232, 8}
-arg.b_grid_desc_k0_n_k1_container_{64, 256, 8}
-arg.c_grid_desc_m_n_container_{ 175232, 256}
-arg.c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_container_( 2738, 4, 2, 2, 4, 2 ) 
-launch_and_time_kernel: grid_dim {2738, 1, 1}, block_dim {256, 1, 1} 
-Warm up
-Start running 1 times...
-arg.a_grid_desc_k0_m_k1_container_{32, 175232, 8}
-arg.b_grid_desc_k0_n_k1_container_{32, 256, 8}
-arg.c_grid_desc_m_n_container_{ 175232, 256}
-arg.c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_container_( 2738, 4, 2, 2, 4, 2 ) 
-launch_and_time_kernel: grid_dim {2738, 1, 1}, block_dim {256, 1, 1} 
-Warm up
-Start running 1 times...
-Perf: 2.45966 ms, 79.5597 TFlops, 169.325 GB/s
-```
diff --git a/example/10_conv2d_bwd_data/conv2d_bwd_data_xdl.cpp b/example/10_conv2d_bwd_data/conv2d_bwd_data_xdl.cpp
deleted file mode 100644
index 340bc657fa..0000000000
--- a/example/10_conv2d_bwd_data/conv2d_bwd_data_xdl.cpp
+++ /dev/null
@@ -1,259 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp"
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-using DeviceConvBwdDataInstance = ck::tensor_operation::device::
-    DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        InDataType,     // InDataType
-        WeiDataType,    // WeiDataType
-        OutDataType,    // OutDataType
-        AccDataType,    // AccDataType
-        InElementOp,    // InElementwiseOperation
-        WeiElementOp,   // WeiElementwiseOperation
-        OutElementOp,   // OutElementwiseOperation
-        ConvBwdDefault, // ConvolutionBackwardDataSpecialization
-        256,            // BlockSize
-        128,            // MPerBlock
-        128,            // NPerBlock
-        4,              // K0PerBlock
-        8,              // K1
-        32,             // MPerXdl
-        32,             // NPerXdl
-        2,              // MXdlPerWave
-        2,              // NXdlPerWave
-        S<4, 64, 1>,    // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,     // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,     // ABlockTransferSrcAccessOrder
-        2,              // ABlockTransferSrcVectorDim
-        8,              // ABlockTransferSrcScalarPerVector
-        8,              // ABlockTransferDstScalarPerVector_K1
-        true,           // ABlockLdsAddExtraM
-        S<4, 64, 1>,    // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<2, 0, 1>,     // BBlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1>,     // BBlockTransferSrcAccessOrder
-        1,              // BBlockTransferSrcVectorDim
-        2,              // BBlockTransferSrcScalarPerVector
-        8,              // BBlockTransferDstScalarPerVector_K1
-        true,           // BBlockLdsAddExtraN
-        7,
-        1>; // GemmCThreadTransferDstScalarPerVector
-
-using ReferenceConvBwdInstance = ck::tensor_operation::host::ReferenceConvBwdData<InDataType,
-                                                                                  WeiDataType,
-                                                                                  OutDataType,
-                                                                                  AccDataType,
-                                                                                  InElementOp,
-                                                                                  WeiElementOp,
-                                                                                  OutElementOp>;
-
-int main(int argc, char* argv[])
-{
-    bool do_verification = true;
-    int init_method      = 1;
-    bool time_kernel     = false;
-
-    // Conv shape
-    ck::index_t N               = 128;
-    ck::index_t K               = 256;
-    ck::index_t C               = 256;
-    ck::index_t Y               = 3;
-    ck::index_t X               = 3;
-    ck::index_t Hi              = 71;
-    ck::index_t Wi              = 71;
-    ck::index_t conv_stride_h   = 2;
-    ck::index_t conv_stride_w   = 2;
-    ck::index_t conv_dilation_h = 1;
-    ck::index_t conv_dilation_w = 1;
-    ck::index_t in_left_pad_h   = 1;
-    ck::index_t in_left_pad_w   = 1;
-    ck::index_t in_right_pad_h  = 1;
-    ck::index_t in_right_pad_w  = 1;
-
-    if(argc == 4)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-    }
-    else if(argc == 19)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-
-        N               = std::stoi(argv[4]);
-        K               = std::stoi(argv[5]);
-        C               = std::stoi(argv[6]);
-        Y               = std::stoi(argv[7]);
-        X               = std::stoi(argv[8]);
-        Hi              = std::stoi(argv[9]);
-        Wi              = std::stoi(argv[10]);
-        conv_stride_h   = std::stoi(argv[11]);
-        conv_stride_w   = std::stoi(argv[12]);
-        conv_dilation_h = std::stoi(argv[13]);
-        conv_dilation_w = std::stoi(argv[14]);
-        in_left_pad_h   = std::stoi(argv[15]);
-        in_left_pad_w   = std::stoi(argv[16]);
-        in_right_pad_h  = std::stoi(argv[17]);
-        in_right_pad_w  = std::stoi(argv[18]);
-    }
-    else
-    {
-        printf("arg1: verification (0=no, 1=yes)\n");
-        printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
-        printf("arg3: time kernel (0=n0, 1=yes)\n");
-        printf("arg4 to 18: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-               "RightPx\n");
-        exit(0);
-    }
-
-    const ck::index_t YEff = (Y - 1) * conv_dilation_h + 1;
-    const ck::index_t XEff = (X - 1) * conv_dilation_w + 1;
-
-    const ck::index_t Ho = (Hi + in_left_pad_h + in_right_pad_h - YEff) / conv_stride_h + 1;
-    const ck::index_t Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
-
-    const std::vector<ck::index_t> conv_filter_strides{{conv_stride_h, conv_stride_w}};
-    const std::vector<ck::index_t> conv_filter_dilations{{conv_dilation_h, conv_dilation_w}};
-    const std::vector<ck::index_t> input_left_pads{{in_left_pad_h, in_left_pad_w}};
-    const std::vector<ck::index_t> input_right_pads{{in_right_pad_h, in_right_pad_w}};
-
-    // tensor layout
-    auto f_host_tensor_descriptor =
-        [](std::size_t N_, std::size_t C_, std::size_t H, std::size_t W) {
-            return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
-                                        std::vector<std::size_t>({C_ * H * W, 1, W * C_, C_}));
-        };
-
-    Tensor<OutDataType> out_n_k_ho_wo(f_host_tensor_descriptor(N, K, Ho, Wo));
-    Tensor<WeiDataType> wei_k_c_y_x(f_host_tensor_descriptor(K, C, Y, X));
-    Tensor<InDataType> in_n_c_hi_wi_host_result(f_host_tensor_descriptor(N, C, Hi, Wi));
-    Tensor<InDataType> in_n_c_hi_wi_device_result(f_host_tensor_descriptor(N, C, Hi, Wi));
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi_host_result.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
-        wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-        break;
-    default:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-        wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_1<WeiDataType>{1});
-    }
-
-    DeviceMem in_device_buf(sizeof(InDataType) *
-                            in_n_c_hi_wi_device_result.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
-
-    out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-    wei_device_buf.ToDevice(wei_k_c_y_x.mData.data());
-
-    // reset input to zero
-    in_n_c_hi_wi_device_result.GenerateTensorValue(GeneratorTensor_1<InDataType>{0});
-    in_device_buf.ToDevice(in_n_c_hi_wi_device_result.mData.data());
-
-    // do GEMM
-    auto conv     = DeviceConvBwdDataInstance{};
-    auto invoker  = conv.MakeInvoker();
-    auto argument = conv.MakeArgument(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                      static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                      static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                      N,
-                                      K,
-                                      C,
-                                      std::vector<ck::index_t>{{Hi, Wi}},
-                                      std::vector<ck::index_t>{{Y, X}},
-                                      std::vector<ck::index_t>{{Ho, Wo}},
-                                      conv_filter_strides,
-                                      conv_filter_dilations,
-                                      input_left_pads,
-                                      input_right_pads,
-                                      InElementOp{},
-                                      WeiElementOp{},
-                                      OutElementOp{});
-
-    if(!conv.IsSupportedArgument(argument))
-    {
-        throw std::runtime_error(
-            "wrong! device_conv with the specified compilation parameters does "
-            "not support this Conv problem");
-    }
-
-    float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = std::size_t(2) * N * K * Ho * Wo * C * Y * X;
-
-    std::size_t num_btype = sizeof(InDataType) * (N * C * Hi * Wi) +
-                            sizeof(WeiDataType) * (K * C * Y * X) +
-                            sizeof(OutDataType) * (N * K * Ho * Wo);
-
-    float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
-
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
-    {
-        auto ref_conv    = ReferenceConvBwdInstance{};
-        auto ref_invoker = ref_conv.MakeInvoker();
-
-        auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi_host_result,
-                                                  wei_k_c_y_x,
-                                                  out_n_k_ho_wo,
-                                                  conv_filter_strides,
-                                                  conv_filter_dilations,
-                                                  input_left_pads,
-                                                  input_right_pads,
-                                                  InElementOp{},
-                                                  WeiElementOp{},
-                                                  OutElementOp{});
-
-        ref_invoker.Run(ref_argument);
-
-        in_device_buf.FromDevice(in_n_c_hi_wi_device_result.mData.data());
-
-        return ck::utils::check_err(in_n_c_hi_wi_device_result.mData,
-                                    in_n_c_hi_wi_host_result.mData)
-                   ? 0
-                   : 1;
-    }
-    return 0;
-}
diff --git a/example/11_conv2d_bwd_weight/CMakeLists.txt b/example/11_conv2d_bwd_weight/CMakeLists.txt
deleted file mode 100644
index 3d771b5569..0000000000
--- a/example/11_conv2d_bwd_weight/CMakeLists.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-add_example_executable(example_conv2d_bwd_weight_xdl conv2d_bwd_weight_xdl.cpp)
-target_link_libraries(example_conv2d_bwd_weight_xdl PRIVATE conv_util)
diff --git a/example/11_conv2d_bwd_weight/README.md b/example/11_conv2d_bwd_weight/README.md
deleted file mode 100644
index c762742784..0000000000
--- a/example/11_conv2d_bwd_weight/README.md
+++ /dev/null
@@ -1,25 +0,0 @@
-# Instructions for ```example_conv2d_bwd_weight_xdl``` Example
-
-## Run ```example_conv2d_bwd_weight_xdl```
-```bash
-#arg1: verification (0=no, 1=yes)
-#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
-#arg3: run kernel # of times (>1)
-#arg4: is show log (0=no, 1=yes)
-#arg5 to 19: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, RightPx, split-k
-./bin/example_conv2d_bwd_weight_xdl 0 1 5 0 4
-```
-
-Result 
-```
-in_n_c_hi_wi: dim 4, lengths {128, 1024, 14, 14}, strides {200704, 1, 14336, 1024}
-wei_k_c_y_x: dim 4, lengths {256, 1024, 3, 3}, strides {9216, 1, 3072, 1024}
-out_n_k_ho_wo: dim 4, lengths {128, 256, 6, 6}, strides {9216, 1, 1536, 256}
-arg.a_grid_desc_kbatch_k0_m_k1_{4, 144, 256, 8}
-arg.b_grid_desc_kbatch_k0_n_k1_{4, 144, 9216, 8}
-arg.c_grid_desc_m_n_{ 256, 9216}
-launch_and_time_kernel: grid_dim {576, 1, 1}, block_dim {256, 1, 1} 
-Warm up
-Start running 5 times...
-Perf: 0.401084 ms, 54.2112 TFlops, 145.75 GB/s
-```
diff --git a/example/11_conv2d_bwd_weight/conv2d_bwd_weight_xdl.cpp b/example/11_conv2d_bwd_weight/conv2d_bwd_weight_xdl.cpp
deleted file mode 100644
index e47ae66152..0000000000
--- a/example/11_conv2d_bwd_weight/conv2d_bwd_weight_xdl.cpp
+++ /dev/null
@@ -1,300 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp"
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InLayout  = ck::tensor_layout::convolution::NHWC;
-using WeiLayout = ck::tensor_layout::convolution::KYXC;
-using OutLayout = ck::tensor_layout::convolution::NHWK;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-
-// clang-format off
-using DeviceConvBwdWeightInstance = ck::tensor_operation::device::
-    DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        InDataType,                       // InDataType
-        WeiDataType,                      // WeiDataType
-        OutDataType,                      // OutDataType
-        AccDataType,                      // AccDataType
-        InElementOp,                      // InElementwiseOperation
-        WeiElementOp,                     // WeiElementwiseOperation
-        OutElementOp,                     // OutElementwiseOperation
-        256,                              // BlockSize
-        128,                              // MPerBlock
-        128,                              // NPerBlock
-        4,                                // K0PerBlock
-        8,                                // K1
-        32,                               // MPerXdl
-        32,                               // NPerXdl
-        2,                                // MXdlPerWave
-        2,                                // NXdlPerWave
-        S<1, 4, 16, 4>,                   // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<0, 3, 1, 2>,                    // ABlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1, 3>,                    // ABlockTransferSrcAccessOrder
-        2,                                // ABlockTransferSrcVectorDim
-        8,                                // ABlockTransferSrcScalarPerVector
-        2,                                // ABlockTransferDstScalarPerVector_K1
-        true,                             // ABlockLdsAddExtraM
-        S<1, 4, 16, 4>,                   // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<0, 3, 1, 2>,                    // BBlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1, 3>,                    // BBlockTransferSrcAccessOrder
-        2,                                // BBlockTransferSrcVectorDim
-        8,                                // BBlockTransferSrcScalarPerVector
-        2,                                // BBlockTransferDstScalarPerVector_K1
-        true,                             // BBlockLdsAddExtraN
-        1,                                // CShuffleMXdlPerWavePerShuffle
-        1,                                // CShuffleNXdlPerWavePerShuffle
-        S<1, 32, 1, 4>,                   // CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock
-        8>;                               // CBlockTransferScalarPerVector_NWaveNPerXdl
-// clang-format on
-
-using ReferenceConvBwdWeightInstance =
-    ck::tensor_operation::host::ReferenceConvBwdWeight<InDataType,
-                                                       WeiDataType,
-                                                       OutDataType,
-                                                       InElementOp,
-                                                       WeiElementOp,
-                                                       OutElementOp>;
-
-int main(int argc, char* argv[])
-{
-    bool do_verification = true;
-    int init_method      = 1;
-    bool time_kernel     = false;
-    int do_log           = 0;
-    int split_k          = 4;
-
-    // Conv shape
-    ck::index_t N               = 128;
-    ck::index_t K               = 256;
-    ck::index_t C               = 1024;
-    ck::index_t Y               = 3;
-    ck::index_t X               = 3;
-    ck::index_t Hi              = 14;
-    ck::index_t Wi              = 14;
-    ck::index_t conv_stride_h   = 2;
-    ck::index_t conv_stride_w   = 2;
-    ck::index_t conv_dilation_h = 1;
-    ck::index_t conv_dilation_w = 1;
-    ck::index_t in_left_pad_h   = 0;
-    ck::index_t in_left_pad_w   = 0;
-    ck::index_t in_right_pad_h  = 0;
-    ck::index_t in_right_pad_w  = 0;
-
-    if(argc == 6)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        do_log          = std::stoi(argv[4]);
-        split_k         = std::stoi(argv[5]);
-    }
-    else if(argc == 21)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        do_log          = std::stoi(argv[4]);
-        split_k         = std::stoi(argv[5]);
-
-        N               = std::stoi(argv[6]);
-        K               = std::stoi(argv[7]);
-        C               = std::stoi(argv[8]);
-        Y               = std::stoi(argv[9]);
-        X               = std::stoi(argv[10]);
-        Hi              = std::stoi(argv[11]);
-        Wi              = std::stoi(argv[12]);
-        conv_stride_h   = std::stoi(argv[13]);
-        conv_stride_w   = std::stoi(argv[14]);
-        conv_dilation_h = std::stoi(argv[15]);
-        conv_dilation_w = std::stoi(argv[16]);
-        in_left_pad_h   = std::stoi(argv[17]);
-        in_left_pad_w   = std::stoi(argv[18]);
-        in_right_pad_h  = std::stoi(argv[19]);
-        in_right_pad_w  = std::stoi(argv[20]);
-    }
-    else
-    {
-        printf("arg1: verification (0=no, 1=yes)\n");
-        printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
-        printf("arg3: time kernel (0=n0, 1=yes)\n");
-        printf("arg4: is show log (0=no, 1=yes)\n");
-        printf("arg5: split-k \n");
-        printf("arg6 to 19: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-               "RightPx\n");
-        exit(0);
-    }
-
-    const ck::index_t YEff = (Y - 1) * conv_dilation_h + 1;
-    const ck::index_t XEff = (X - 1) * conv_dilation_w + 1;
-
-    const ck::index_t Ho = (Hi + in_left_pad_h + in_right_pad_h - YEff) / conv_stride_h + 1;
-    const ck::index_t Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
-
-    const std::vector<ck::index_t> conv_filter_strides{{conv_stride_h, conv_stride_w}};
-    const std::vector<ck::index_t> conv_filter_dilations{{conv_dilation_h, conv_dilation_w}};
-    const std::vector<ck::index_t> input_left_pads{{in_left_pad_h, in_left_pad_w}};
-    const std::vector<ck::index_t> input_right_pads{{in_right_pad_h, in_right_pad_w}};
-
-    // tensor layout
-    auto f_host_tensor_descriptor = [](std::size_t N_,
-                                       std::size_t C_,
-                                       std::size_t H,
-                                       std::size_t W,
-                                       auto layout) {
-        if constexpr(ck::is_same<decltype(layout), ck::tensor_layout::convolution::NCHW>::value ||
-                     ck::is_same<decltype(layout), ck::tensor_layout::convolution::KCYX>::value ||
-                     ck::is_same<decltype(layout), ck::tensor_layout::convolution::NKHW>::value)
-        {
-            return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
-                                        std::vector<std::size_t>({C_ * H * W, H * W, W, 1}));
-        }
-        else if constexpr(ck::is_same<decltype(layout),
-                                      ck::tensor_layout::convolution::NHWC>::value ||
-                          ck::is_same<decltype(layout),
-                                      ck::tensor_layout::convolution::KYXC>::value ||
-                          ck::is_same<decltype(layout),
-                                      ck::tensor_layout::convolution::NHWK>::value)
-        {
-            return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
-                                        std::vector<std::size_t>({C_ * H * W, 1, W * C_, C_}));
-        }
-    };
-
-    Tensor<InDataType> in_n_c_hi_wi(f_host_tensor_descriptor(N, C, Hi, Wi, InLayout{}));
-    Tensor<WeiDataType> wei_k_c_y_x_host_result(f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
-    Tensor<WeiDataType> wei_k_c_y_x_device_result(
-        f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
-    Tensor<OutDataType> out_n_k_ho_wo(f_host_tensor_descriptor(N, K, Ho, Wo, OutLayout{}));
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_host_result.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
-        break;
-    default:
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_1<InDataType>{1});
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-    }
-    wei_k_c_y_x_device_result.GenerateTensorValue(GeneratorTensor_1<WeiDataType>{0});
-
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) *
-                             wei_k_c_y_x_device_result.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
-    out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-    wei_device_buf.ToDevice(wei_k_c_y_x_device_result.mData.data());
-
-    // do GEMM
-    auto conv     = DeviceConvBwdWeightInstance{};
-    auto invoker  = conv.MakeInvoker();
-    auto argument = conv.MakeArgument(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                      static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                      static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                      N,
-                                      K,
-                                      C,
-                                      std::vector<ck::index_t>{{Hi, Wi}},
-                                      std::vector<ck::index_t>{{Y, X}},
-                                      std::vector<ck::index_t>{{Ho, Wo}},
-                                      conv_filter_strides,
-                                      conv_filter_dilations,
-                                      input_left_pads,
-                                      input_right_pads,
-                                      InElementOp{},
-                                      WeiElementOp{},
-                                      OutElementOp{},
-                                      split_k);
-
-    if(!conv.IsSupportedArgument(argument))
-    {
-        std::cout << "wrong! device_conv with the specified compilation parameters does "
-                     "not support this Conv problem"
-                  << std::endl;
-        return 1;
-    }
-
-    float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = std::size_t(2) * N * K * Ho * Wo * C * Y * X;
-
-    std::size_t num_btype = sizeof(InDataType) * (N * C * Hi * Wi) +
-                            sizeof(WeiDataType) * (K * C * Y * X) +
-                            sizeof(OutDataType) * (N * K * Ho * Wo);
-
-    float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
-
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
-    {
-        auto ref_conv    = ReferenceConvBwdWeightInstance{};
-        auto ref_invoker = ref_conv.MakeInvoker();
-
-        auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi,
-                                                  wei_k_c_y_x_host_result,
-                                                  out_n_k_ho_wo,
-                                                  conv_filter_strides,
-                                                  conv_filter_dilations,
-                                                  input_left_pads,
-                                                  input_right_pads,
-                                                  InElementOp{},
-                                                  WeiElementOp{},
-                                                  OutElementOp{});
-
-        ref_invoker.Run(ref_argument);
-
-        wei_device_buf.FromDevice(wei_k_c_y_x_device_result.mData.data());
-
-        if(do_log)
-        {
-            LogRangeAsType<float>(std::cout << "out: ", out_n_k_ho_wo.mData, ",") << std::endl;
-            LogRangeAsType<float>(std::cout << "in : ", in_n_c_hi_wi.mData, ",") << std::endl;
-            LogRangeAsType<float>(
-                std::cout << "wei_device(after): ", wei_k_c_y_x_device_result.mData, ",")
-                << std::endl;
-            LogRangeAsType<float>(std::cout << "wei_host  : ", wei_k_c_y_x_host_result.mData, ",")
-                << std::endl;
-        }
-        return ck::utils::check_err(wei_k_c_y_x_device_result.mData, wei_k_c_y_x_host_result.mData)
-                   ? 0
-                   : 1;
-    }
-    return 0;
-}
diff --git a/example/12_reduce/reduce_blockwise.cpp b/example/12_reduce/reduce_blockwise.cpp
index 0a93af5358..a410f2a055 100644
--- a/example/12_reduce/reduce_blockwise.cpp
+++ b/example/12_reduce/reduce_blockwise.cpp
@@ -13,11 +13,11 @@
 #include "ck/tensor_operation/gpu/device/device_reduce_multiblock.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/host_tensor/host_common_util.hpp"
-#include "ck/library/host_tensor/host_reduction.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/host_common_util.hpp"
+#include "ck/library/utility/host_reduction.hpp"
 
 using namespace ck;
 using namespace ck::tensor_operation::device;
@@ -230,13 +230,13 @@ int main(int argc, char* argv[])
         }
 
         if(beta != 0.0f)
-            for(size_t i = 0; i < out_ref.mDesc.GetElementSpace(); i++)
+            for(size_t i = 0; i < out_ref.mDesc.GetElementSpaceSize(); i++)
                 out.mData[i] = out_ref.mData[i];
     };
 
     // these buffers are usually provided by the user application
-    DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpace());
-    DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpace());
+    DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+    DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
 
     in_dev.ToDevice(in.mData.data());
 
diff --git a/example/12_reduce/reduce_blockwise_two_call.cpp b/example/12_reduce/reduce_blockwise_two_call.cpp
index 727c5877c5..df58cc276b 100644
--- a/example/12_reduce/reduce_blockwise_two_call.cpp
+++ b/example/12_reduce/reduce_blockwise_two_call.cpp
@@ -14,11 +14,11 @@
 #include "ck/tensor_operation/gpu/device/device_reduce_multiblock.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/host_tensor/host_common_util.hpp"
-#include "ck/library/host_tensor/host_reduction.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/host_common_util.hpp"
+#include "ck/library/utility/host_reduction.hpp"
 
 using namespace ck;
 using namespace ck::tensor_operation::device;
@@ -171,13 +171,13 @@ int main(int argc, char* argv[])
         }
 
         if(beta != 0.0f)
-            for(size_t i = 0; i < out_ref.mDesc.GetElementSpace(); i++)
+            for(size_t i = 0; i < out_ref.mDesc.GetElementSpaceSize(); i++)
                 out.mData[i] = out_ref.mData[i];
     };
 
-    DeviceMem in_1_dev(sizeof(InOutDataType) * in_1.mDesc.GetElementSpace());
-    DeviceMem in_2_dev(sizeof(InOutDataType) * in_2.mDesc.GetElementSpace());
-    DeviceMem out_dev(sizeof(InOutDataType) * out.mDesc.GetElementSpace());
+    DeviceMem in_1_dev(sizeof(InOutDataType) * in_1.mDesc.GetElementSpaceSize());
+    DeviceMem in_2_dev(sizeof(InOutDataType) * in_2.mDesc.GetElementSpaceSize());
+    DeviceMem out_dev(sizeof(InOutDataType) * out.mDesc.GetElementSpaceSize());
 
     in_1_dev.ToDevice(in_1.mData.data());
 
diff --git a/example/13_pool2d_fwd/pool2d_fwd_common.hpp b/example/13_pool2d_fwd/pool2d_fwd_common.hpp
index ac1d0f3a41..32b66934a0 100644
--- a/example/13_pool2d_fwd/pool2d_fwd_common.hpp
+++ b/example/13_pool2d_fwd/pool2d_fwd_common.hpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 template <typename InDataType,
           typename OutDataType,
@@ -204,10 +204,11 @@ bool pool_test(bool do_verification,
     default: in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_3<InDataType>{-5.0, 5.0});
     }
 
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_c_ho_wo_device.mDesc.GetElementSpace());
+    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) *
+                             out_n_c_ho_wo_device.mDesc.GetElementSpaceSize());
     DeviceMem out_indices_device_buf(sizeof(IndexDataType) *
-                                     out_indices_n_c_ho_wo_device.mDesc.GetElementSpace());
+                                     out_indices_n_c_ho_wo_device.mDesc.GetElementSpaceSize());
 
     in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
 
diff --git a/example/14_gemm_xdl_requant_relu_requant/gemm_xdl_requant_relu_requant_int8.cpp b/example/14_gemm_xdl_requant_relu_requant/gemm_xdl_requant_relu_requant_int8.cpp
index 379be22ad1..d3afa3865d 100644
--- a/example/14_gemm_xdl_requant_relu_requant/gemm_xdl_requant_relu_requant_int8.cpp
+++ b/example/14_gemm_xdl_requant_relu_requant/gemm_xdl_requant_relu_requant_int8.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_xdl_cshuffle.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -190,9 +190,9 @@ int main(int argc, char* argv[])
         b_k_n.GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/15_grouped_gemm/grouped_gemm_xdl_fp16.cpp b/example/15_grouped_gemm/grouped_gemm_xdl_fp16.cpp
index b3ef605f68..a107b6b8c8 100644
--- a/example/15_grouped_gemm/grouped_gemm_xdl_fp16.cpp
+++ b/example/15_grouped_gemm/grouped_gemm_xdl_fp16.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -36,9 +36,10 @@ using CShuffleDataType = F16;
 using DsDataType       = ck::Tuple<>;
 using EDataType        = F16;
 
-using ALayout = Row;
-using BLayout = Col;
-using ELayout = Row;
+using ALayout  = Row;
+using BLayout  = Col;
+using DsLayout = ck::Tuple<>;
+using ELayout  = Row;
 
 using AElementOp   = PassThrough;
 using BElementOp   = PassThrough;
@@ -46,13 +47,13 @@ using CDEElementOp = PassThrough;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
-using DeviceGemmInstance = ck::tensor_operation::device::DeviceGroupedGemmXdl
+using DeviceGemmInstance = ck::tensor_operation::device::DeviceGroupedGemm_Xdl
     // clang-format off
-//######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-//######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-//######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-//######|        |        |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
+//######| ALayout| BLayout| DsLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+//######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+//######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 using ReferenceGemmInstance = ck::tensor_operation::host::ReferenceGemm<ADataType,
@@ -178,12 +179,12 @@ int main(int argc, char* argv[])
 
     for(std::size_t i = 0; i < gemm_descs.size(); i++)
     {
-        a_tensors_device.emplace_back(
-            std::make_unique<DeviceMem>(sizeof(ADataType) * a_tensors[i].mDesc.GetElementSpace()));
-        b_tensors_device.emplace_back(
-            std::make_unique<DeviceMem>(sizeof(BDataType) * b_tensors[i].mDesc.GetElementSpace()));
+        a_tensors_device.emplace_back(std::make_unique<DeviceMem>(
+            sizeof(ADataType) * a_tensors[i].mDesc.GetElementSpaceSize()));
+        b_tensors_device.emplace_back(std::make_unique<DeviceMem>(
+            sizeof(BDataType) * b_tensors[i].mDesc.GetElementSpaceSize()));
         c_tensors_device.emplace_back(std::make_unique<DeviceMem>(
-            sizeof(EDataType) * c_device_tensors[i].mDesc.GetElementSpace()));
+            sizeof(EDataType) * c_device_tensors[i].mDesc.GetElementSpaceSize()));
 
         a_tensors_device[i]->ToDevice(a_tensors[i].mData.data());
         b_tensors_device[i]->ToDevice(b_tensors[i].mData.data());
diff --git a/example/16_gemm_reduce/gemm_reduce_xdl_max_fp16.cpp b/example/16_gemm_reduce/gemm_reduce_xdl_max_fp16.cpp
index d20c863c4b..457a7ef492 100644
--- a/example/16_gemm_reduce/gemm_reduce_xdl_max_fp16.cpp
+++ b/example/16_gemm_reduce/gemm_reduce_xdl_max_fp16.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_reduce_xdl_cshuffle.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -173,11 +173,11 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce_device_buf(sizeof(ReduceDataType) *
-                                reduce_m_device_result.mDesc.GetElementSpace());
+                                reduce_m_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/16_gemm_reduce/gemm_reduce_xdl_mean_squaremean_fp16.cpp b/example/16_gemm_reduce/gemm_reduce_xdl_mean_squaremean_fp16.cpp
index ddfaa9d752..2ebd096679 100644
--- a/example/16_gemm_reduce/gemm_reduce_xdl_mean_squaremean_fp16.cpp
+++ b/example/16_gemm_reduce/gemm_reduce_xdl_mean_squaremean_fp16.cpp
@@ -14,9 +14,9 @@
 #include "ck/utility/reduction_operator.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -188,13 +188,13 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce0_device_buf(sizeof(ReduceDataType) *
-                                 reduce0_m_device_result.mDesc.GetElementSpace());
+                                 reduce0_m_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce1_device_buf(sizeof(ReduceDataType) *
-                                 reduce1_m_device_result.mDesc.GetElementSpace());
+                                 reduce1_m_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/17_convnd_bwd_data/CMakeLists.txt b/example/17_convnd_bwd_data/CMakeLists.txt
new file mode 100644
index 0000000000..35f320bd34
--- /dev/null
+++ b/example/17_convnd_bwd_data/CMakeLists.txt
@@ -0,0 +1,2 @@
+add_example_executable(example_convnd_bwd_data_xdl_fp16 convnd_bwd_data_xdl_fp16.cpp)
+target_link_libraries(example_convnd_bwd_data_xdl_fp16 PRIVATE utility)
diff --git a/example/17_convnd_bwd_data_xdl/README.md b/example/17_convnd_bwd_data/README.md
similarity index 100%
rename from example/17_convnd_bwd_data_xdl/README.md
rename to example/17_convnd_bwd_data/README.md
diff --git a/example/17_convnd_bwd_data/convnd_bwd_data_common.hpp b/example/17_convnd_bwd_data/convnd_bwd_data_common.hpp
new file mode 100644
index 0000000000..061c6e9eb1
--- /dev/null
+++ b/example/17_convnd_bwd_data/convnd_bwd_data_common.hpp
@@ -0,0 +1,149 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp"
+
+void print_helper_msg()
+{
+    std::cout << "arg1: verification (0=no, 1=yes)\n"
+              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
+              << "arg3: time kernel (0=no, 1=yes)\n"
+              << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+}
+
+template <ck::index_t NDimSpatial,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementOp,
+          typename WeiElementOp,
+          typename OutElementOp,
+          typename DeviceConvNdBwdDataInstance>
+int run_conv_bwd_data(bool do_verification,
+                      int init_method,
+                      bool time_kernel,
+                      const ck::utils::conv::ConvParam& conv_param,
+                      const HostTensorDescriptor& in_g_n_c_wis_desc,
+                      const HostTensorDescriptor& wei_g_k_c_xs_desc,
+                      const HostTensorDescriptor& out_g_n_k_wos_desc,
+                      const InElementOp& in_element_op,
+                      const WeiElementOp& wei_element_op,
+                      const OutElementOp& out_element_op)
+{
+    Tensor<InDataType> in_host(in_g_n_c_wis_desc);
+    Tensor<InDataType> in_device(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> wei(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> out(out_g_n_k_wos_desc);
+
+    std::cout << "in: " << in_host.mDesc << std::endl;
+    std::cout << "wei: " << wei.mDesc << std::endl;
+    std::cout << "out: " << out.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        out.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
+        wei.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        break;
+    default:
+        out.GenerateTensorValue(GeneratorTensor_3<OutDataType>{0.0, 1.0});
+        wei.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * in_device.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
+
+    out_device_buf.ToDevice(out.mData.data());
+    wei_device_buf.ToDevice(wei.mData.data());
+
+    // reset input to zero
+    in_device_buf.SetZero();
+
+    // do GEMM
+    auto conv     = DeviceConvNdBwdDataInstance{};
+    auto invoker  = conv.MakeInvoker();
+    auto argument = conv.MakeArgument(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
+                                      static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
+                                      static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
+                                      conv_param.N_,
+                                      conv_param.K_,
+                                      conv_param.C_,
+                                      conv_param.input_spatial_lengths_,
+                                      conv_param.filter_spatial_lengths_,
+                                      conv_param.GetOutputSpatialLengths(),
+                                      conv_param.conv_filter_strides_,
+                                      conv_param.conv_filter_dilations_,
+                                      conv_param.input_left_pads_,
+                                      conv_param.input_right_pads_,
+                                      in_element_op,
+                                      wei_element_op,
+                                      out_element_op);
+
+    if(!conv.IsSupportedArgument(argument))
+    {
+        throw std::runtime_error(
+            "wrong! device_conv with the specified compilation parameters does "
+            "not support this Conv problem");
+    }
+
+    float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
+
+    std::size_t flop      = conv_param.GetFlops();
+    std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+    float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
+
+    float gb_per_sec = num_btype / 1.E6 / ave_time;
+
+    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
+              << std::endl;
+
+    if(do_verification)
+    {
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvBwdData<NDimSpatial,
+                                                                         InDataType,
+                                                                         WeiDataType,
+                                                                         OutDataType,
+                                                                         InElementOp,
+                                                                         WeiElementOp,
+                                                                         OutElementOp>();
+
+        auto ref_invoker = ref_conv.MakeInvoker();
+
+        auto ref_argument = ref_conv.MakeArgument(in_host,
+                                                  wei,
+                                                  out,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  in_element_op,
+                                                  wei_element_op,
+                                                  out_element_op);
+
+        ref_invoker.Run(ref_argument);
+
+        in_device_buf.FromDevice(in_device.mData.data());
+
+        return ck::utils::check_err(in_device.mData, in_host.mData) ? 0 : 1;
+    }
+
+    return 0;
+}
diff --git a/example/17_convnd_bwd_data/convnd_bwd_data_xdl_fp16.cpp b/example/17_convnd_bwd_data/convnd_bwd_data_xdl_fp16.cpp
new file mode 100644
index 0000000000..392e961b06
--- /dev/null
+++ b/example/17_convnd_bwd_data/convnd_bwd_data_xdl_fp16.cpp
@@ -0,0 +1,207 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "convnd_bwd_data_common.hpp"
+
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+
+using InDataType  = ck::half_t;
+using WeiDataType = ck::half_t;
+using OutDataType = ck::half_t;
+using AccDataType = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+template <ck::index_t NDimSpatial>
+using DeviceConvNdBwdDataInstance = ck::tensor_operation::device::DeviceConvNdBwdDataNwcKxcNwk_Xdl<
+    NDimSpatial,    // NDimSpatial
+    InDataType,     // InDataType
+    WeiDataType,    // WeiDataType
+    OutDataType,    // OutDataType
+    AccDataType,    // AccDataType
+    InElementOp,    // InElementwiseOperation
+    WeiElementOp,   // WeiElementwiseOperation
+    OutElementOp,   // OutElementwiseOperation
+    ConvBwdDefault, // ConvolutionBackwardDataSpecialization
+    256,            // BlockSize
+    128,            // MPerBlock
+    128,            // NPerBlock
+    4,              // K0PerBlock
+    8,              // K1
+    32,             // MPerXdl
+    32,             // NPerXdl
+    2,              // MXdlPerWave
+    2,              // NXdlPerWave
+    S<4, 64, 1>,    // ABlockTransferThreadClusterLengths_K0_M_K1
+    S<1, 0, 2>,     // ABlockTransferThreadClusterArrangeOrder
+    S<1, 0, 2>,     // ABlockTransferSrcAccessOrder
+    2,              // ABlockTransferSrcVectorDim
+    8,              // ABlockTransferSrcScalarPerVector
+    8,              // ABlockTransferDstScalarPerVector_K1
+    true,           // ABlockLdsAddExtraM
+    S<4, 64, 1>,    // BBlockTransferThreadClusterLengths_K0_N_K1
+    S<2, 0, 1>,     // BBlockTransferThreadClusterArrangeOrder
+    S<0, 2, 1>,     // BBlockTransferSrcAccessOrder
+    1,              // BBlockTransferSrcVectorDim
+    2,              // BBlockTransferSrcScalarPerVector
+    8,              // BBlockTransferDstScalarPerVector_K1
+    true,           // BBlockLdsAddExtraN
+    7,
+    1>; // GemmCThreadTransferDstScalarPerVector
+
+int main(int argc, char* argv[])
+{
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
+
+    bool do_verification = true;
+    int init_method      = 1;
+    bool time_kernel     = false;
+
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 128, 256, 256, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
+
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
+    {
+        do_verification = std::stoi(argv[1]);
+        init_method     = std::stoi(argv[2]);
+        time_kernel     = std::stoi(argv[3]);
+    }
+    else
+    {
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
+    }
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    if(conv_param.num_dim_spatial_ == 1)
+    {
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_data<1,
+                                 InDataType,
+                                 WeiDataType,
+                                 OutDataType,
+                                 InElementOp,
+                                 WeiElementOp,
+                                 OutElementOp,
+                                 DeviceConvNdBwdDataInstance<1>>(do_verification,
+                                                                 init_method,
+                                                                 time_kernel,
+                                                                 conv_param,
+                                                                 in_g_n_c_wis_desc,
+                                                                 wei_g_k_c_xs_desc,
+                                                                 out_g_n_k_wos_desc,
+                                                                 in_element_op,
+                                                                 wei_element_op,
+                                                                 out_element_op);
+    }
+    else if(conv_param.num_dim_spatial_ == 2)
+    {
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_data<2,
+                                 InDataType,
+                                 WeiDataType,
+                                 OutDataType,
+                                 InElementOp,
+                                 WeiElementOp,
+                                 OutElementOp,
+                                 DeviceConvNdBwdDataInstance<2>>(do_verification,
+                                                                 init_method,
+                                                                 time_kernel,
+                                                                 conv_param,
+                                                                 in_g_n_c_wis_desc,
+                                                                 wei_g_k_c_xs_desc,
+                                                                 out_g_n_k_wos_desc,
+                                                                 in_element_op,
+                                                                 wei_element_op,
+                                                                 out_element_op);
+    }
+    else if(conv_param.num_dim_spatial_ == 3)
+    {
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_data<3,
+                                 InDataType,
+                                 WeiDataType,
+                                 OutDataType,
+                                 InElementOp,
+                                 WeiElementOp,
+                                 OutElementOp,
+                                 DeviceConvNdBwdDataInstance<3>>(do_verification,
+                                                                 init_method,
+                                                                 time_kernel,
+                                                                 conv_param,
+                                                                 in_g_n_c_wis_desc,
+                                                                 wei_g_k_c_xs_desc,
+                                                                 out_g_n_k_wos_desc,
+                                                                 in_element_op,
+                                                                 wei_element_op,
+                                                                 out_element_op);
+    }
+
+    return 0;
+}
diff --git a/example/17_convnd_bwd_data_xdl/CMakeLists.txt b/example/17_convnd_bwd_data_xdl/CMakeLists.txt
deleted file mode 100644
index 963f311703..0000000000
--- a/example/17_convnd_bwd_data_xdl/CMakeLists.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-add_example_executable(example_convnd_bwd_data_xdl convnd_bwd_data_xdl.cpp)
-target_link_libraries(example_convnd_bwd_data_xdl PRIVATE conv_util)
diff --git a/example/17_convnd_bwd_data_xdl/convnd_bwd_data_xdl.cpp b/example/17_convnd_bwd_data_xdl/convnd_bwd_data_xdl.cpp
deleted file mode 100644
index 5e3a87e2e4..0000000000
--- a/example/17_convnd_bwd_data_xdl/convnd_bwd_data_xdl.cpp
+++ /dev/null
@@ -1,352 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp"
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-using DeviceConvBwdDataBasePtr =
-    ck::tensor_operation::device::DeviceConvBwdDataPtr<InElementOp, WeiElementOp, OutElementOp>;
-
-template <ck::index_t NumDimSpatial>
-using DeviceConvNDBwdDataInstance = ck::tensor_operation::device::
-    DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<
-        InDataType,     // InDataType
-        WeiDataType,    // WeiDataType
-        OutDataType,    // OutDataType
-        AccDataType,    // AccDataType
-        InElementOp,    // InElementwiseOperation
-        WeiElementOp,   // WeiElementwiseOperation
-        OutElementOp,   // OutElementwiseOperation
-        ConvBwdDefault, // ConvolutionBackwardDataSpecialization
-        NumDimSpatial,  // NumDimSpatial
-        256,            // BlockSize
-        128,            // MPerBlock
-        128,            // NPerBlock
-        4,              // K0PerBlock
-        8,              // K1
-        32,             // MPerXdl
-        32,             // NPerXdl
-        2,              // MXdlPerWave
-        2,              // NXdlPerWave
-        S<4, 64, 1>,    // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,     // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,     // ABlockTransferSrcAccessOrder
-        2,              // ABlockTransferSrcVectorDim
-        8,              // ABlockTransferSrcScalarPerVector
-        8,              // ABlockTransferDstScalarPerVector_K1
-        true,           // ABlockLdsAddExtraM
-        S<4, 64, 1>,    // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<2, 0, 1>,     // BBlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1>,     // BBlockTransferSrcAccessOrder
-        1,              // BBlockTransferSrcVectorDim
-        2,              // BBlockTransferSrcScalarPerVector
-        8,              // BBlockTransferDstScalarPerVector_K1
-        true,           // BBlockLdsAddExtraN
-        7,
-        1>; // GemmCThreadTransferDstScalarPerVector
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvBwdDataInstance =
-    ck::tensor_operation::host::ReferenceConvBwdData<InDataType,
-                                                     WeiDataType,
-                                                     OutDataType,
-                                                     AccDataType,
-                                                     InElementOp,
-                                                     WeiElementOp,
-                                                     OutElementOp,
-                                                     NumDimSpatial>;
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=random value, 2= init to 1 )\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "arg4: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 5;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-DeviceConvBwdDataBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvNDBwdDataInstance<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvNDBwdDataInstance<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvNDBwdDataInstance<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-int main(int argc, char* argv[])
-{
-    bool do_verification = true;
-    int init_method      = 1;
-    bool time_kernel     = false;
-    int num_dim_spatial  = 2;
-
-    ck::utils::conv::ConvParams params;
-    params.C_ = 128;
-
-    if(argc == 4)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-    }
-    else if(argc > 4)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        num_dim_spatial = std::stoi(argv[4]);
-        // check args number
-        int conv_args     = 3 + num_dim_spatial * 6;
-        int cmdline_nargs = conv_args + 5;
-        if(cmdline_nargs != argc)
-        {
-            print_use_msg();
-            exit(1);
-        }
-
-        params = parse_conv_params(num_dim_spatial, argv);
-    }
-    else if(argc != 1)
-    {
-        print_use_msg();
-        exit(1);
-    }
-
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
-
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> in_n_c_hi_wi_host_result(
-        ck::utils::conv::get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<InDataType> in_n_c_hi_wi_device_result(
-        ck::utils::conv::get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> wei_k_c_y_x(
-        ck::utils::conv::get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> out_n_k_ho_wo(
-        ck::utils::conv::get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi_host_result.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_3<OutDataType>{-0.2, 0.2});
-        wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.2, 0.2});
-        break;
-    default:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-        wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_1<WeiDataType>{1});
-    }
-
-    DeviceMem in_device_buf(sizeof(InDataType) *
-                            in_n_c_hi_wi_device_result.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
-
-    out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-    wei_device_buf.ToDevice(wei_k_c_y_x.mData.data());
-    // reset input to zero
-    in_device_buf.SetZero();
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{});
-
-    if(!conv->IsSupportedArgument(argument.get()))
-    {
-        throw std::runtime_error(
-            "wrong! device_conv with the specified compilation parameters does "
-            "not support this Conv problem");
-    }
-
-    float ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = ck::utils::conv::get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype = ck::utils::conv::get_btype<InDataType, WeiDataType, OutDataType>(
-        params.N_,
-        params.C_,
-        params.K_,
-        params.input_spatial_lengths_,
-        params.filter_spatial_lengths_,
-        output_spatial_lengths);
-
-    float tflops     = static_cast<float>(flop) / 1.E9 / ave_time;
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
-    {
-        auto verify_f = [&](const auto& ref_conv) {
-            auto ref_invoker = ref_conv.MakeInvoker();
-
-            auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi_host_result,
-                                                      wei_k_c_y_x,
-                                                      out_n_k_ho_wo,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
-
-            ref_invoker.Run(ref_argument);
-
-            in_device_buf.FromDevice(in_n_c_hi_wi_device_result.mData.data());
-
-            return ck::utils::check_err(in_n_c_hi_wi_device_result.mData,
-                                        in_n_c_hi_wi_host_result.mData)
-                       ? 0
-                       : 1;
-        };
-
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvBwdDataInstance<3>();
-            return verify_f(ref_conv);
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvBwdDataInstance<2>();
-            return verify_f(ref_conv);
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvBwdDataInstance<1>();
-            return verify_f(ref_conv);
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
-    }
-    return 0;
-}
diff --git a/example/18_batched_gemm_reduce/batched_gemm_reduce_xdl_fp16.cpp b/example/18_batched_gemm_reduce/batched_gemm_reduce_xdl_fp16.cpp
index 53bf671514..eaea725efa 100644
--- a/example/18_batched_gemm_reduce/batched_gemm_reduce_xdl_fp16.cpp
+++ b/example/18_batched_gemm_reduce/batched_gemm_reduce_xdl_fp16.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -174,13 +174,13 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_g_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_g_m_n_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce0_device_buf(sizeof(ReduceDataType) *
-                                 d0_g_m_device_result.mDesc.GetElementSpace());
+                                 d0_g_m_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce1_device_buf(sizeof(ReduceDataType) *
-                                 d1_g_m_device_result.mDesc.GetElementSpace());
+                                 d1_g_m_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_g_m_k.mData.data());
     b_device_buf.ToDevice(b_g_k_n.mData.data());
diff --git a/example/19_binary_elementwise/broadcast_add_2d_amn_bn.cpp b/example/19_binary_elementwise/broadcast_add_2d_amn_bn.cpp
index aecd84cb8d..58ee6f7537 100644
--- a/example/19_binary_elementwise/broadcast_add_2d_amn_bn.cpp
+++ b/example/19_binary_elementwise/broadcast_add_2d_amn_bn.cpp
@@ -9,9 +9,9 @@
 #include "ck/tensor_operation/gpu/device/device_binary_elementwise.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;
@@ -92,9 +92,9 @@ int main()
     a_m_n.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
     b_n.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
 
-    DeviceMem a_m_n_device_buf(sizeof(ABDataType) * a_m_n.mDesc.GetElementSpace());
-    DeviceMem b_n_device_buf(sizeof(ABDataType) * b_n.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n.mDesc.GetElementSpace());
+    DeviceMem a_m_n_device_buf(sizeof(ABDataType) * a_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem b_n_device_buf(sizeof(ABDataType) * b_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * c_m_n.mDesc.GetElementSpaceSize());
 
     a_m_n_device_buf.ToDevice(a_m_n.mData.data());
     b_n_device_buf.ToDevice(b_n.mData.data());
diff --git a/example/19_binary_elementwise/broadcast_add_3d_am_bmnk.cpp b/example/19_binary_elementwise/broadcast_add_3d_am_bmnk.cpp
index 89def92d26..ac44673d56 100644
--- a/example/19_binary_elementwise/broadcast_add_3d_am_bmnk.cpp
+++ b/example/19_binary_elementwise/broadcast_add_3d_am_bmnk.cpp
@@ -9,9 +9,9 @@
 #include "ck/tensor_operation/gpu/device/device_binary_elementwise.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;
@@ -74,9 +74,9 @@ int main()
     a_m.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
     b_m_n_k.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
 
-    DeviceMem a_m_device_buf(sizeof(ABDataType) * a_m.mDesc.GetElementSpace());
-    DeviceMem b_m_n_k_device_buf(sizeof(ABDataType) * b_m_n_k.mDesc.GetElementSpace());
-    DeviceMem c_m_n_k_device_buf(sizeof(CDataType) * c_m_n_k.mDesc.GetElementSpace());
+    DeviceMem a_m_device_buf(sizeof(ABDataType) * a_m.mDesc.GetElementSpaceSize());
+    DeviceMem b_m_n_k_device_buf(sizeof(ABDataType) * b_m_n_k.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_k_device_buf(sizeof(CDataType) * c_m_n_k.mDesc.GetElementSpaceSize());
 
     a_m_device_buf.ToDevice(a_m.mData.data());
     b_m_n_k_device_buf.ToDevice(b_m_n_k.mData.data());
diff --git a/example/19_binary_elementwise/elementwise_add_1d.cpp b/example/19_binary_elementwise/elementwise_add_1d.cpp
index aab60146a3..18c12c3e4d 100644
--- a/example/19_binary_elementwise/elementwise_add_1d.cpp
+++ b/example/19_binary_elementwise/elementwise_add_1d.cpp
@@ -8,9 +8,9 @@
 #include "ck/tensor_operation/gpu/device/device_binary_elementwise.hpp"
 #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;
@@ -72,9 +72,9 @@ int main()
     a_m.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
     b_m.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
 
-    DeviceMem a_m_device_buf(sizeof(ABDataType) * a_m.mDesc.GetElementSpace());
-    DeviceMem b_m_device_buf(sizeof(ABDataType) * b_m.mDesc.GetElementSpace());
-    DeviceMem c_m_device_buf(sizeof(CDataType) * c_m.mDesc.GetElementSpace());
+    DeviceMem a_m_device_buf(sizeof(ABDataType) * a_m.mDesc.GetElementSpaceSize());
+    DeviceMem b_m_device_buf(sizeof(ABDataType) * b_m.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_device_buf(sizeof(CDataType) * c_m.mDesc.GetElementSpaceSize());
 
     a_m_device_buf.ToDevice(a_m.mData.data());
     b_m_device_buf.ToDevice(b_m.mData.data());
diff --git a/example/19_binary_elementwise/elementwise_add_4d.cpp b/example/19_binary_elementwise/elementwise_add_4d.cpp
index a4a703a71c..9817208ae4 100644
--- a/example/19_binary_elementwise/elementwise_add_4d.cpp
+++ b/example/19_binary_elementwise/elementwise_add_4d.cpp
@@ -9,9 +9,9 @@
 #include "ck/tensor_operation/gpu/device/device_binary_elementwise.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 using F16 = ck::half_t;
 using F32 = float;
@@ -74,9 +74,9 @@ int main()
     a.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
     b.GenerateTensorValue(GeneratorTensor_3<ABDataType>{0.0, 1.0});
 
-    DeviceMem a_device_buf(sizeof(ABDataType) * a.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(ABDataType) * b.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ABDataType) * a.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(ABDataType) * b.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a.mData.data());
     b_device_buf.ToDevice(b.mData.data());
diff --git a/example/20_convnd_bwd_weight/CMakeLists.txt b/example/20_convnd_bwd_weight/CMakeLists.txt
new file mode 100644
index 0000000000..29a0e312ce
--- /dev/null
+++ b/example/20_convnd_bwd_weight/CMakeLists.txt
@@ -0,0 +1,5 @@
+add_example_executable(example_convnd_bwd_weight_xdl_fp16 convnd_bwd_weight_xdl_fp16.cpp)
+add_example_executable(example_convnd_bwd_weight_xdl_bf16 convnd_bwd_weight_xdl_bf16.cpp)
+
+target_link_libraries(example_convnd_bwd_weight_xdl_fp16 PRIVATE utility)
+target_link_libraries(example_convnd_bwd_weight_xdl_bf16 PRIVATE utility)
diff --git a/example/20_convnd_bwd_weight/convnd_bwd_weight_common.hpp b/example/20_convnd_bwd_weight/convnd_bwd_weight_common.hpp
new file mode 100644
index 0000000000..c9f6c33660
--- /dev/null
+++ b/example/20_convnd_bwd_weight/convnd_bwd_weight_common.hpp
@@ -0,0 +1,152 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_bwd_weight.hpp"
+
+void print_helper_msg()
+{
+    std::cout << "arg1: verification (0=no, 1=yes)\n"
+              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
+              << "arg3: time kernel (0=no, 1=yes)\n"
+              << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+}
+
+template <ck::index_t NDimSpatial,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementOp,
+          typename WeiElementOp,
+          typename OutElementOp,
+          typename DeviceConvBwdWeightInstance>
+int run_conv_bwd_weight(bool do_verification,
+                        int init_method,
+                        bool time_kernel,
+                        const ck::utils::conv::ConvParam& conv_param,
+                        const HostTensorDescriptor& in_g_n_c_wis_desc,
+                        const HostTensorDescriptor& wei_g_k_c_xs_desc,
+                        const HostTensorDescriptor& out_g_n_k_wos_desc,
+                        const InElementOp& in_element_op,
+                        const WeiElementOp& wei_element_op,
+                        const OutElementOp& out_element_op,
+                        ck::index_t split_k)
+{
+    Tensor<InDataType> in(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> wei_host_result(wei_g_k_c_xs_desc);
+    Tensor<WeiDataType> wei_device_result(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> out(out_g_n_k_wos_desc);
+
+    std::cout << "in: " << in.mDesc << std::endl;
+    std::cout << "wei: " << wei_host_result.mDesc << std::endl;
+    std::cout << "out: " << out.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        in.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
+        out.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
+        break;
+    default:
+        in.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
+        out.GenerateTensorValue(GeneratorTensor_3<OutDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
+
+    in_device_buf.ToDevice(in.mData.data());
+    out_device_buf.ToDevice(out.mData.data());
+
+    // init to 0
+    wei_device_buf.SetZero();
+
+    // do GEMM
+    auto conv     = DeviceConvBwdWeightInstance{};
+    auto invoker  = conv.MakeInvoker();
+    auto argument = conv.MakeArgument(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
+                                      static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
+                                      static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
+                                      conv_param.N_,
+                                      conv_param.K_,
+                                      conv_param.C_,
+                                      conv_param.input_spatial_lengths_,
+                                      conv_param.filter_spatial_lengths_,
+                                      conv_param.output_spatial_lengths_,
+                                      conv_param.conv_filter_strides_,
+                                      conv_param.conv_filter_dilations_,
+                                      conv_param.input_left_pads_,
+                                      conv_param.input_right_pads_,
+                                      in_element_op,
+                                      wei_element_op,
+                                      out_element_op,
+                                      split_k);
+
+    if(!conv.IsSupportedArgument(argument))
+    {
+        std::cout << "wrong! device_conv with the specified compilation parameters does "
+                     "not support this Conv problem"
+                  << std::endl;
+        return 1;
+    }
+
+    float avg_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
+
+    std::size_t flop      = conv_param.GetFlops();
+    std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+    float tflops = static_cast<float>(flop) / 1.E9 / avg_time;
+
+    float gb_per_sec = num_btype / 1.E6 / avg_time;
+
+    std::cout << "Perf: " << avg_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
+              << conv.GetTypeString() << std::endl;
+
+    if(do_verification)
+    {
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvBwdWeight<NDimSpatial,
+                                                                           InDataType,
+                                                                           WeiDataType,
+                                                                           OutDataType,
+                                                                           InElementOp,
+                                                                           WeiElementOp,
+                                                                           OutElementOp>{};
+
+        auto ref_invoker = ref_conv.MakeInvoker();
+
+        auto ref_argument = ref_conv.MakeArgument(in,
+                                                  wei_host_result,
+                                                  out,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  InElementOp{},
+                                                  WeiElementOp{},
+                                                  OutElementOp{});
+
+        ref_invoker.Run(ref_argument);
+
+        wei_device_buf.FromDevice(wei_device_result.mData.data());
+
+        return ck::utils::check_err(wei_device_result.mData, wei_host_result.mData) ? 0 : 1;
+    }
+
+    return 0;
+}
diff --git a/example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_bf16.cpp b/example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_bf16.cpp
new file mode 100644
index 0000000000..d9409d7c40
--- /dev/null
+++ b/example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_bf16.cpp
@@ -0,0 +1,219 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "convnd_bwd_weight_common.hpp"
+
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+
+using InDataType = ck::bhalf_t;
+// bf16 kernel use fp32 atomic add to accumulate Weight tensor into global memory
+using WeiDataType = float;
+using OutDataType = ck::bhalf_t;
+using AccDataType = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+template <ck::index_t NDimSpatial>
+using DeviceConvndBwdWeightInstance =
+    ck::tensor_operation::device::DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<
+        NDimSpatial,          // NDimSpatial
+        InDataType,           // InDataType
+        WeiDataType,          // WeiDataType
+        OutDataType,          // OutDataType
+        AccDataType,          // AccDataType
+        InElementOp,          // InElementwiseOperation
+        WeiElementOp,         // WeiElementwiseOperation
+        OutElementOp,         // OutElementwiseOperation
+        ConvBwdWeightDefault, // ConvolutionBackwardWeightSpecialization
+        256,                  // BlockSize
+        128,                  // MPerBlock
+        128,                  // NPerBlock
+        4,                    // K0PerBlock
+        8,                    // K1
+        32,                   // MPerXdl
+        32,                   // NPerXdl
+        2,                    // MXdlPerWave
+        2,                    // NXdlPerWave
+        S<1, 4, 16, 4>,       // ABlockTransferThreadClusterLengths_K0_M_K1
+        S<0, 3, 1, 2>,        // ABlockTransferThreadClusterArrangeOrder
+        S<0, 2, 1, 3>,        // ABlockTransferSrcAccessOrder
+        2,                    // ABlockTransferSrcVectorDim
+        8,                    // ABlockTransferSrcScalarPerVector
+        2,                    // ABlockTransferDstScalarPerVector_K1
+        true,                 // ABlockLdsAddExtraM
+        S<1, 4, 16, 4>,       // BBlockTransferThreadClusterLengths_K0_N_K1
+        S<0, 3, 1, 2>,        // BBlockTransferThreadClusterArrangeOrder
+        S<0, 2, 1, 3>,        // BBlockTransferSrcAccessOrder
+        2,                    // BBlockTransferSrcVectorDim
+        8,                    // BBlockTransferSrcScalarPerVector
+        2,                    // BBlockTransferDstScalarPerVector_K1
+        true,                 // BBlockLdsAddExtraN
+        1,                    // CShuffleMXdlPerWavePerShuffle
+        1,                    // CShuffleNXdlPerWavePerShuffle
+        S<1, 32, 1, 4>,       // CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock
+        4>;                   // CBlockTransferScalarPerVector_NWaveNPerXdl
+
+int main(int argc, char* argv[])
+{
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
+
+    bool do_verification = true;
+    int init_method      = 1;
+    bool time_kernel     = false;
+
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 32, 256, 1024, {3, 3}, {14, 14}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
+
+    ck::index_t split_k = 4;
+
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
+    {
+        do_verification = std::stoi(argv[1]);
+        init_method     = std::stoi(argv[2]);
+        time_kernel     = std::stoi(argv[3]);
+    }
+    else
+    {
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
+
+        split_k = std::stoi(argv[5 + 3 + 6 * num_dim_spatial - 1]);
+        split_k = std::max(1, split_k);
+    }
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    if(conv_param.num_dim_spatial_ == 1)
+    {
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_weight<1,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   InElementOp,
+                                   WeiElementOp,
+                                   OutElementOp,
+                                   DeviceConvndBwdWeightInstance<1>>(do_verification,
+                                                                     init_method,
+                                                                     time_kernel,
+                                                                     conv_param,
+                                                                     in_g_n_c_wis_desc,
+                                                                     wei_g_k_c_xs_desc,
+                                                                     out_g_n_k_wos_desc,
+                                                                     in_element_op,
+                                                                     wei_element_op,
+                                                                     out_element_op,
+                                                                     split_k);
+    }
+    else if(conv_param.num_dim_spatial_ == 2)
+    {
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_weight<2,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   InElementOp,
+                                   WeiElementOp,
+                                   OutElementOp,
+                                   DeviceConvndBwdWeightInstance<2>>(do_verification,
+                                                                     init_method,
+                                                                     time_kernel,
+                                                                     conv_param,
+                                                                     in_g_n_c_wis_desc,
+                                                                     wei_g_k_c_xs_desc,
+                                                                     out_g_n_k_wos_desc,
+                                                                     in_element_op,
+                                                                     wei_element_op,
+                                                                     out_element_op,
+                                                                     split_k);
+    }
+    else if(conv_param.num_dim_spatial_ == 3)
+    {
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_weight<3,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   InElementOp,
+                                   WeiElementOp,
+                                   OutElementOp,
+                                   DeviceConvndBwdWeightInstance<3>>(do_verification,
+                                                                     init_method,
+                                                                     time_kernel,
+                                                                     conv_param,
+                                                                     in_g_n_c_wis_desc,
+                                                                     wei_g_k_c_xs_desc,
+                                                                     out_g_n_k_wos_desc,
+                                                                     in_element_op,
+                                                                     wei_element_op,
+                                                                     out_element_op,
+                                                                     split_k);
+    }
+
+    return 0;
+}
diff --git a/example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_fp16.cpp b/example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_fp16.cpp
new file mode 100644
index 0000000000..39476eb040
--- /dev/null
+++ b/example/20_convnd_bwd_weight/convnd_bwd_weight_xdl_fp16.cpp
@@ -0,0 +1,216 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "convnd_bwd_weight_common.hpp"
+
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+
+using InDataType  = ck::half_t;
+using WeiDataType = ck::half_t;
+using OutDataType = ck::half_t;
+using AccDataType = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+template <ck::index_t NDimSpatial>
+using DeviceConvndBwdWeightInstance =
+    ck::tensor_operation::device::DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<
+        NDimSpatial,          // NDimSpatial
+        InDataType,           // InDataType
+        WeiDataType,          // WeiDataType
+        OutDataType,          // OutDataType
+        AccDataType,          // AccDataType
+        InElementOp,          // InElementwiseOperation
+        WeiElementOp,         // WeiElementwiseOperation
+        OutElementOp,         // OutElementwiseOperation
+        ConvBwdWeightDefault, // ConvolutionBackwardWeightSpecialization
+        256,                  // BlockSize
+        128,                  // MPerBlock
+        128,                  // NPerBlock
+        4,                    // K0PerBlock
+        8,                    // K1
+        32,                   // MPerXdl
+        32,                   // NPerXdl
+        2,                    // MXdlPerWave
+        2,                    // NXdlPerWave
+        S<1, 4, 16, 4>,       // ABlockTransferThreadClusterLengths_K0_M_K1
+        S<0, 3, 1, 2>,        // ABlockTransferThreadClusterArrangeOrder
+        S<0, 2, 1, 3>,        // ABlockTransferSrcAccessOrder
+        2,                    // ABlockTransferSrcVectorDim
+        8,                    // ABlockTransferSrcScalarPerVector
+        2,                    // ABlockTransferDstScalarPerVector_K1
+        true,                 // ABlockLdsAddExtraM
+        S<1, 4, 16, 4>,       // BBlockTransferThreadClusterLengths_K0_N_K1
+        S<0, 3, 1, 2>,        // BBlockTransferThreadClusterArrangeOrder
+        S<0, 2, 1, 3>,        // BBlockTransferSrcAccessOrder
+        2,                    // BBlockTransferSrcVectorDim
+        8,                    // BBlockTransferSrcScalarPerVector
+        2,                    // BBlockTransferDstScalarPerVector_K1
+        true,                 // BBlockLdsAddExtraN
+        1,                    // CShuffleMXdlPerWavePerShuffle
+        1,                    // CShuffleNXdlPerWavePerShuffle
+        S<1, 32, 1, 4>,       // CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock
+        8>;                   // CBlockTransferScalarPerVector_NWaveNPerXdl
+
+int main(int argc, char* argv[])
+{
+    namespace ctc = ck::tensor_layout::convolution;
+
+    bool do_verification = true;
+    int init_method      = 1;
+    bool time_kernel     = false;
+
+    ck::utils::conv::ConvParam conv_param{
+        2, 1, 32, 256, 1024, {3, 3}, {14, 14}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
+
+    ck::index_t split_k = 4;
+
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
+    {
+        do_verification = std::stoi(argv[1]);
+        init_method     = std::stoi(argv[2]);
+        time_kernel     = std::stoi(argv[3]);
+    }
+    else
+    {
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
+
+        split_k = std::stoi(argv[5 + 3 + 6 * num_dim_spatial - 1]);
+        split_k = std::max(1, split_k);
+    }
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    if(conv_param.num_dim_spatial_ == 1)
+    {
+        using InLayout  = ctc::GNWC;
+        using WeiLayout = ctc::GKXC;
+        using OutLayout = ctc::GNWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_weight<1,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   InElementOp,
+                                   WeiElementOp,
+                                   OutElementOp,
+                                   DeviceConvndBwdWeightInstance<1>>(do_verification,
+                                                                     init_method,
+                                                                     time_kernel,
+                                                                     conv_param,
+                                                                     in_g_n_c_wis_desc,
+                                                                     wei_g_k_c_xs_desc,
+                                                                     out_g_n_k_wos_desc,
+                                                                     in_element_op,
+                                                                     wei_element_op,
+                                                                     out_element_op,
+                                                                     split_k);
+    }
+    else if(conv_param.num_dim_spatial_ == 2)
+    {
+        using InLayout  = ctc::GNHWC;
+        using WeiLayout = ctc::GKYXC;
+        using OutLayout = ctc::GNHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_weight<2,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   InElementOp,
+                                   WeiElementOp,
+                                   OutElementOp,
+                                   DeviceConvndBwdWeightInstance<2>>(do_verification,
+                                                                     init_method,
+                                                                     time_kernel,
+                                                                     conv_param,
+                                                                     in_g_n_c_wis_desc,
+                                                                     wei_g_k_c_xs_desc,
+                                                                     out_g_n_k_wos_desc,
+                                                                     in_element_op,
+                                                                     wei_element_op,
+                                                                     out_element_op,
+                                                                     split_k);
+    }
+    else if(conv_param.num_dim_spatial_ == 3)
+    {
+        using InLayout  = ctc::GNDHWC;
+        using WeiLayout = ctc::GKZYXC;
+        using OutLayout = ctc::GNDHWK;
+
+        const auto in_g_n_c_wis_desc =
+            ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(
+                conv_param);
+
+        const auto wei_g_k_c_xs_desc =
+            ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(
+                conv_param);
+
+        const auto out_g_n_k_wos_desc =
+            ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(
+                conv_param);
+
+        return run_conv_bwd_weight<3,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   InElementOp,
+                                   WeiElementOp,
+                                   OutElementOp,
+                                   DeviceConvndBwdWeightInstance<3>>(do_verification,
+                                                                     init_method,
+                                                                     time_kernel,
+                                                                     conv_param,
+                                                                     in_g_n_c_wis_desc,
+                                                                     wei_g_k_c_xs_desc,
+                                                                     out_g_n_k_wos_desc,
+                                                                     in_element_op,
+                                                                     wei_element_op,
+                                                                     out_element_op,
+                                                                     split_k);
+    }
+
+    return 0;
+}
diff --git a/example/20_convnd_bwd_weight_xdl/CMakeLists.txt b/example/20_convnd_bwd_weight_xdl/CMakeLists.txt
deleted file mode 100644
index 66fdef625a..0000000000
--- a/example/20_convnd_bwd_weight_xdl/CMakeLists.txt
+++ /dev/null
@@ -1,4 +0,0 @@
-add_example_executable(example_convnd_bwd_weight_xdl convnd_bwd_weight_xdl.cpp)
-add_example_executable(example_convnd_bwd_weight_xdl_bf16_splitk convnd_bwd_weight_xdl_bf16_splitk.cpp)
-target_link_libraries(example_convnd_bwd_weight_xdl PRIVATE conv_util)
-target_link_libraries(example_convnd_bwd_weight_xdl_bf16_splitk PRIVATE conv_util)
\ No newline at end of file
diff --git a/example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl.cpp b/example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl.cpp
deleted file mode 100644
index e6d64e5964..0000000000
--- a/example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl.cpp
+++ /dev/null
@@ -1,385 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp"
-
-using InDataType  = ck::half_t;
-using WeiDataType = ck::half_t;
-using OutDataType = ck::half_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-using DeviceConvBwdWeightBasePtr =
-    ck::tensor_operation::device::DeviceConvBwdWeightPtr<InElementOp, WeiElementOp, OutElementOp>;
-
-// clang-format off
-template <ck::index_t NumDimSpatial>
-using DeviceConvndBwdWeightInstance = ck::tensor_operation::device::
-    DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        InDataType,                       // InDataType
-        WeiDataType,                      // WeiDataType
-        OutDataType,                      // OutDataType
-        AccDataType,                      // AccDataType
-        InElementOp,                      // InElementwiseOperation
-        WeiElementOp,                     // WeiElementwiseOperation
-        OutElementOp,                     // OutElementwiseOperation
-        ConvBwdWeightDefault,             // ConvolutionBackwardWeightSpecialization
-        NumDimSpatial,                    // NumDimSpatial
-        256,                              // BlockSize
-        128,                              // MPerBlock
-        128,                              // NPerBlock
-        4,                                // K0PerBlock
-        8,                                // K1
-        32,                               // MPerXdl
-        32,                               // NPerXdl
-        2,                                // MXdlPerWave
-        2,                                // NXdlPerWave
-        S<1, 4, 16, 4>,                   // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<0, 3, 1, 2>,                    // ABlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1, 3>,                    // ABlockTransferSrcAccessOrder
-        2,                                // ABlockTransferSrcVectorDim
-        8,                                // ABlockTransferSrcScalarPerVector
-        2,                                // ABlockTransferDstScalarPerVector_K1
-        true,                             // ABlockLdsAddExtraM
-        S<1, 4, 16, 4>,                   // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<0, 3, 1, 2>,                    // BBlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1, 3>,                    // BBlockTransferSrcAccessOrder
-        2,                                // BBlockTransferSrcVectorDim
-        8,                                // BBlockTransferSrcScalarPerVector
-        2,                                // BBlockTransferDstScalarPerVector_K1
-        true,                             // BBlockLdsAddExtraN
-        1,                                // CShuffleMXdlPerWavePerShuffle
-        1,                                // CShuffleNXdlPerWavePerShuffle
-        S<1, 32, 1, 4>,                   // CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock
-        8>;                               // CBlockTransferScalarPerVector_NWaveNPerXdl
-// clang-format on
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvBwdWeightInstance =
-    ck::tensor_operation::host::ReferenceConvBwdWeight<InDataType,
-                                                       WeiDataType,
-                                                       OutDataType,
-                                                       InElementOp,
-                                                       WeiElementOp,
-                                                       OutElementOp,
-                                                       NumDimSpatial>;
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=random value, 2= init to 1 )\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "arg4: is show log (0=no, 1=yes)\n"
-              << "arg5: split-k \n"
-              << "arg6: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 7;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-DeviceConvBwdWeightBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvndBwdWeightInstance<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvndBwdWeightInstance<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvndBwdWeightInstance<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-int main(int argc, char* argv[])
-{
-    bool do_verification = true;
-    int init_method      = 1;
-    bool time_kernel     = false;
-    int num_dim_spatial  = 2;
-    int do_log           = 0;
-    int split_k          = 1;
-
-    ck::utils::conv::ConvParams params;
-    params.C_ = 128;
-
-    if(argc == 6)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        do_log          = std::stoi(argv[4]);
-        split_k         = std::stoi(argv[5]);
-    }
-    else if(argc > 6)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        do_log          = std::stoi(argv[4]);
-        split_k         = std::stoi(argv[5]);
-        num_dim_spatial = std::stoi(argv[6]);
-        // check args number
-        int conv_args     = 3 + num_dim_spatial * 6;
-        int cmdline_nargs = conv_args + 7;
-        if(cmdline_nargs != argc)
-        {
-            print_use_msg();
-            exit(1);
-        }
-
-        params = parse_conv_params(num_dim_spatial, argv);
-    }
-    else if(argc != 1)
-    {
-        print_use_msg();
-        exit(1);
-    }
-
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
-
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> in_n_c_hi_wi(
-        ck::utils::conv::get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> wei_k_c_y_x_host_result(
-        ck::utils::conv::get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<WeiDataType> wei_k_c_y_x_device_result(
-        ck::utils::conv::get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> out_n_k_ho_wo(
-        ck::utils::conv::get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_device_result.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_host_result.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-2, 2});
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_2<InDataType>{-2, 2});
-        break;
-    default:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_1<InDataType>{1});
-    }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) *
-                             wei_k_c_y_x_device_result.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
-    out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-    // reset input to zero
-    wei_device_buf.SetZero();
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{},
-                                  split_k);
-
-    // alloc work space
-    float ave_time = 0.f;
-    if(!conv->IsSupportedArgument(argument.get()))
-    {
-        std::cout << "wrong! device_conv with the specified compilation parameters does "
-                     "not support this Conv problem"
-                  << std::endl;
-        return 1;
-    }
-    ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = ck::utils::conv::get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype = ck::utils::conv::get_btype<InDataType, WeiDataType, OutDataType>(
-        params.N_,
-        params.C_,
-        params.K_,
-        params.input_spatial_lengths_,
-        params.filter_spatial_lengths_,
-        output_spatial_lengths);
-
-    float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
-
-    float gb_per_sec = num_btype / 1.E6 / ave_time;
-
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
-              << std::endl;
-
-    if(do_verification)
-    {
-        auto verify_f = [&](const auto& ref_conv) {
-            auto ref_invoker = ref_conv.MakeInvoker();
-
-            auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi,
-                                                      wei_k_c_y_x_host_result,
-                                                      out_n_k_ho_wo,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
-
-            ref_invoker.Run(ref_argument);
-
-            wei_device_buf.FromDevice(wei_k_c_y_x_device_result.mData.data());
-
-            if(do_log)
-            {
-                LogRangeAsType<float>(std::cout << "out: ", out_n_k_ho_wo.mData, ",") << std::endl;
-                LogRangeAsType<float>(std::cout << "in : ", in_n_c_hi_wi.mData, ",") << std::endl;
-                LogRangeAsType<float>(
-                    std::cout << "wei_device(after): ", wei_k_c_y_x_device_result.mData, ",")
-                    << std::endl;
-                LogRangeAsType<float>(
-                    std::cout << "wei_host  : ", wei_k_c_y_x_host_result.mData, ",")
-                    << std::endl;
-            }
-
-            return ck::utils::check_err(wei_k_c_y_x_device_result.mData,
-                                        wei_k_c_y_x_host_result.mData)
-                       ? 0
-                       : 1;
-        };
-
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvBwdWeightInstance<3>();
-            return verify_f(ref_conv);
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvBwdWeightInstance<2>();
-            return verify_f(ref_conv);
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvBwdWeightInstance<1>();
-            return verify_f(ref_conv);
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
-    }
-    return 0;
-}
diff --git a/example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl_bf16_splitk.cpp b/example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl_bf16_splitk.cpp
deleted file mode 100644
index 34377bab94..0000000000
--- a/example/20_convnd_bwd_weight_xdl/convnd_bwd_weight_xdl_bf16_splitk.cpp
+++ /dev/null
@@ -1,427 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/device/device_unary_elementwise.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp"
-
-using InDataType  = ck::bhalf_t;
-using WeiDataType = ck::bhalf_t;
-using OutDataType = ck::bhalf_t;
-using AccDataType = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-
-using UnaryTypeConvert = ck::tensor_operation::element_wise::UnaryTypeConvert<ck::bhalf_t, float>;
-
-using DeviceUnaryElementwiseTypeConvertInstance = ck::tensor_operation::device::
-    DeviceUnaryElementwise<AccDataType, WeiDataType, UnaryTypeConvert, 1, 4>;
-
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-using DeviceConvBwdWeightBasePtr =
-    ck::tensor_operation::device::DeviceConvBwdWeightPtr<InElementOp, WeiElementOp, OutElementOp>;
-
-// clang-format off
-template <ck::index_t NumDimSpatial>
-using DeviceConvndBwdWeightInstance_bf16_splitk = ck::tensor_operation::device::
-    DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        InDataType,                       // InDataType
-        AccDataType,                      // WeiDataType
-        OutDataType,                      // OutDataType
-        AccDataType,                      // AccDataType
-        InElementOp,                      // InElementwiseOperation
-        WeiElementOp,                     // WeiElementwiseOperation
-        OutElementOp,                     // OutElementwiseOperation
-        ConvBwdWeightDefault,             // ConvolutionBackwardWeightSpecialization
-        NumDimSpatial,                    // NumDimSpatial
-        256,                              // BlockSize
-        128,                              // MPerBlock
-        128,                              // NPerBlock
-        4,                                // K0PerBlock
-        8,                                // K1
-        32,                               // MPerXdl
-        32,                               // NPerXdl
-        2,                                // MXdlPerWave
-        2,                                // NXdlPerWave
-        S<1, 4, 16, 4>,                   // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<0, 3, 1, 2>,                    // ABlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1, 3>,                    // ABlockTransferSrcAccessOrder
-        2,                                // ABlockTransferSrcVectorDim
-        8,                                // ABlockTransferSrcScalarPerVector
-        2,                                // ABlockTransferDstScalarPerVector_K1
-        true,                             // ABlockLdsAddExtraM
-        S<1, 4, 16, 4>,                   // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<0, 3, 1, 2>,                    // BBlockTransferThreadClusterArrangeOrder
-        S<0, 2, 1, 3>,                    // BBlockTransferSrcAccessOrder
-        2,                                // BBlockTransferSrcVectorDim
-        8,                                // BBlockTransferSrcScalarPerVector
-        2,                                // BBlockTransferDstScalarPerVector_K1
-        true,                             // BBlockLdsAddExtraN
-        1,                                // CShuffleMXdlPerWavePerShuffle
-        1,                                // CShuffleNXdlPerWavePerShuffle
-        S<1, 32, 1, 4>,                   // CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock
-        4>;                               // CBlockTransferScalarPerVector_NWaveNPerXdl
-// clang-format on
-
-template <ck::index_t NumDimSpatial>
-using ReferenceConvBwdWeightInstance =
-    ck::tensor_operation::host::ReferenceConvBwdWeight<InDataType,
-                                                       WeiDataType,
-                                                       OutDataType,
-                                                       InElementOp,
-                                                       WeiElementOp,
-                                                       OutElementOp,
-                                                       NumDimSpatial>;
-
-template <typename HostTensorB, typename HostTensorA, typename Functor>
-void host_elementwise(HostTensorB& B,
-                      const HostTensorA& A,
-                      const std::vector<std::size_t>& shape,
-                      Functor functor)
-{
-    size_t tensor_size = std::accumulate(shape.begin(), shape.end(), 1, std::multiplies<int>{});
-    std::cout << __LINE__ << ":" << tensor_size << ", " << A.mData[0] << std::endl;
-    for(std::size_t n = 0; n < tensor_size; ++n)
-    {
-        B.mData[n] = functor(A.mData[n]);
-    }
-}
-
-void print_use_msg()
-{
-    std::cout << "arg1: verification (0=no, 1=yes)\n"
-              << "arg2: initialization (0=no init, 1=random value, 2= init to 1 )\n"
-              << "arg3: time kernel (0=n0, 1=yes)\n"
-              << "arg4: is show log (0=no, 1=yes)\n"
-              << "arg5: split-k : in this example split-k must be larger than 1\n"
-              << "arg6: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    ck::utils::conv::ConvParams params;
-    int arg_idx = 7;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-DeviceConvBwdWeightBasePtr get_conv_instance(int num_dim_spatial)
-{
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return std::make_unique<DeviceConvndBwdWeightInstance_bf16_splitk<3>>();
-    }
-    case 2: {
-        return std::make_unique<DeviceConvndBwdWeightInstance_bf16_splitk<2>>();
-    }
-    case 1: {
-        return std::make_unique<DeviceConvndBwdWeightInstance_bf16_splitk<1>>();
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-int main(int argc, char* argv[])
-{
-    bool do_verification = true;
-    int init_method      = 1;
-    bool time_kernel     = false;
-    int num_dim_spatial  = 2;
-    int do_log           = 0;
-    int split_k          = 2;
-
-    ck::utils::conv::ConvParams params;
-    params.C_ = 128;
-
-    if(argc == 6)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        do_log          = std::stoi(argv[4]);
-        split_k         = std::stoi(argv[5]);
-    }
-    else if(argc > 6)
-    {
-        do_verification = std::stoi(argv[1]);
-        init_method     = std::stoi(argv[2]);
-        time_kernel     = std::stoi(argv[3]);
-        do_log          = std::stoi(argv[4]);
-        split_k         = std::stoi(argv[5]);
-        num_dim_spatial = std::stoi(argv[6]);
-        // check args number
-        int conv_args     = 3 + num_dim_spatial * 6;
-        int cmdline_nargs = conv_args + 7;
-        if(cmdline_nargs != argc)
-        {
-            print_use_msg();
-            exit(1);
-        }
-
-        params = parse_conv_params(num_dim_spatial, argv);
-    }
-    else if(argc != 1)
-    {
-        print_use_msg();
-        exit(1);
-    }
-
-    if(split_k <= 1)
-    {
-        print_use_msg();
-        exit(1);
-    }
-
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
-
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
-
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
-
-    Tensor<InDataType> in_n_c_hi_wi(
-        ck::utils::conv::get_input_host_tensor_descriptor(input_dims, num_dim_spatial));
-    Tensor<WeiDataType> wei_k_c_y_x_host_result(
-        ck::utils::conv::get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<WeiDataType> wei_k_c_y_x_device_result(
-        ck::utils::conv::get_filters_host_tensor_descriptor(filter_dims, num_dim_spatial));
-    Tensor<OutDataType> out_n_k_ho_wo(
-        ck::utils::conv::get_output_host_tensor_descriptor(output_dims, num_dim_spatial));
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_device_result.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_host_result.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-2, 2});
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_2<InDataType>{-2, 2});
-        break;
-    default:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_1<InDataType>{1});
-    }
-
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) *
-                             wei_k_c_y_x_device_result.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
-
-    in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
-    out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-    // reset input to zero
-    wei_device_buf.SetZero();
-
-    // do GEMM
-    auto conv    = get_conv_instance(num_dim_spatial);
-    auto invoker = conv->MakeInvokerPointer();
-    auto argument =
-        conv->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                                  static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                                  static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                                  params.N_,
-                                  params.K_,
-                                  params.C_,
-                                  params.input_spatial_lengths_,
-                                  params.filter_spatial_lengths_,
-                                  output_spatial_lengths,
-                                  params.conv_filter_strides_,
-                                  params.conv_filter_dilations_,
-                                  params.input_left_pads_,
-                                  params.input_right_pads_,
-                                  InElementOp{},
-                                  WeiElementOp{},
-                                  OutElementOp{},
-                                  split_k);
-
-    // alloc work space
-    size_t bwd_weight_workspace_size = conv->GetWorkSpaceSize(argument.get());
-    if(bwd_weight_workspace_size <= 0)
-    {
-        print_use_msg();
-        exit(1);
-    }
-
-    float conv_ave_time = 0.f;
-
-    DeviceMem wei_work_space_device_buf(bwd_weight_workspace_size);
-    wei_work_space_device_buf.SetZero();
-    conv->SetWorkSpacePointer(argument.get(), wei_work_space_device_buf.GetDeviceBuffer());
-
-    if(!conv->IsSupportedArgument(argument.get()))
-    {
-        std::cout << "wrong! device_conv with the specified compilation parameters does "
-                     "not support this Conv problem"
-                  << std::endl;
-        return 1;
-    }
-
-    conv_ave_time = invoker->Run(argument.get(), StreamConfig{nullptr, time_kernel});
-
-    std::size_t flop = ck::utils::conv::get_flops(
-        params.N_, params.C_, params.K_, params.filter_spatial_lengths_, output_spatial_lengths);
-    std::size_t num_btype = ck::utils::conv::get_btype<InDataType, WeiDataType, OutDataType>(
-        params.N_,
-        params.C_,
-        params.K_,
-        params.input_spatial_lengths_,
-        params.filter_spatial_lengths_,
-        output_spatial_lengths);
-
-    float tflops = static_cast<float>(flop) / 1.E9 / conv_ave_time;
-
-    float gb_per_sec = num_btype / 1.E6 / conv_ave_time;
-
-    std::cout << "Perf: conv: " << conv_ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec
-              << " GB/s" << std::endl;
-
-    if(do_verification)
-    {
-        auto verify_f = [&](const auto& ref_conv) {
-            auto ref_invoker = ref_conv.MakeInvoker();
-
-            auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi,
-                                                      wei_k_c_y_x_host_result,
-                                                      out_n_k_ho_wo,
-                                                      params.conv_filter_strides_,
-                                                      params.conv_filter_dilations_,
-                                                      params.input_left_pads_,
-                                                      params.input_right_pads_,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
-
-            ref_invoker.Run(ref_argument);
-
-            wei_device_buf.FromDevice(wei_k_c_y_x_device_result.mData.data());
-
-            if(do_log)
-            {
-                LogRangeAsType<float>(std::cout << "out: ", out_n_k_ho_wo.mData, ",") << std::endl;
-                LogRangeAsType<float>(std::cout << "in : ", in_n_c_hi_wi.mData, ",") << std::endl;
-                LogRangeAsType<float>(
-                    std::cout << "wei_device(after): ", wei_k_c_y_x_device_result.mData, ",")
-                    << std::endl;
-                LogRangeAsType<float>(
-                    std::cout << "wei_host  : ", wei_k_c_y_x_host_result.mData, ",")
-                    << std::endl;
-            }
-
-            return ck::utils::check_err(wei_k_c_y_x_device_result.mData,
-                                        wei_k_c_y_x_host_result.mData)
-                       ? 0
-                       : 1;
-        };
-
-        switch(num_dim_spatial)
-        {
-        case 3: {
-            auto ref_conv = ReferenceConvBwdWeightInstance<3>();
-            verify_f(ref_conv);
-            break;
-        }
-        case 2: {
-            auto ref_conv = ReferenceConvBwdWeightInstance<2>();
-            verify_f(ref_conv);
-            break;
-        }
-        case 1: {
-            auto ref_conv = ReferenceConvBwdWeightInstance<1>();
-            verify_f(ref_conv);
-            break;
-        }
-        default: {
-            throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-        }
-        }
-    }
-    return 0;
-}
diff --git a/example/21_gemm_layernorm/gemm_bias_relu_add_layernorm_xdl_fp16.cpp b/example/21_gemm_layernorm/gemm_bias_relu_add_layernorm_xdl_fp16.cpp
index 6c64cfcf01..1f853ca8c8 100644
--- a/example/21_gemm_layernorm/gemm_bias_relu_add_layernorm_xdl_fp16.cpp
+++ b/example/21_gemm_layernorm/gemm_bias_relu_add_layernorm_xdl_fp16.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -281,18 +281,19 @@ int main()
     gamma_n.GenerateTensorValue(GeneratorTensor_3<GammaDataType>{-1, 1});
     beta_n.GenerateTensorValue(GeneratorTensor_3<BetaDataType>{-1, 1});
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n.mDesc.GetElementSpace());
-    DeviceMem bias_device_buf(sizeof(BiasDataType) * bias_n.mDesc.GetElementSpace());
-    DeviceMem d0_device_buf(sizeof(D0DataType) * c1_m_n.mDesc.GetElementSpace());
-    DeviceMem reduceMean_device_buf(sizeof(ReduceDataType) * reduceMean_m.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem bias_device_buf(sizeof(BiasDataType) * bias_n.mDesc.GetElementSpaceSize());
+    DeviceMem d0_device_buf(sizeof(D0DataType) * c1_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem reduceMean_device_buf(sizeof(ReduceDataType) *
+                                    reduceMean_m.mDesc.GetElementSpaceSize());
     DeviceMem reduceMeanSquare_device_buf(sizeof(ReduceDataType) *
-                                          reduceMeanSquare_m.mDesc.GetElementSpace());
-    DeviceMem gamma_device_buf(sizeof(GammaDataType) * gamma_n.mDesc.GetElementSpace());
-    DeviceMem beta_device_buf(sizeof(BetaDataType) * beta_n.mDesc.GetElementSpace());
+                                          reduceMeanSquare_m.mDesc.GetElementSpaceSize());
+    DeviceMem gamma_device_buf(sizeof(GammaDataType) * gamma_n.mDesc.GetElementSpaceSize());
+    DeviceMem beta_device_buf(sizeof(BetaDataType) * beta_n.mDesc.GetElementSpaceSize());
     DeviceMem layerNorm_device_buf(sizeof(LayerNormOutDataType) *
-                                   layerNorm_m_n.mDesc.GetElementSpace());
+                                   layerNorm_m_n.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/21_gemm_layernorm/gemm_layernorm_xdl_fp16.cpp b/example/21_gemm_layernorm/gemm_layernorm_xdl_fp16.cpp
index 24f049a6dc..d19c495f75 100644
--- a/example/21_gemm_layernorm/gemm_layernorm_xdl_fp16.cpp
+++ b/example/21_gemm_layernorm/gemm_layernorm_xdl_fp16.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -249,16 +249,17 @@ int main()
     gamma_n.GenerateTensorValue(GeneratorTensor_3<GammaDataType>{-1, 1});
     beta_n.GenerateTensorValue(GeneratorTensor_3<BetaDataType>{-1, 1});
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n.mDesc.GetElementSpace());
-    DeviceMem reduceMean_device_buf(sizeof(ReduceDataType) * reduceMean_m.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem reduceMean_device_buf(sizeof(ReduceDataType) *
+                                    reduceMean_m.mDesc.GetElementSpaceSize());
     DeviceMem reduceMeanSquare_device_buf(sizeof(ReduceDataType) *
-                                          reduceMeanSquare_m.mDesc.GetElementSpace());
-    DeviceMem gamma_device_buf(sizeof(GammaDataType) * gamma_n.mDesc.GetElementSpace());
-    DeviceMem beta_device_buf(sizeof(BetaDataType) * beta_n.mDesc.GetElementSpace());
+                                          reduceMeanSquare_m.mDesc.GetElementSpaceSize());
+    DeviceMem gamma_device_buf(sizeof(GammaDataType) * gamma_n.mDesc.GetElementSpaceSize());
+    DeviceMem beta_device_buf(sizeof(BetaDataType) * beta_n.mDesc.GetElementSpaceSize());
     DeviceMem layerNorm_device_buf(sizeof(LayerNormOutDataType) *
-                                   layerNorm_m_n.mDesc.GetElementSpace());
+                                   layerNorm_m_n.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/21_gemm_layernorm/gemm_xdl_layernorm_single_kernel_fp16.cpp b/example/21_gemm_layernorm/gemm_xdl_layernorm_single_kernel_fp16.cpp
index 06506cab8e..a6d15b00ad 100644
--- a/example/21_gemm_layernorm/gemm_xdl_layernorm_single_kernel_fp16.cpp
+++ b/example/21_gemm_layernorm/gemm_xdl_layernorm_single_kernel_fp16.cpp
@@ -7,9 +7,9 @@
 
 #include "ck/ck.hpp"
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/device_gemm_xdl_layernorm_cshuffle.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
@@ -185,13 +185,13 @@ int main(int argc, char* argv[])
     c_m_n_host_result.GenerateTensorValue(GeneratorTensor_1<CDataType>{0});
     acc_m_n_host_result.GenerateTensorValue(GeneratorTensor_1<AccDataType>{0});
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
-    DeviceMem c0_bias_buf(sizeof(C0DataType) * c0_n_bias.mDesc.GetElementSpace());
-    DeviceMem c0_add_buf(sizeof(C0DataType) * c0_m_n_add.mDesc.GetElementSpace());
-    DeviceMem c0_gamma_buf(sizeof(C0DataType) * c0_n_gamma.mDesc.GetElementSpace());
-    DeviceMem c0_beta_buf(sizeof(C0DataType) * c0_n_beta.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem c0_bias_buf(sizeof(C0DataType) * c0_n_bias.mDesc.GetElementSpaceSize());
+    DeviceMem c0_add_buf(sizeof(C0DataType) * c0_m_n_add.mDesc.GetElementSpaceSize());
+    DeviceMem c0_gamma_buf(sizeof(C0DataType) * c0_n_gamma.mDesc.GetElementSpaceSize());
+    DeviceMem c0_beta_buf(sizeof(C0DataType) * c0_n_beta.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/22_cgemm/cgemm_xdl_fp16.cpp b/example/22_cgemm/cgemm_xdl_fp16.cpp
index a1dbf0b6c4..8796dbfb08 100644
--- a/example/22_cgemm/cgemm_xdl_fp16.cpp
+++ b/example/22_cgemm/cgemm_xdl_fp16.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/device/device_cgemm_4gemm_xdl_cshuffle.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_cgemm.hpp"
 
 template <ck::index_t... Is>
@@ -177,14 +177,14 @@ int main(int argc, char* argv[])
 
     auto cgemm = DeviceCGemmInstance{};
 
-    DeviceMem a_m_k_real_device_buf(sizeof(ADataType) * a_m_k_real.mDesc.GetElementSpace());
-    DeviceMem a_m_k_imag_device_buf(sizeof(ADataType) * a_m_k_imag.mDesc.GetElementSpace());
-    DeviceMem b_k_n_real_device_buf(sizeof(BDataType) * b_k_n_real.mDesc.GetElementSpace());
-    DeviceMem b_k_n_imag_device_buf(sizeof(BDataType) * b_k_n_imag.mDesc.GetElementSpace());
+    DeviceMem a_m_k_real_device_buf(sizeof(ADataType) * a_m_k_real.mDesc.GetElementSpaceSize());
+    DeviceMem a_m_k_imag_device_buf(sizeof(ADataType) * a_m_k_imag.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_real_device_buf(sizeof(BDataType) * b_k_n_real.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_imag_device_buf(sizeof(BDataType) * b_k_n_imag.mDesc.GetElementSpaceSize());
     DeviceMem c_m_n_real_device_buf(sizeof(CDataType) *
-                                    c_m_n_real_device_result.mDesc.GetElementSpace());
+                                    c_m_n_real_device_result.mDesc.GetElementSpaceSize());
     DeviceMem c_m_n_imag_device_buf(sizeof(CDataType) *
-                                    c_m_n_imag_device_result.mDesc.GetElementSpace());
+                                    c_m_n_imag_device_result.mDesc.GetElementSpaceSize());
     DeviceMem workspace_device_buf(cgemm.GetWorkspaceSize(M, N, K, StrideA, StrideB, StrideC));
 
     a_m_k_real_device_buf.ToDevice(a_m_k_real.mData.data());
diff --git a/example/23_softmax/softmax_blockwise.cpp b/example/23_softmax/softmax_blockwise.cpp
index 613a86cb0b..fa2e4cbf49 100644
--- a/example/23_softmax/softmax_blockwise.cpp
+++ b/example/23_softmax/softmax_blockwise.cpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_common_util.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_common_util.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_softmax.hpp"
 
 using namespace ck;
@@ -177,7 +177,7 @@ int main(int argc, char* argv[])
         }
 
         if(beta != 0.0f)
-            for(size_t i = 0; i < out_ref.mDesc.GetElementSpace(); i++)
+            for(size_t i = 0; i < out_ref.mDesc.GetElementSpaceSize(); i++)
                 out.mData[i] = out_ref.mData[i];
     };
     // std::cout << "beta = " << beta << std::endl;
@@ -185,8 +185,8 @@ int main(int argc, char* argv[])
     // LogRangeAsType<float>(std::cout << "tensor prior out: " , out.mData, ",") << std::endl;
 
     // these buffers are usually provided by the user application
-    DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpace());
-    DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpace());
+    DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+    DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
 
     in_dev.ToDevice(in.mData.data());
 
diff --git a/example/24_batched_gemm_c_permute/CMakeLists.txt b/example/24_batched_gemm_c_permute/CMakeLists.txt
deleted file mode 100644
index 79c612d053..0000000000
--- a/example/24_batched_gemm_c_permute/CMakeLists.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-add_example_executable(example_batched_gemm_c_permute_xdl_fp16 batched_gemm_c_permute_xdl_fp16.cpp)
-
diff --git a/example/24_batched_gemm_e_permute/CMakeLists.txt b/example/24_batched_gemm_e_permute/CMakeLists.txt
new file mode 100644
index 0000000000..3c5d39784b
--- /dev/null
+++ b/example/24_batched_gemm_e_permute/CMakeLists.txt
@@ -0,0 +1,2 @@
+add_example_executable(example_batched_gemm_e_permute_xdl_fp16 batched_gemm_e_permute_xdl_fp16.cpp)
+
diff --git a/example/24_batched_gemm_c_permute/batched_gemm_c_permute_xdl_fp16.cpp b/example/24_batched_gemm_e_permute/batched_gemm_e_permute_xdl_fp16.cpp
similarity index 67%
rename from example/24_batched_gemm_c_permute/batched_gemm_c_permute_xdl_fp16.cpp
rename to example/24_batched_gemm_e_permute/batched_gemm_e_permute_xdl_fp16.cpp
index 7c69ac72b2..e377530584 100644
--- a/example/24_batched_gemm_c_permute/batched_gemm_c_permute_xdl_fp16.cpp
+++ b/example/24_batched_gemm_e_permute/batched_gemm_e_permute_xdl_fp16.cpp
@@ -6,13 +6,13 @@
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_batched_gemm_c_permute_xdl.hpp"
+#include "ck/tensor_operation/gpu/device/device_batched_gemm_e_permute_xdl.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -30,7 +30,6 @@ using ADataType        = F16;
 using BDataType        = F16;
 using AccDataType      = F32;
 using CShuffleDataType = F16;
-using DsDataType       = ck::Tuple<>;
 using EDataType        = F16;
 
 using ALayout = Row;
@@ -42,16 +41,14 @@ using BElementOp   = PassThrough;
 using CDEElementOp = PassThrough;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-// static constexpr auto MNPadding = ck::tensor_operation::device::GemmSpecialization::MNPadding;
-// static constexpr auto MNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
-// clang-format off
-using DeviceGemmInstance = ck::tensor_operation::device::DeviceBatchedGemmCPermuteXdl
-//######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-//######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-//######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-//######|        |        |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
+using DeviceGemmInstance = ck::tensor_operation::device::DeviceBatchedGemmEPermuteXdl
+    // clang-format off
+//######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+//######|        |        |        |      Type|      Type|        Type|         DataType|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+//######|        |        |        |          |          |            |                 |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|        |        |        |          |          |            |                 |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        < ALayout, BLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 using ReferenceBatchedGemmInstance = ck::tensor_operation::host::
@@ -99,7 +96,7 @@ int main(int argc, char* argv[])
     }
 
     // GEMM shape
-    ck::tensor_operation::device::BatchedGemmCPermuteDesc batched_gemm_c_permute_desc{
+    ck::tensor_operation::device::BatchedGemmEPermuteDesc batched_gemm_e_permute_desc{
         G0, G1, M, N, stride_G0, stride_G1, stride_M, stride_N};
 
     auto f_host_tensor_descriptor = [](std::size_t batch_count_,
@@ -125,7 +122,7 @@ int main(int argc, char* argv[])
     Tensor<BDataType> b_g_k_n(
         f_host_tensor_descriptor(batch_count, K, N, stride_B, batch_stride_B, BLayout{}));
 
-    auto f_host_c_tensor_descriptor = [](std::size_t G0_,
+    auto f_host_e_tensor_descriptor = [](std::size_t G0_,
                                          std::size_t G1_,
                                          std::size_t M_,
                                          std::size_t N_,
@@ -138,15 +135,15 @@ int main(int argc, char* argv[])
             std::vector<std::size_t>({stride_G0_, stride_G1_, stride_M_, stride_N_}));
     };
 
-    Tensor<EDataType> c_g0_g1_m_n_host_result(
-        f_host_c_tensor_descriptor(G0, G1, M, N, stride_G0, stride_G1, stride_M, stride_N));
+    Tensor<EDataType> e_g0_g1_m_n_host_result(
+        f_host_e_tensor_descriptor(G0, G1, M, N, stride_G0, stride_G1, stride_M, stride_N));
 
-    Tensor<EDataType> c_g0_g1_m_n_device_result(
-        f_host_c_tensor_descriptor(G0, G1, M, N, stride_G0, stride_G1, stride_M, stride_N));
+    Tensor<EDataType> e_g0_g1_m_n_device_result(
+        f_host_e_tensor_descriptor(G0, G1, M, N, stride_G0, stride_G1, stride_M, stride_N));
 
     std::cout << "a_g_m_k: " << a_g_m_k.mDesc << std::endl;
     std::cout << "b_g_k_n: " << b_g_k_n.mDesc << std::endl;
-    std::cout << "c_g0_g1_m_n: " << c_g0_g1_m_n_host_result.mDesc << std::endl;
+    std::cout << "e_g0_g1_m_n: " << e_g0_g1_m_n_host_result.mDesc << std::endl;
 
     switch(init_method)
     {
@@ -161,9 +158,10 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(EDataType) * c_g0_g1_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) *
+                           e_g0_g1_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_g_m_k.mData.data());
     b_device_buf.ToDevice(b_g_k_n.mData.data());
@@ -178,7 +176,7 @@ int main(int argc, char* argv[])
     // do GEM
     auto argument = gemm.MakeArgument(static_cast<ADataType*>(a_device_buf.GetDeviceBuffer()),
                                       static_cast<BDataType*>(b_device_buf.GetDeviceBuffer()),
-                                      static_cast<EDataType*>(c_device_buf.GetDeviceBuffer()),
+                                      static_cast<EDataType*>(e_device_buf.GetDeviceBuffer()),
                                       M,
                                       N,
                                       K,
@@ -186,7 +184,7 @@ int main(int argc, char* argv[])
                                       stride_B,
                                       batch_stride_A,
                                       batch_stride_B,
-                                      batched_gemm_c_permute_desc,
+                                      batched_gemm_e_permute_desc,
                                       batch_count,
                                       a_element_op,
                                       b_element_op,
@@ -217,7 +215,7 @@ int main(int argc, char* argv[])
 
     if(do_verification)
     {
-        c_device_buf.FromDevice(c_g0_g1_m_n_device_result.mData.data());
+        e_device_buf.FromDevice(e_g0_g1_m_n_device_result.mData.data());
 
         auto ref_batched_gemm = ReferenceBatchedGemmInstance{};
         auto ref_invoker      = ref_batched_gemm.MakeInvoker();
@@ -238,15 +236,16 @@ int main(int argc, char* argv[])
                 {
                     for(int n = 0; n < N; n++)
                     {
-                        int g                                 = g0 * G1 + g1;
-                        c_g0_g1_m_n_host_result(g0, g1, m, n) = c_g_m_n_host_result(g, m, n);
+                        int g = g0 * G1 + g1;
+
+                        e_g0_g1_m_n_host_result(g0, g1, m, n) = c_g_m_n_host_result(g, m, n);
                     }
                 }
             }
         }
 
-        pass = ck::utils::check_err(c_g0_g1_m_n_host_result.mData,
-                                    c_g0_g1_m_n_device_result.mData,
+        pass = ck::utils::check_err(e_g0_g1_m_n_host_result.mData,
+                                    e_g0_g1_m_n_device_result.mData,
                                     "Error: Incorrect results c");
     }
 
diff --git a/example/25_gemm_bias_c_permute/CMakeLists.txt b/example/25_gemm_bias_c_permute/CMakeLists.txt
deleted file mode 100644
index 29b1d94b3c..0000000000
--- a/example/25_gemm_bias_c_permute/CMakeLists.txt
+++ /dev/null
@@ -1 +0,0 @@
-add_example_executable(example_gemm_bias_c_permute_xdl_fp16 gemm_bias_c_permute_xdl_fp16.cpp)
diff --git a/example/25_gemm_bias_e_permute/CMakeLists.txt b/example/25_gemm_bias_e_permute/CMakeLists.txt
new file mode 100644
index 0000000000..0a1a435dbe
--- /dev/null
+++ b/example/25_gemm_bias_e_permute/CMakeLists.txt
@@ -0,0 +1 @@
+add_example_executable(example_gemm_bias_e_permute_xdl_fp16 gemm_bias_e_permute_xdl_fp16.cpp)
diff --git a/example/25_gemm_bias_c_permute/gemm_bias_c_permute_xdl_fp16.cpp b/example/25_gemm_bias_e_permute/gemm_bias_e_permute_xdl_fp16.cpp
similarity index 96%
rename from example/25_gemm_bias_c_permute/gemm_bias_c_permute_xdl_fp16.cpp
rename to example/25_gemm_bias_e_permute/gemm_bias_e_permute_xdl_fp16.cpp
index e7a439ca34..e4e840d1b8 100644
--- a/example/25_gemm_bias_c_permute/gemm_bias_c_permute_xdl_fp16.cpp
+++ b/example/25_gemm_bias_e_permute/gemm_bias_e_permute_xdl_fp16.cpp
@@ -9,12 +9,12 @@
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_bias_c_permute_xdl.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_bias_e_permute_xdl.hpp"
 #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
 
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 #include "ck/library/utility/check_err.hpp"
 
@@ -49,7 +49,7 @@ using CDEElementOp = Add;
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
 // clang-format off
-using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmBiasCPermute_Xdl
+using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmBiasEPermute_Xdl
 //######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
 //######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
 //######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
@@ -186,12 +186,12 @@ int main(int argc, char* argv[])
         d_m0_m1_m2_n0_n1.GenerateTensorValue(GeneratorTensor_3<DDataType>{0.0, 1.0});
     }
 
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
     DeviceMem d_m0_m1_m2_n0_n1_device_buf(sizeof(DDataType) *
-                                          d_m0_m1_m2_n0_n1.mDesc.GetElementSpace());
-    DeviceMem e_m0_m1_m2_n0_n1_device_buf(sizeof(EDataType) *
-                                          e_m0_m1_m2_n0_n1_device_result.mDesc.GetElementSpace());
+                                          d_m0_m1_m2_n0_n1.mDesc.GetElementSpaceSize());
+    DeviceMem e_m0_m1_m2_n0_n1_device_buf(
+        sizeof(EDataType) * e_m0_m1_m2_n0_n1_device_result.mDesc.GetElementSpaceSize());
 
     a_m_k_device_buf.ToDevice(a_m_k.mData.data());
     b_k_n_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/example/26_contraction/contraction_bilinear_xdl_fp32.cpp b/example/26_contraction/contraction_bilinear_xdl_fp32.cpp
index ed3f2c0e82..070703b4fe 100644
--- a/example/26_contraction/contraction_bilinear_xdl_fp32.cpp
+++ b/example/26_contraction/contraction_bilinear_xdl_fp32.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -324,10 +324,10 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_ms_ks.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_ns_ks.mDesc.GetElementSpace());
-    DeviceMem d_device_buf(sizeof(DDataType) * d_ms_ns.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_ms_ns_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_ms_ks.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_ns_ks.mDesc.GetElementSpaceSize());
+    DeviceMem d_device_buf(sizeof(DDataType) * d_ms_ns.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_ms_ns_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_ms_ks.mData.data());
     b_device_buf.ToDevice(b_ns_ks.mData.data());
diff --git a/example/26_contraction/contraction_scale_xdl_fp32.cpp b/example/26_contraction/contraction_scale_xdl_fp32.cpp
index e5337c45a7..0c8061352c 100644
--- a/example/26_contraction/contraction_scale_xdl_fp32.cpp
+++ b/example/26_contraction/contraction_scale_xdl_fp32.cpp
@@ -12,9 +12,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -260,16 +260,16 @@ int main(int argc, char* argv[])
 
         e_ms_ns_lengths = {M0, M1, N0, N1};
         e_ms_ns_strides = {
-            std::stoi(argv[22]), std::stoi(argv[23]), std::stoi(argv[24]), std::stoi(argv[25])};
+            std::stoi(argv[18]), std::stoi(argv[19]), std::stoi(argv[20]), std::stoi(argv[21])};
 
-        scale = std::stof(argv[26]);
+        scale = std::stof(argv[22]);
     }
     else
     {
         printf("arg1: verification (0=no, 1=yes)\n");
         printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
         printf("arg3: time kernel (0=no, 1=yes)\n");
-        printf("arg4 to 7: M0, M1, N0, N1, K0, K1\n");
+        printf("arg4 to 9: M0, M1, N0, N1, K0, K1\n");
         printf("arg10 to 13: Stride_A_M0, Stride_A_M1, Stride_A_K0, Stride_A_K1\n");
         printf("arg14 to 17: Stride_B_N0, Stride_B_N1, Stride_B_K0, Stride_B_K1\n");
         printf("arg18 to 21: Stride_E_M0, Stride_E_M1, Stride_E_N0, Stride_E_N1\n");
@@ -307,9 +307,9 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_ms_ks.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_ns_ks.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_ms_ns_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_ms_ks.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_ns_ks.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_ms_ns_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_ms_ks.mData.data());
     b_device_buf.ToDevice(b_ns_ks.mData.data());
diff --git a/example/27_layernorm/layernorm_blockwise.cpp b/example/27_layernorm/layernorm_blockwise.cpp
index 9ed1dae838..e2625a7772 100644
--- a/example/27_layernorm/layernorm_blockwise.cpp
+++ b/example/27_layernorm/layernorm_blockwise.cpp
@@ -13,10 +13,10 @@
 #include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_common_util.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_common_util.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp"
 
 using XDataType     = ck::half_t;
@@ -75,10 +75,10 @@ int main()
     gamma.GenerateTensorValue(GeneratorTensor_3<GammaDataType>{0.0, 1.0});
     beta.GenerateTensorValue(GeneratorTensor_3<BetaDataType>{0.0, 1.0});
 
-    DeviceMem x_dev(sizeof(XDataType) * x.mDesc.GetElementSpace());
-    DeviceMem gamma_dev(sizeof(GammaDataType) * gamma.mDesc.GetElementSpace());
-    DeviceMem beta_dev(sizeof(BetaDataType) * beta.mDesc.GetElementSpace());
-    DeviceMem y_dev(sizeof(YDataType) * y.mDesc.GetElementSpace());
+    DeviceMem x_dev(sizeof(XDataType) * x.mDesc.GetElementSpaceSize());
+    DeviceMem gamma_dev(sizeof(GammaDataType) * gamma.mDesc.GetElementSpaceSize());
+    DeviceMem beta_dev(sizeof(BetaDataType) * beta.mDesc.GetElementSpaceSize());
+    DeviceMem y_dev(sizeof(YDataType) * y.mDesc.GetElementSpaceSize());
 
     x_dev.ToDevice(x.mData.data());
     gamma_dev.ToDevice(gamma.mData.data());
diff --git a/example/28_grouped_gemm_bias/grouped_gemm_bias_xdl_fp16.cpp b/example/28_grouped_gemm_bias/grouped_gemm_bias_xdl_fp16.cpp
index de226df690..b7c2dc92ee 100644
--- a/example/28_grouped_gemm_bias/grouped_gemm_bias_xdl_fp16.cpp
+++ b/example/28_grouped_gemm_bias/grouped_gemm_bias_xdl_fp16.cpp
@@ -13,9 +13,9 @@
 #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -34,13 +34,15 @@ using ADataType        = F16;
 using BDataType        = F16;
 using AccDataType      = F32;
 using CShuffleDataType = F16;
-using D0DataType       = F16;
-using DsDataType       = ck::Tuple<D0DataType>;
+using DDataType        = F16;
+using DsDataType       = ck::Tuple<DDataType>;
 using EDataType        = F16;
 
-using ALayout = Row;
-using BLayout = Col;
-using ELayout = Row;
+using ALayout  = Row;
+using BLayout  = Col;
+using DLayout  = Row;
+using DsLayout = ck::Tuple<DLayout>;
+using ELayout  = Row;
 
 using AElementOp   = PassThrough;
 using BElementOp   = PassThrough;
@@ -48,13 +50,13 @@ using CDEElementOp = Add;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
-using DeviceGemmInstance = ck::tensor_operation::device::DeviceGroupedGemmXdl
+using DeviceGemmInstance = ck::tensor_operation::device::DeviceGroupedGemm_Xdl
     // clang-format off
-//######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-//######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-//######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-//######|        |        |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
+//######| ALayout| BLayout| DsLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+//######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+//######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 int main(int argc, char* argv[])
@@ -118,24 +120,24 @@ int main(int argc, char* argv[])
 
     std::vector<Tensor<ADataType>> a_tensors;
     std::vector<Tensor<BDataType>> b_tensors;
-    std::vector<Tensor<D0DataType>> d0_tensors;
+    std::vector<Tensor<DDataType>> d_tensors;
     std::vector<Tensor<EDataType>> e_host_tensors;
     std::vector<Tensor<EDataType>> e_device_tensors;
 
     a_tensors.reserve(group_count);
     b_tensors.reserve(group_count);
-    d0_tensors.reserve(group_count);
+    d_tensors.reserve(group_count);
     e_host_tensors.reserve(group_count);
     e_device_tensors.reserve(group_count);
 
     using DeviceMemPtr = std::unique_ptr<DeviceMem>;
 
-    std::vector<DeviceMemPtr> a_tensors_device, b_tensors_device, d0_tensors_device,
+    std::vector<DeviceMemPtr> a_tensors_device, b_tensors_device, d_tensors_device,
         e_tensors_device;
 
     a_tensors_device.reserve(group_count);
     b_tensors_device.reserve(group_count);
-    d0_tensors_device.reserve(group_count);
+    d_tensors_device.reserve(group_count);
     e_tensors_device.reserve(group_count);
 
     std::size_t flop = 0, num_btype = 0;
@@ -146,7 +148,7 @@ int main(int argc, char* argv[])
             gemm_descs[i].M_, gemm_descs[i].K_, gemm_descs[i].stride_A_, ALayout{})));
         b_tensors.push_back(Tensor<BDataType>(f_host_tensor_descriptor(
             gemm_descs[i].K_, gemm_descs[i].N_, gemm_descs[i].stride_B_, BLayout{})));
-        d0_tensors.push_back(Tensor<D0DataType>(f_host_tensor_descriptor(
+        d_tensors.push_back(Tensor<DDataType>(f_host_tensor_descriptor(
             gemm_descs[i].M_, gemm_descs[i].N_, gemm_descs[i].stride_Ds_[0], ELayout{})));
         e_host_tensors.push_back(Tensor<EDataType>(f_host_tensor_descriptor(
             gemm_descs[i].M_, gemm_descs[i].N_, gemm_descs[i].stride_C_, ELayout{})));
@@ -168,38 +170,38 @@ int main(int argc, char* argv[])
         case 1:
             a_tensors[i].GenerateTensorValue(GeneratorTensor_2<ADataType>{-5, 5});
             b_tensors[i].GenerateTensorValue(GeneratorTensor_2<BDataType>{-5, 5});
-            d0_tensors[i].GenerateTensorValue(GeneratorTensor_2<BDataType>{-5, 5});
+            d_tensors[i].GenerateTensorValue(GeneratorTensor_2<BDataType>{-5, 5});
             break;
         case 2:
             a_tensors[i].GenerateTensorValue(GeneratorTensor_3<ADataType>{0.0, 1.0});
             b_tensors[i].GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
-            d0_tensors[i].GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
+            d_tensors[i].GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5});
             break;
         default:
             a_tensors[i].GenerateTensorValue(GeneratorTensor_Sequential<0>{});
             b_tensors[i].GenerateTensorValue(GeneratorTensor_Sequential<1>{});
-            d0_tensors[i].GenerateTensorValue(GeneratorTensor_Sequential<0>{});
+            d_tensors[i].GenerateTensorValue(GeneratorTensor_Sequential<0>{});
         }
     }
 
     for(std::size_t i = 0; i < gemm_descs.size(); i++)
     {
-        a_tensors_device.emplace_back(
-            std::make_unique<DeviceMem>(sizeof(ADataType) * a_tensors[i].mDesc.GetElementSpace()));
-        b_tensors_device.emplace_back(
-            std::make_unique<DeviceMem>(sizeof(BDataType) * b_tensors[i].mDesc.GetElementSpace()));
-        d0_tensors_device.emplace_back(std::make_unique<DeviceMem>(
-            sizeof(D0DataType) * d0_tensors[i].mDesc.GetElementSpace()));
+        a_tensors_device.emplace_back(std::make_unique<DeviceMem>(
+            sizeof(ADataType) * a_tensors[i].mDesc.GetElementSpaceSize()));
+        b_tensors_device.emplace_back(std::make_unique<DeviceMem>(
+            sizeof(BDataType) * b_tensors[i].mDesc.GetElementSpaceSize()));
+        d_tensors_device.emplace_back(std::make_unique<DeviceMem>(
+            sizeof(DDataType) * d_tensors[i].mDesc.GetElementSpaceSize()));
         e_tensors_device.emplace_back(std::make_unique<DeviceMem>(
-            sizeof(EDataType) * e_device_tensors[i].mDesc.GetElementSpace()));
+            sizeof(EDataType) * e_device_tensors[i].mDesc.GetElementSpaceSize()));
 
         a_tensors_device[i]->ToDevice(a_tensors[i].mData.data());
         b_tensors_device[i]->ToDevice(b_tensors[i].mData.data());
-        d0_tensors_device[i]->ToDevice(d0_tensors[i].mData.data());
+        d_tensors_device[i]->ToDevice(d_tensors[i].mData.data());
 
         p_a.push_back(a_tensors_device[i]->GetDeviceBuffer());
         p_b.push_back(b_tensors_device[i]->GetDeviceBuffer());
-        p_ds.push_back({d0_tensors_device[i]->GetDeviceBuffer()});
+        p_ds.push_back({d_tensors_device[i]->GetDeviceBuffer()});
         p_c.push_back(e_tensors_device[i]->GetDeviceBuffer());
     }
 
@@ -266,7 +268,7 @@ int main(int argc, char* argv[])
                 for(int n = 0; n < gemm_descs[i].N_; ++n)
                 {
                     cde_element_op(
-                        e_host_tensors[i](m, n), e_host_tensors[i](m, n), d0_tensors[i](m, n));
+                        e_host_tensors[i](m, n), e_host_tensors[i](m, n), d_tensors[i](m, n));
                 }
             }
 
diff --git a/example/29_batched_gemm_multi_d/batched_gemm_bias_xdl_fp16.cpp b/example/29_batched_gemm_multi_d/batched_gemm_bias_xdl_fp16.cpp
index 2f988a6b18..badc3fecb9 100644
--- a/example/29_batched_gemm_multi_d/batched_gemm_bias_xdl_fp16.cpp
+++ b/example/29_batched_gemm_multi_d/batched_gemm_bias_xdl_fp16.cpp
@@ -10,9 +10,9 @@
 #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -37,7 +37,9 @@ using EDataType        = F16;
 
 using ALayout  = Row;
 using BLayout  = Col;
-using DELayout = Row;
+using DLayout  = Row;
+using DsLayout = ck::Tuple<DLayout>;
+using ELayout  = Row;
 
 using AElementOp   = PassThrough;
 using BElementOp   = PassThrough;
@@ -48,12 +50,12 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 // static constexpr auto MNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
 // clang-format off
-using DeviceGemmInstance = ck::tensor_operation::device::DeviceBatchedGemmMultiDXdl
-//######| ALayout| BLayout| DELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-//######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-//######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-//######|        |        |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, DELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
+using DeviceGemmInstance = ck::tensor_operation::device::DeviceBatchedGemmMultiD_Xdl
+//######| ALayout| BLayout| DsLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+//######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+//######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 int main(int argc, char* argv[])
@@ -117,10 +119,10 @@ int main(int argc, char* argv[])
         f_host_tensor_descriptor(batch_count, K, N, stride_B, batch_stride_B, BLayout{}));
 
     Tensor<DDataType> d_g_m_n(
-        f_host_tensor_descriptor(batch_count, M, N, stride_D, batch_stride_D, DELayout{}));
+        f_host_tensor_descriptor(batch_count, M, N, stride_D, batch_stride_D, DLayout{}));
 
     Tensor<EDataType> e_g_m_n_device_result(
-        f_host_tensor_descriptor(batch_count, M, N, stride_E, batch_stride_E, DELayout{}));
+        f_host_tensor_descriptor(batch_count, M, N, stride_E, batch_stride_E, ELayout{}));
 
     std::cout << "a_g_m_k: " << a_g_m_k.mDesc << std::endl;
     std::cout << "b_g_k_n: " << b_g_k_n.mDesc << std::endl;
@@ -142,10 +144,10 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpace());
-    DeviceMem d_device_buf(sizeof(DDataType) * d_g_m_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(EDataType) * e_g_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem d_device_buf(sizeof(DDataType) * d_g_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(EDataType) * e_g_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_g_m_k.mData.data());
     b_device_buf.ToDevice(b_g_k_n.mData.data());
@@ -166,6 +168,7 @@ int main(int argc, char* argv[])
                                       M,
                                       N,
                                       K,
+                                      batch_count,
                                       stride_A,
                                       stride_B,
                                       {stride_D},
@@ -174,7 +177,6 @@ int main(int argc, char* argv[])
                                       batch_stride_B,
                                       {batch_stride_D},
                                       batch_stride_E,
-                                      batch_count,
                                       a_element_op,
                                       b_element_op,
                                       cde_element_op);
@@ -218,7 +220,7 @@ int main(int argc, char* argv[])
         auto ref_invoker      = ref_batched_gemm.MakeInvoker();
 
         Tensor<EDataType> e_g_m_n_host_result(
-            f_host_tensor_descriptor(batch_count, M, N, stride_E, batch_stride_E, DELayout{}));
+            f_host_tensor_descriptor(batch_count, M, N, stride_E, batch_stride_E, ELayout{}));
 
         auto ref_argument = ref_batched_gemm.MakeArgument(
             a_g_m_k, b_g_k_n, e_g_m_n_host_result, a_element_op, b_element_op, PassThrough{});
diff --git a/example/29_batched_gemm_multi_d/batched_gemm_xdl_fp16.cpp b/example/29_batched_gemm_multi_d/batched_gemm_xdl_fp16.cpp
index 8b04781cbd..cb6b8d10fb 100644
--- a/example/29_batched_gemm_multi_d/batched_gemm_xdl_fp16.cpp
+++ b/example/29_batched_gemm_multi_d/batched_gemm_xdl_fp16.cpp
@@ -10,9 +10,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp"
 
 template <ck::index_t... Is>
@@ -33,9 +33,10 @@ using CShuffleDataType = F16;
 using DsDataType       = ck::Tuple<>;
 using EDataType        = F16;
 
-using ALayout = Row;
-using BLayout = Col;
-using ELayout = Row;
+using ALayout  = Row;
+using BLayout  = Col;
+using DsLayout = ck::Tuple<>;
+using ELayout  = Row;
 
 using AElementOp   = PassThrough;
 using BElementOp   = PassThrough;
@@ -46,12 +47,12 @@ static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecializa
 // static constexpr auto MNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
 
 // clang-format off
-using DeviceGemmInstance = ck::tensor_operation::device::DeviceBatchedGemmMultiDXdl
-//######| ALayout| BLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-//######|        |        |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-//######|        |        |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-//######|        |        |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
+using DeviceGemmInstance = ck::tensor_operation::device::DeviceBatchedGemmMultiD_Xdl
+//######| ALayout| BLayout| DsLayout| ELayout|     AData|     BData|     AccData|         CShuffle|     DsData|     EData|           A|           B|          CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+//######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+//######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+//######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 using ReferenceBatchedGemmInstance = ck::tensor_operation::host::
@@ -135,9 +136,9 @@ int main(int argc, char* argv[])
         break;
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(EDataType) * e_g_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(EDataType) * e_g_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_g_m_k.mData.data());
     b_device_buf.ToDevice(b_g_k_n.mData.data());
@@ -157,6 +158,7 @@ int main(int argc, char* argv[])
                                       M,
                                       N,
                                       K,
+                                      batch_count,
                                       stride_A,
                                       stride_B,
                                       {},
@@ -165,7 +167,6 @@ int main(int argc, char* argv[])
                                       batch_stride_B,
                                       {},
                                       batch_stride_C,
-                                      batch_count,
                                       a_element_op,
                                       b_element_op,
                                       cde_element_op);
diff --git a/example/30_grouped_convnd_fwd_bias_relu/CMakeLists.txt b/example/30_grouped_convnd_fwd_bias_relu/CMakeLists.txt
new file mode 100644
index 0000000000..cd91cc80ee
--- /dev/null
+++ b/example/30_grouped_convnd_fwd_bias_relu/CMakeLists.txt
@@ -0,0 +1,2 @@
+add_example_executable(example_grouped_convnd_fwd_bias_relu_xdl_fp16 grouped_convnd_fwd_bias_relu_xdl_fp16.cpp)
+target_link_libraries(example_grouped_convnd_fwd_bias_relu_xdl_fp16 PRIVATE utility)
diff --git a/example/30_grouped_convnd_fwd_bias_relu/README.md b/example/30_grouped_convnd_fwd_bias_relu/README.md
new file mode 100644
index 0000000000..b9865ea1cb
--- /dev/null
+++ b/example/30_grouped_convnd_fwd_bias_relu/README.md
@@ -0,0 +1,28 @@
+```bash
+#arg1: verification (0=no, 1=yes)
+#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
+#arg3: time kernel (0=no, 1=yes)
+#Following arguments (depending on number of spatial dims):
+# N spatial dimensions
+# G, N, K, C,
+# <filter spatial dimensions>, (ie Y, X for 2D)
+# <input image spatial dimensions>, (ie Hi, Wi for 2D)
+# <strides>, (ie Sy, Sx for 2D)
+# <dilations>, (ie Dy, Dx for 2D)
+# <left padding>, (ie LeftPy, LeftPx for 2D)
+# <right padding>, (ie RightPy, RightPx for 2D)
+
+bin/example_grouped_convnd_fwd_bias_relu_xdl_fp16 1 1 1
+```
+
+Result (MI100)
+```
+in: dim 5, lengths {1, 128, 192, 71, 71}, strides {6912, 967872, 1, 13632, 192}
+wei: dim 5, lengths {1, 256, 192, 3, 3}, strides {192, 1728, 1, 576, 192}
+bias: dim 5, lengths {1, 128, 256, 36, 36}, strides {256, 0, 1, 0, 0}
+out: dim 5, lengths {1, 128, 256, 36, 36}, strides {256, 331776, 1, 9216, 256}
+launch_and_time_kernel: grid_dim {1296, 1, 1}, block_dim {256, 1, 1}
+Warm up 1 time
+Start running 10 times...
+Perf: 1.19215 ms, 123.112 TFlops, 279.827 GB/s, DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<256, 128, 256, 32, Default>
+```
diff --git a/example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_common.hpp b/example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_common.hpp
new file mode 100644
index 0000000000..63f41b5932
--- /dev/null
+++ b/example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_common.hpp
@@ -0,0 +1,192 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <numeric>
+#include <type_traits>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+
+void print_helper_msg()
+{
+    std::cout << "arg1: verification (0=no, 1=yes)\n"
+              << "arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n"
+              << "arg3: time kernel (0=no, 1=yes)\n"
+              << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+}
+
+template <ck::index_t NDimSpatial,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementOp,
+          typename WeiElementOp,
+          typename OutElementOp,
+          typename DeviceConvNDFwdInstance>
+int run_grouped_conv_fwd_bias(bool do_verification,
+                              int init_method,
+                              bool time_kernel,
+                              const ck::utils::conv::ConvParam& conv_param,
+                              const HostTensorDescriptor& in_g_n_c_wis_desc,
+                              const HostTensorDescriptor& wei_g_k_c_xs_desc,
+                              const HostTensorDescriptor& bias_g_n_k_wos_desc,
+                              const HostTensorDescriptor& out_g_n_k_wos_desc,
+                              const InElementOp& in_element_op,
+                              const WeiElementOp& wei_element_op,
+                              const OutElementOp& out_element_op)
+{
+    Tensor<InDataType> in(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> wei(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> bias(bias_g_n_k_wos_desc);
+    Tensor<OutDataType> out_host(out_g_n_k_wos_desc);
+    Tensor<OutDataType> out_device(out_g_n_k_wos_desc);
+
+    std::cout << "in: " << in.mDesc << std::endl;
+    std::cout << "wei: " << wei.mDesc << std::endl;
+    std::cout << "bias: " << bias.mDesc << std::endl;
+    std::cout << "out: " << out_host.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        in.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
+        wei.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        bias.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
+        break;
+    default:
+        in.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
+        wei.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+        bias.GenerateTensorValue(GeneratorTensor_3<OutDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei.mDesc.GetElementSpaceSize());
+    DeviceMem bias_device_buf(sizeof(OutDataType) * bias.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * out_device.mDesc.GetElementSpaceSize());
+
+    in_device_buf.ToDevice(in.mData.data());
+    wei_device_buf.ToDevice(wei.mData.data());
+    bias_device_buf.ToDevice(bias.mData.data());
+
+    std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> d_g_n_k_wos_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> d_g_n_k_wos_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_strides{};
+    std::array<ck::index_t, NDimSpatial> conv_filter_strides{};
+    std::array<ck::index_t, NDimSpatial> conv_filter_dilations{};
+    std::array<ck::index_t, NDimSpatial> input_left_pads{};
+    std::array<ck::index_t, NDimSpatial> input_right_pads{};
+
+    auto copy = [](auto& x, auto& y) { std::copy(x.begin(), x.end(), y.begin()); };
+
+    copy(in_g_n_c_wis_desc.GetLengths(), a_g_n_c_wis_lengths);
+    copy(in_g_n_c_wis_desc.GetStrides(), a_g_n_c_wis_strides);
+    copy(wei_g_k_c_xs_desc.GetLengths(), b_g_k_c_xs_lengths);
+    copy(wei_g_k_c_xs_desc.GetStrides(), b_g_k_c_xs_strides);
+    copy(bias_g_n_k_wos_desc.GetLengths(), d_g_n_k_wos_lengths);
+    copy(bias_g_n_k_wos_desc.GetStrides(), d_g_n_k_wos_strides);
+    copy(out_g_n_k_wos_desc.GetLengths(), e_g_n_k_wos_lengths);
+    copy(out_g_n_k_wos_desc.GetStrides(), e_g_n_k_wos_strides);
+    copy(conv_param.conv_filter_strides_, conv_filter_strides);
+    copy(conv_param.conv_filter_dilations_, conv_filter_dilations);
+    copy(conv_param.input_left_pads_, input_left_pads);
+    copy(conv_param.input_right_pads_, input_right_pads);
+
+    // do Conv
+    auto conv     = DeviceConvNDFwdInstance{};
+    auto invoker  = conv.MakeInvoker();
+    auto argument = conv.MakeArgument(
+        in_device_buf.GetDeviceBuffer(),
+        wei_device_buf.GetDeviceBuffer(),
+        std::array<const void*, 1>{bias_device_buf.GetDeviceBuffer()},
+        out_device_buf.GetDeviceBuffer(),
+        a_g_n_c_wis_lengths,
+        a_g_n_c_wis_strides,
+        b_g_k_c_xs_lengths,
+        b_g_k_c_xs_strides,
+        std::array<std::array<ck::index_t, NDimSpatial + 3>, 1>{{d_g_n_k_wos_lengths}},
+        std::array<std::array<ck::index_t, NDimSpatial + 3>, 1>{{d_g_n_k_wos_strides}},
+        e_g_n_k_wos_lengths,
+        e_g_n_k_wos_strides,
+        conv_filter_strides,
+        conv_filter_dilations,
+        input_left_pads,
+        input_right_pads,
+        in_element_op,
+        wei_element_op,
+        out_element_op);
+
+    if(!conv.IsSupportedArgument(argument))
+    {
+        throw std::runtime_error(
+            "wrong! device_conv with the specified compilation parameters does "
+            "not support this Conv problem");
+    }
+
+    float avg_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
+
+    std::size_t flop      = conv_param.GetFlops();
+    std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+    float tflops     = static_cast<float>(flop) / 1.E9 / avg_time;
+    float gb_per_sec = num_btype / 1.E6 / avg_time;
+    std::cout << "Perf: " << avg_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s, "
+              << conv.GetTypeString() << std::endl;
+
+    if(do_verification)
+    {
+        using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+        Tensor<OutDataType> c_host(out_g_n_k_wos_desc);
+
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvFwd<NDimSpatial,
+                                                                     InDataType,
+                                                                     WeiDataType,
+                                                                     OutDataType,
+                                                                     InElementOp,
+                                                                     WeiElementOp,
+                                                                     PassThrough>();
+
+        auto ref_invoker  = ref_conv.MakeInvoker();
+        auto ref_argument = ref_conv.MakeArgument(in,
+                                                  wei,
+                                                  c_host,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  in_element_op,
+                                                  wei_element_op,
+                                                  PassThrough{});
+
+        ref_invoker.Run(ref_argument);
+
+        // TODO: implement elementwise operation for host
+        out_host.ForEach(
+            [&](auto&, auto idx) { out_element_op(out_host(idx), c_host(idx), bias(idx)); });
+
+        out_device_buf.FromDevice(out_device.mData.data());
+
+        return ck::utils::check_err(
+                   out_device.mData, out_host.mData, "Error: incorrect results!", 1e-5f, 1e-4f)
+                   ? 0
+                   : 1;
+    }
+
+    return 0;
+}
diff --git a/example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_relu_xdl_fp16.cpp b/example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_relu_xdl_fp16.cpp
new file mode 100644
index 0000000000..6331386cc4
--- /dev/null
+++ b/example/30_grouped_convnd_fwd_bias_relu/grouped_convnd_fwd_bias_relu_xdl_fp16.cpp
@@ -0,0 +1,370 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "grouped_convnd_fwd_bias_common.hpp"
+
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+
+using InDataType       = ck::half_t;
+using WeiDataType      = ck::half_t;
+using AccDataType      = float;
+using CShuffleDataType = ck::half_t;
+using BiasDataType     = ck::half_t;
+using OutDataType      = ck::half_t;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::AddRelu;
+
+static constexpr auto ConvSpec =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+template <ck::index_t NDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename BiasLayout,
+          typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<BiasLayout>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<BiasDataType>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        256,         // NPerBlock
+        32,          // KPerBlock
+        8,           // AK1
+        8,           // BK1
+        32,          // MPerXdl
+        32,          // NPerXdl
+        2,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        8,           // ABlockTransferSrcScalarPerVector
+        8,           // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        8,           // BBlockTransferSrcScalarPerVector
+        8,           // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 32, 1, 8>,
+        8>;
+
+int main(int argc, char* argv[])
+{
+    namespace ctc = ck::tensor_layout::convolution;
+
+    print_helper_msg();
+
+    bool do_verification = true;
+    int init_method      = 1;
+    bool time_kernel     = false;
+
+    ck::utils::conv::ConvParam conv_param{
+        2, 2, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1}};
+
+    if(argc == 1)
+    {
+        // use default
+    }
+    else if(argc == 4)
+    {
+        do_verification = std::stoi(argv[1]);
+        init_method     = std::stoi(argv[2]);
+        time_kernel     = std::stoi(argv[3]);
+    }
+    else
+    {
+        do_verification                   = std::stoi(argv[1]);
+        init_method                       = std::stoi(argv[2]);
+        time_kernel                       = std::stoi(argv[3]);
+        const ck::index_t num_dim_spatial = std::stoi(argv[4]);
+
+        conv_param = ck::utils::conv::parse_conv_param(num_dim_spatial, 5, argv);
+    }
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    if(conv_param.num_dim_spatial_ == 1)
+    {
+        using InLayout   = ctc::G_NW_C;
+        using WeiLayout  = ctc::G_K_X_C;
+        using BiasLayout = ctc::G_NW_K;
+        using OutLayout  = ctc::G_NW_K;
+
+        const auto in_g_n_c_wis_desc = HostTensorDescriptor(
+            {conv_param.G_, conv_param.N_, conv_param.C_, conv_param.input_spatial_lengths_[0]},
+            {
+                conv_param.C_,                                                        // g
+                conv_param.input_spatial_lengths_[0] * conv_param.G_ * conv_param.C_, // n
+                1,                                                                    // c
+                conv_param.G_ * conv_param.C_                                         // wi
+            });
+
+        const auto wei_g_k_c_xs_desc = HostTensorDescriptor(
+            {conv_param.G_, conv_param.K_, conv_param.C_, conv_param.filter_spatial_lengths_[0]},
+            {
+                conv_param.C_,                                                         // g
+                conv_param.filter_spatial_lengths_[0] * conv_param.G_ * conv_param.C_, // k
+                1,                                                                     // c
+                conv_param.G_ * conv_param.C_                                          // x
+            });
+
+        const auto bias_g_n_k_wos_desc = HostTensorDescriptor(
+            {conv_param.G_, conv_param.N_, conv_param.K_, conv_param.output_spatial_lengths_[0]},
+            {
+                conv_param.K_, // g
+                0,             // k
+                1,             // c
+                0              // x
+            });
+
+        const auto out_g_n_k_wos_desc = HostTensorDescriptor(
+            {conv_param.G_, conv_param.N_, conv_param.K_, conv_param.output_spatial_lengths_[0]},
+            {
+                conv_param.K_,                                                         // g
+                conv_param.output_spatial_lengths_[0] * conv_param.G_ * conv_param.K_, // n
+                1,                                                                     // k
+                conv_param.G_ * conv_param.K_                                          // wo
+            });
+
+        return run_grouped_conv_fwd_bias<
+            1,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<1, InLayout, WeiLayout, BiasLayout, OutLayout>>(
+            do_verification,
+            init_method,
+            time_kernel,
+            conv_param,
+            in_g_n_c_wis_desc,
+            wei_g_k_c_xs_desc,
+            bias_g_n_k_wos_desc,
+            out_g_n_k_wos_desc,
+            in_element_op,
+            wei_element_op,
+            out_element_op);
+    }
+    else if(conv_param.num_dim_spatial_ == 2)
+    {
+        using InLayout   = ctc::G_NHW_C;
+        using WeiLayout  = ctc::G_K_YX_C;
+        using BiasLayout = ctc::G_NHW_K;
+        using OutLayout  = ctc::G_NHW_K;
+
+        const auto in_g_n_c_wis_desc = HostTensorDescriptor(
+            {conv_param.G_,
+             conv_param.N_,
+             conv_param.C_,
+             conv_param.input_spatial_lengths_[0],
+             conv_param.input_spatial_lengths_[1]},
+            {
+                conv_param.output_spatial_lengths_[0] * conv_param.C_, // g
+                conv_param.input_spatial_lengths_[0] * conv_param.input_spatial_lengths_[1] *
+                    conv_param.G_ * conv_param.C_,                                    // n
+                1,                                                                    // c
+                conv_param.input_spatial_lengths_[1] * conv_param.G_ * conv_param.C_, // hi
+                conv_param.G_ * conv_param.C_                                         // wi
+            });
+
+        const auto wei_g_k_c_xs_desc = HostTensorDescriptor(
+            {conv_param.G_,
+             conv_param.K_,
+             conv_param.C_,
+             conv_param.filter_spatial_lengths_[0],
+             conv_param.filter_spatial_lengths_[1]},
+            {
+                conv_param.C_, // g
+                conv_param.filter_spatial_lengths_[0] * conv_param.filter_spatial_lengths_[1] *
+                    conv_param.G_ * conv_param.C_,                                     // k
+                1,                                                                     // c
+                conv_param.filter_spatial_lengths_[1] * conv_param.G_ * conv_param.C_, // y
+                conv_param.G_ * conv_param.C_                                          // x
+            });
+
+        const auto bias_g_n_k_wos_desc =
+            HostTensorDescriptor({conv_param.G_,
+                                  conv_param.N_,
+                                  conv_param.K_,
+                                  conv_param.output_spatial_lengths_[0],
+                                  conv_param.output_spatial_lengths_[1]},
+                                 {
+                                     conv_param.K_, // g
+                                     0,             // n
+                                     1,             // k
+                                     0,             // ho
+                                     0              // wo
+                                 });
+
+        const auto out_g_n_k_wos_desc = HostTensorDescriptor(
+            {conv_param.G_,
+             conv_param.N_,
+             conv_param.K_,
+             conv_param.output_spatial_lengths_[0],
+             conv_param.output_spatial_lengths_[1]},
+            {
+                conv_param.K_, // g
+                conv_param.output_spatial_lengths_[0] * conv_param.output_spatial_lengths_[1] *
+                    conv_param.G_ * conv_param.K_,                                     // n
+                1,                                                                     // k
+                conv_param.output_spatial_lengths_[1] * conv_param.G_ * conv_param.K_, // ho
+                conv_param.G_ * conv_param.K_                                          // wo
+            });
+
+        return run_grouped_conv_fwd_bias<
+            2,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<2, InLayout, WeiLayout, BiasLayout, OutLayout>>(
+            do_verification,
+            init_method,
+            time_kernel,
+            conv_param,
+            in_g_n_c_wis_desc,
+            wei_g_k_c_xs_desc,
+            bias_g_n_k_wos_desc,
+            out_g_n_k_wos_desc,
+            in_element_op,
+            wei_element_op,
+            out_element_op);
+    }
+    else if(conv_param.num_dim_spatial_ == 3)
+    {
+        using InLayout   = ctc::G_NDHW_C;
+        using WeiLayout  = ctc::G_K_ZYX_C;
+        using BiasLayout = ctc::G_NDHW_K;
+        using OutLayout  = ctc::G_NDHW_K;
+
+        const auto in_g_n_c_wis_desc = HostTensorDescriptor(
+            {conv_param.G_,
+             conv_param.N_,
+             conv_param.C_,
+             conv_param.input_spatial_lengths_[0],
+             conv_param.input_spatial_lengths_[1],
+             conv_param.input_spatial_lengths_[2]},
+            {
+                conv_param.output_spatial_lengths_[0] * conv_param.C_, // g
+                conv_param.input_spatial_lengths_[0] * conv_param.input_spatial_lengths_[1] *
+                    conv_param.input_spatial_lengths_[2] * conv_param.G_ * conv_param.C_, // n
+                1,                                                                        // c
+                conv_param.input_spatial_lengths_[1] * conv_param.input_spatial_lengths_[2] *
+                    conv_param.G_ * conv_param.C_,                                    // di
+                conv_param.input_spatial_lengths_[2] * conv_param.G_ * conv_param.C_, // hi
+                conv_param.G_ * conv_param.C_                                         // wi
+            });
+
+        const auto wei_g_k_c_xs_desc = HostTensorDescriptor(
+            {conv_param.G_,
+             conv_param.K_,
+             conv_param.C_,
+             conv_param.filter_spatial_lengths_[0],
+             conv_param.filter_spatial_lengths_[1],
+             conv_param.filter_spatial_lengths_[2]},
+            {
+                conv_param.C_, // g
+                conv_param.filter_spatial_lengths_[0] * conv_param.filter_spatial_lengths_[1] *
+                    conv_param.filter_spatial_lengths_[2] * conv_param.G_ * conv_param.C_, // k
+                1,                                                                         // c
+                conv_param.filter_spatial_lengths_[1] * conv_param.filter_spatial_lengths_[2] *
+                    conv_param.G_ * conv_param.C_,                                     // z
+                conv_param.filter_spatial_lengths_[2] * conv_param.G_ * conv_param.C_, // y
+                conv_param.G_ * conv_param.C_                                          // x
+            });
+
+        const auto bias_g_n_k_wos_desc =
+            HostTensorDescriptor({conv_param.G_,
+                                  conv_param.N_,
+                                  conv_param.K_,
+                                  conv_param.output_spatial_lengths_[0],
+                                  conv_param.output_spatial_lengths_[1],
+                                  conv_param.output_spatial_lengths_[2]},
+                                 {
+                                     conv_param.K_, // g
+                                     0,             // n
+                                     1,             // k
+                                     0,             // z
+                                     0,             // y
+                                     0              // x
+                                 });
+
+        const auto out_g_n_k_wos_desc = HostTensorDescriptor(
+            {conv_param.G_,
+             conv_param.N_,
+             conv_param.K_,
+             conv_param.output_spatial_lengths_[0],
+             conv_param.output_spatial_lengths_[1],
+             conv_param.output_spatial_lengths_[2]},
+            {
+                conv_param.K_, // g
+                conv_param.output_spatial_lengths_[0] * conv_param.output_spatial_lengths_[1] *
+                    conv_param.output_spatial_lengths_[2] * conv_param.G_ * conv_param.K_, // n
+                1,                                                                         // k
+                conv_param.output_spatial_lengths_[1] * conv_param.output_spatial_lengths_[2] *
+                    conv_param.G_ * conv_param.K_,                                     // do
+                conv_param.output_spatial_lengths_[2] * conv_param.G_ * conv_param.K_, // ho
+                conv_param.G_ * conv_param.K_                                          // wo
+            });
+
+        return run_grouped_conv_fwd_bias<
+            3,
+            InDataType,
+            WeiDataType,
+            OutDataType,
+            InElementOp,
+            WeiElementOp,
+            OutElementOp,
+            DeviceGroupedConvNDFwdInstance<3, InLayout, WeiLayout, BiasLayout, OutLayout>>(
+            do_verification,
+            init_method,
+            time_kernel,
+            conv_param,
+            in_g_n_c_wis_desc,
+            wei_g_k_c_xs_desc,
+            bias_g_n_k_wos_desc,
+            out_g_n_k_wos_desc,
+            in_element_op,
+            wei_element_op,
+            out_element_op);
+    }
+
+    return 0;
+}
diff --git a/example/CMakeLists.txt b/example/CMakeLists.txt
index f1996898f9..9e5843ce0c 100644
--- a/example/CMakeLists.txt
+++ b/example/CMakeLists.txt
@@ -8,7 +8,7 @@ add_custom_target(examples)
 function(add_example_executable EXAMPLE_NAME FILE_NAME)
     message("adding example ${EXAMPLE_NAME}")
     add_executable(${EXAMPLE_NAME} ${FILE_NAME})
-    target_link_libraries(${EXAMPLE_NAME} PRIVATE host_tensor)
+    target_link_libraries(${EXAMPLE_NAME} PRIVATE utility)
     add_test(NAME ${EXAMPLE_NAME} COMMAND $<TARGET_FILE:${EXAMPLE_NAME}> ${ARGN})
     add_dependencies(examples ${EXAMPLE_NAME})
     add_dependencies(check ${EXAMPLE_NAME})
@@ -17,7 +17,7 @@ endfunction(add_example_executable EXAMPLE_NAME)
 function(add_example_executable_no_testing EXAMPLE_NAME FILE_NAME)
     message("adding example ${EXAMPLE_NAME}")
     add_executable(${EXAMPLE_NAME} ${FILE_NAME})
-    target_link_libraries(${EXAMPLE_NAME} PRIVATE host_tensor)
+    target_link_libraries(${EXAMPLE_NAME} PRIVATE utility)
     add_dependencies(examples ${EXAMPLE_NAME})
 endfunction(add_example_executable_no_testing EXAMPLE_NAME)
 
@@ -25,26 +25,23 @@ add_subdirectory(01_gemm)
 add_subdirectory(02_gemm_bilinear)
 add_subdirectory(03_gemm_bias_relu)
 add_subdirectory(04_gemm_add_add_fastgelu)
-add_subdirectory(06_conv2d_fwd_bias_relu)
-add_subdirectory(07_conv2d_fwd_bias_relu_add)
 add_subdirectory(09_convnd_fwd)
-add_subdirectory(10_conv2d_bwd_data)
-add_subdirectory(11_conv2d_bwd_weight)
 add_subdirectory(12_reduce)
 add_subdirectory(13_pool2d_fwd)
 add_subdirectory(14_gemm_xdl_requant_relu_requant)
 add_subdirectory(15_grouped_gemm)
 add_subdirectory(16_gemm_reduce)
-add_subdirectory(17_convnd_bwd_data_xdl)
+add_subdirectory(17_convnd_bwd_data)
 add_subdirectory(18_batched_gemm_reduce)
 add_subdirectory(19_binary_elementwise)
-add_subdirectory(20_convnd_bwd_weight_xdl)
+add_subdirectory(20_convnd_bwd_weight)
 add_subdirectory(21_gemm_layernorm)
 add_subdirectory(22_cgemm)
 add_subdirectory(23_softmax)
-add_subdirectory(24_batched_gemm_c_permute)
-add_subdirectory(25_gemm_bias_c_permute)
+add_subdirectory(24_batched_gemm_e_permute)
+add_subdirectory(25_gemm_bias_e_permute)
 add_subdirectory(26_contraction)
 add_subdirectory(27_layernorm)
 add_subdirectory(28_grouped_gemm_bias)
-add_subdirectory(29_batched_gemm_multi_d)
\ No newline at end of file
+add_subdirectory(29_batched_gemm_multi_d)
+add_subdirectory(30_grouped_convnd_fwd_bias_relu)
diff --git a/include/ck/ck.hpp b/include/ck/ck.hpp
index 2721f7d1f8..fcaec592e8 100644
--- a/include/ck/ck.hpp
+++ b/include/ck/ck.hpp
@@ -146,7 +146,7 @@
 
 // workaround: verifaction failure, due to compiler regression, for conv bwd-data fp16 using some
 // tuning parameter
-#define CK_WORKAROUND_SWDEV_325164 1
+#define CK_WORKAROUND_SWDEV_325164 0
 
 namespace ck {
 
diff --git a/include/ck/device_utility/device_prop.hpp b/include/ck/host_utility/device_prop.hpp
similarity index 100%
rename from include/ck/device_utility/device_prop.hpp
rename to include/ck/host_utility/device_prop.hpp
diff --git a/include/ck/device_utility/hip_check_error.hpp b/include/ck/host_utility/hip_check_error.hpp
similarity index 100%
rename from include/ck/device_utility/hip_check_error.hpp
rename to include/ck/host_utility/hip_check_error.hpp
diff --git a/include/ck/host_utility/io.hpp b/include/ck/host_utility/io.hpp
new file mode 100644
index 0000000000..ac8719592d
--- /dev/null
+++ b/include/ck/host_utility/io.hpp
@@ -0,0 +1,41 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+#include <iostream>
+#include <vector>
+#include <iterator>
+
+#include "ck/tensor_description/tensor_descriptor.hpp"
+
+template <typename T>
+std::ostream& operator<<(std::ostream& os, const std::vector<T>& v)
+{
+    std::copy(std::begin(v), std::end(v), std::ostream_iterator<T>(os, " "));
+    return os;
+}
+
+template <typename T, std::size_t N>
+std::ostream& operator<<(std::ostream& os, const std::array<T, N>& v)
+{
+    std::copy(std::begin(v), std::end(v), std::ostream_iterator<T>(os, " "));
+    return os;
+}
+
+template <typename... Ts>
+std::ostream& operator<<(std::ostream& os, const ck::TensorDescriptor<Ts...>& desc)
+{
+    constexpr ck::index_t nDim = ck::remove_cvref_t<decltype(desc)>::GetNumOfDimension();
+
+    os << "{";
+
+    ck::static_for<0, nDim - 1, 1>{}([&](auto i) { os << desc.GetLength(i) << ", "; });
+
+    os << desc.GetLength(ck::Number<nDim - 1>{});
+
+    os << "}";
+
+    return os;
+}
diff --git a/include/ck/device_utility/kernel_launch.hpp b/include/ck/host_utility/kernel_launch.hpp
similarity index 97%
rename from include/ck/device_utility/kernel_launch.hpp
rename to include/ck/host_utility/kernel_launch.hpp
index 5879f9995e..ed6e2f0ba1 100644
--- a/include/ck/device_utility/kernel_launch.hpp
+++ b/include/ck/host_utility/kernel_launch.hpp
@@ -7,7 +7,7 @@
 
 #include "ck/ck.hpp"
 #include "ck/stream_config.hpp"
-#include "ck/device_utility/hip_check_error.hpp"
+#include "ck/host_utility/hip_check_error.hpp"
 
 template <typename... Args, typename F>
 float launch_and_time_kernel(const StreamConfig& stream_config,
diff --git a/include/ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp b/include/ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp
index 6a226b0c53..a4a29f5d5e 100644
--- a/include/ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp
+++ b/include/ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp
@@ -1,8 +1,7 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#ifndef CONVOLUTION_BACKWARD_DATA_SPECIALIZATION
-#define CONVOLUTION_BACKWARD_DATA_SPECIALIZATION
+#pragma once
 
 namespace ck {
 namespace tensor_operation {
@@ -14,7 +13,18 @@ enum struct ConvolutionBackwardDataSpecialization
     Filter1x1Stride1Pad0,
 };
 
+inline std::string
+getConvBackwardDataSpecializationString(const ConvolutionBackwardDataSpecialization& s)
+{
+    switch(s)
+    {
+    case ConvolutionBackwardDataSpecialization::Default: return "Default";
+    case ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0:
+        return "FFilter1x1Stride1Pad0";
+    default: return "Unrecognized specialization!";
+    }
+}
+
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
-#endif
diff --git a/include/ck/tensor_operation/gpu/device/convolution_backward_weight_specialization.hpp b/include/ck/tensor_operation/gpu/device/convolution_backward_weight_specialization.hpp
index f4607ee612..20b2a152b9 100644
--- a/include/ck/tensor_operation/gpu/device/convolution_backward_weight_specialization.hpp
+++ b/include/ck/tensor_operation/gpu/device/convolution_backward_weight_specialization.hpp
@@ -15,6 +15,19 @@ enum struct ConvolutionBackwardWeightSpecialization
     OddC,
 };
 
+inline std::string
+getConvBackwardWeightSpecializationString(const ConvolutionBackwardWeightSpecialization& s)
+{
+    switch(s)
+    {
+    case ConvolutionBackwardWeightSpecialization::Default: return "Default";
+    case ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0:
+        return "Filter1x1Stride1Pad0";
+    case ConvolutionBackwardWeightSpecialization::Filter1x1Pad0: return "Filter1x1Pad0";
+    case ConvolutionBackwardWeightSpecialization::OddC: return "OddC";
+    default: return "Unrecognized specialization!";
+    }
+}
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp b/include/ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp
index ea60a4e6d9..953ff1e06e 100644
--- a/include/ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp
+++ b/include/ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp
@@ -1,8 +1,7 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#ifndef CONVOLUTION_FORWARD_SPECIALIZATION
-#define CONVOLUTION_FORWARD_SPECIALIZATION
+#pragma once
 
 #include <string>
 
@@ -33,4 +32,3 @@ inline std::string getConvForwardSpecializationString(const ConvolutionForwardSp
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
-#endif
diff --git a/include/ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp b/include/ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp
index c228045bdb..bd8d7756d2 100644
--- a/include/ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_5ary_elementwise.hpp
@@ -12,8 +12,8 @@
 #include "ck/tensor_description/tensor_descriptor_helper.hpp"
 #include "ck/tensor_operation/gpu/device/device_elementwise.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_5ary_Elementwise_1d.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute_xdl.hpp
index 432dcb5d57..6b5e0dc565 100644
--- a/include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute_xdl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute_xdl.hpp
@@ -11,8 +11,8 @@
 #include "ck/tensor_operation/gpu/device/device_batched_gemm_multi_d_xdl.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute.hpp
similarity index 52%
rename from include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute.hpp
rename to include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute.hpp
index 7041954097..acd779b2db 100644
--- a/include/ck/tensor_operation/gpu/device/device_batched_gemm_c_permute.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute.hpp
@@ -8,7 +8,7 @@ namespace ck {
 namespace tensor_operation {
 namespace device {
 
-struct BatchedGemmCPermuteDesc
+struct BatchedGemmEPermuteDesc
 {
     ck::index_t G0_, G1_, M_, N_;
     ck::index_t stride_G0_, stride_G1_, stride_M_, stride_N_;
@@ -23,12 +23,12 @@ template <typename ALayout,
           typename AElementwiseOperation,
           typename BElementwiseOperation,
           typename CDEElementwiseOperation>
-struct DeviceBatchedGemmCPermute : public BaseOperator
+struct DeviceBatchedGemmEPermute : public BaseOperator
 {
     virtual std::unique_ptr<BaseArgument>
     MakeArgumentPointer(const void* p_a,
                         const void* p_b,
-                        void* p_c,
+                        void* p_e,
                         index_t M,
                         index_t N,
                         index_t K,
@@ -36,35 +36,15 @@ struct DeviceBatchedGemmCPermute : public BaseOperator
                         index_t stride_B,
                         index_t batch_stride_A,
                         index_t batch_stride_B,
-                        BatchedGemmCPermuteDesc batched_gemm_c_permute_desc,
+                        BatchedGemmEPermuteDesc batched_gemm_e_permute_desc,
                         index_t BatchCount,
                         AElementwiseOperation a_element_op,
                         BElementwiseOperation b_element_op,
-                        CDEElementwiseOperation c_element_op) = 0;
+                        CDEElementwiseOperation cde_element_op) = 0;
 
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename ALayout,
-          typename BLayout,
-          typename DELayout,
-          typename ADataType,
-          typename BDataType,
-          typename EDataType,
-          typename AElementwiseOperation,
-          typename BElementwiseOperation,
-          typename CDEElementwiseOperation>
-using DeviceBatchedGemmCPermutePtr =
-    std::unique_ptr<DeviceBatchedGemmCPermute<ALayout,
-                                              BLayout,
-                                              DELayout,
-                                              ADataType,
-                                              BDataType,
-                                              EDataType,
-                                              AElementwiseOperation,
-                                              BElementwiseOperation,
-                                              CDEElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute_xdl.hpp
new file mode 100644
index 0000000000..8c5dc7de1f
--- /dev/null
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_e_permute_xdl.hpp
@@ -0,0 +1,682 @@
+#pragma once
+
+#include <iostream>
+#include <sstream>
+
+#include "ck/utility/common_header.hpp"
+#include "ck/tensor_description/tensor_descriptor.hpp"
+#include "ck/tensor_description/tensor_descriptor_helper.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_batched_gemm_e_permute.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
+#include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+
+/*
+ * \brief Wrapper function of GridwiseGemm::Run to realize BatchedGEMM.
+ *
+ * \tparam ComputePtrOffsetOfBatch Class that computes the base pointer offsets of A, B, C matrix
+ * given the batch. For example, ComputePtrOffsetOfStridedBatch() computes the offsets of evenly
+ * strided batched, but we can easily extend to other layouts. The returned offset can be either \p
+ * index_t or \p long_index_t. If it returns \p long_index_t, we are not subject to the 2GB
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
+ * limitations.
+ *
+ * \tparam Block2ETileMap Block2ETileMap::CalculateBottomIndex() takes in id of a workgroup and
+ * returns the 2D index of the tile that it computes. \see
+ * GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3::Run().
+ * \note Using \p ComputePtrOffsetOfBatch gives us the flexibility that 2 workgroups can compute 2
+ * tiles from different matrices. Keep in mind that these 2 matrices can share the same grid
+ * descriptor (like in BatchedGEMM), or use their own grid descriptors (in GroupedGemm). \link
+ * device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp kernel_gemm_xdlops_v2r3_for_conv3d \endlink for \link
+ * DeviceConv3d \endlink uses the same concept, but currently does NOT encapsulate the computing of
+ * pointer offset into \p ComputePtrOffsetOfStridedBatch.
+ *
+ * \note \p Block2ETileMap allows customized mapping between a workgroup and the C-tile it computes.
+ * Together with \p ComputePtrOffsetOfBatch, we can reuse GridwiseGemm (and GridwiseGemm fusion ) to
+ * realize BatchedGemmCPermute and GroupedGemm (and the corresponding GEMM fusion).
+ *
+ */
+template <typename GridwiseGemm,
+          typename ABDataType,
+          typename EDataType,
+          typename AGridDesc_AK0_M_AK1,
+          typename BGridDesc_BK0_N_BK1,
+          typename EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock,
+          typename AElementwiseOperation,
+          typename BElementwiseOperation,
+          typename CDEElementwiseOperation,
+          typename ComputePtrOffsetOfBatch,
+          typename Block2ETileMap,
+          bool HasMainKBlockLoop>
+__global__ void
+#if CK_USE_LAUNCH_BOUNDS
+    __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU)
+#endif
+        kernel_batched_gemm_e_permute_xdl(const ABDataType* __restrict__ p_a_grid,
+                                          const ABDataType* __restrict__ p_b_grid,
+                                          EDataType* __restrict__ p_e_grid,
+                                          const index_t batch_count,
+                                          const AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1,
+                                          const BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1,
+                                          const EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock
+                                              e_grid_desc_mblock_mperblock_nblock_nperblock,
+                                          const AElementwiseOperation a_element_op,
+                                          const BElementwiseOperation b_element_op,
+                                          const CDEElementwiseOperation cde_element_op,
+                                          const ComputePtrOffsetOfBatch compute_ptr_offset_of_batch,
+                                          const Block2ETileMap block_2_etile_map)
+{
+#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__))
+    const index_t num_blocks_per_batch =
+        __builtin_amdgcn_readfirstlane(get_grid_size() / batch_count);
+    const index_t g_idx = __builtin_amdgcn_readfirstlane(get_block_1d_id() / num_blocks_per_batch);
+
+    const long_index_t a_batch_offset = __builtin_amdgcn_readfirstlane(
+        static_cast<long_index_t>(compute_ptr_offset_of_batch.GetAPtrOffset(g_idx)));
+    const long_index_t b_batch_offset = __builtin_amdgcn_readfirstlane(
+        static_cast<long_index_t>(compute_ptr_offset_of_batch.GetBPtrOffset(g_idx)));
+    const long_index_t e_batch_offset = __builtin_amdgcn_readfirstlane(
+        static_cast<long_index_t>(compute_ptr_offset_of_batch.GetCPtrOffset(g_idx)));
+
+    __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
+
+    GridwiseGemm::template Run<HasMainKBlockLoop>(p_a_grid + a_batch_offset,
+                                                  p_b_grid + b_batch_offset,
+                                                  ck::Tuple<>{},
+                                                  p_e_grid + e_batch_offset,
+                                                  p_shared,
+                                                  a_element_op,
+                                                  b_element_op,
+                                                  cde_element_op,
+                                                  a_grid_desc_ak0_m_ak1,
+                                                  b_grid_desc_bk0_n_bk1,
+                                                  ck::Tuple<>{},
+                                                  e_grid_desc_mblock_mperblock_nblock_nperblock,
+                                                  block_2_etile_map);
+#else
+    ignore = p_a_grid;
+    ignore = p_b_grid;
+    ignore = p_e_grid;
+    ignore = batch_count;
+    ignore = a_grid_desc_ak0_m_ak1;
+    ignore = b_grid_desc_bk0_n_bk1;
+    ignore = e_grid_desc_mblock_mperblock_nblock_nperblock;
+    ignore = a_element_op;
+    ignore = b_element_op;
+    ignore = cde_element_op;
+    ignore = compute_ptr_offset_of_batch;
+    ignore = block_2_etile_map;
+#endif
+}
+
+template <typename ALayout,
+          typename BLayout,
+          typename ELayout,
+          typename ADataType,
+          typename BDataType,
+          typename AccDataType,
+          typename CShuffleDataType,
+          typename EDataType,
+          typename AElementwiseOperation,
+          typename BElementwiseOperation,
+          typename CDEElementwiseOperation,
+          GemmSpecialization GemmSpec,
+          index_t NumPrefetch,
+          index_t BlockSize,
+          index_t MPerBlock,
+          index_t NPerBlock,
+          index_t KPerBlock,
+          index_t AK1,
+          index_t BK1,
+          index_t MPerXDL,
+          index_t NPerXDL,
+          index_t MXdlPerWave,
+          index_t NXdlPerWave,
+          typename ABlockTransferThreadClusterLengths_K0_M_K1,
+          typename ABlockTransferThreadClusterArrangeOrder,
+          typename ABlockTransferSrcAccessOrder,
+          index_t ABlockTransferSrcVectorDim,
+          index_t ABlockTransferSrcScalarPerVector,
+          index_t ABlockTransferDstScalarPerVector_K1,
+          index_t ABlockLdsExtraM,
+          typename BBlockTransferThreadClusterLengths_K0_N_K1,
+          typename BBlockTransferThreadClusterArrangeOrder,
+          typename BBlockTransferSrcAccessOrder,
+          index_t BBlockTransferSrcVectorDim,
+          index_t BBlockTransferSrcScalarPerVector,
+          index_t BBlockTransferDstScalarPerVector_K1,
+          index_t BBlockLdsExtraN,
+          index_t CShuffleMXdlPerWavePerShuffle,
+          index_t CShuffleNXdlPerWavePerShuffle,
+          typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
+          index_t CDEBlockTransferScalarPerVector_NPerBlock,
+          LoopScheduler LoopSched = make_default_loop_scheduler()>
+struct DeviceBatchedGemmEPermuteXdl : public DeviceBatchedGemmEPermute<ALayout,
+                                                                       BLayout,
+                                                                       ELayout,
+                                                                       ADataType,
+                                                                       BDataType,
+                                                                       EDataType,
+                                                                       AElementwiseOperation,
+                                                                       BElementwiseOperation,
+                                                                       CDEElementwiseOperation>
+{
+    using DeviceOp = DeviceBatchedGemmEPermuteXdl;
+
+    static constexpr auto I0 = Number<0>{};
+    static constexpr auto I1 = Number<1>{};
+    static constexpr auto I2 = Number<2>{};
+
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
+
+    static auto MakeAGridDescriptor_M_K(index_t MRaw, index_t KRaw, index_t StrideA)
+    {
+        const auto a_grid_desc_mraw_kraw = [&]() {
+            if constexpr(is_same_v<tensor_layout::gemm::RowMajor, ALayout>)
+            {
+                return make_naive_tensor_descriptor(make_tuple(MRaw, KRaw),
+                                                    make_tuple(StrideA, I1));
+            }
+            else if constexpr(is_same_v<tensor_layout::gemm::ColumnMajor, ALayout>)
+            {
+                return make_naive_tensor_descriptor(make_tuple(MRaw, KRaw),
+                                                    make_tuple(I1, StrideA));
+            }
+        }();
+
+        return matrix_padder.PadADescriptor_M_K(a_grid_desc_mraw_kraw);
+    }
+
+    static auto MakeBGridDescriptor_N_K(index_t KRaw, index_t NRaw, index_t StrideB)
+    {
+        const auto b_grid_desc_nraw_kraw = [&]() {
+            if constexpr(is_same<tensor_layout::gemm::RowMajor, BLayout>::value)
+            {
+                return make_naive_tensor_descriptor(make_tuple(NRaw, KRaw),
+                                                    make_tuple(I1, StrideB));
+            }
+            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, BLayout>::value)
+            {
+                return make_naive_tensor_descriptor(make_tuple(NRaw, KRaw),
+                                                    make_tuple(StrideB, I1));
+            }
+        }();
+
+        return matrix_padder.PadBDescriptor_N_K(b_grid_desc_nraw_kraw);
+    }
+
+    static auto
+    MakeEGridDescriptor_M_N(index_t MRaw, index_t NRaw, index_t stride_M, index_t stride_N)
+    {
+        const auto e_grid_desc_mraw_nraw =
+            make_naive_tensor_descriptor(make_tuple(MRaw, NRaw), make_tuple(stride_M, stride_N));
+
+        return matrix_padder.PadCDescriptor_M_N(e_grid_desc_mraw_nraw);
+    }
+
+    static auto MakeEGridDescriptor_G0_G1_M_N(index_t G0,
+                                              index_t G1,
+                                              index_t MRaw,
+                                              index_t NRaw,
+                                              index_t stride_G0,
+                                              index_t stride_G1,
+                                              index_t stride_M,
+                                              index_t stride_N)
+    {
+        const auto e_grid_desc_g0_g1_mraw_nraw = [&]() {
+            return make_naive_tensor_descriptor(
+                make_tuple(G0, G1, MRaw, NRaw),
+                make_tuple(stride_G0, stride_G1, stride_M, stride_N));
+        }();
+
+        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
+        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
+
+        const auto MPad = M - MRaw;
+        const auto NPad = N - NRaw;
+
+        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
+                     GemmSpec == GemmSpecialization::MNKPadding)
+        {
+            // pad M and N
+            return transform_tensor_descriptor(
+                e_grid_desc_g0_g1_mraw_nraw,
+                make_tuple(make_pass_through_transform(G0),
+                           make_pass_through_transform(G1),
+                           make_right_pad_transform(MRaw, MPad),
+                           make_right_pad_transform(NRaw, NPad)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
+                          GemmSpec == GemmSpecialization::MKPadding)
+        {
+            // pad M, but not N
+            return transform_tensor_descriptor(
+                e_grid_desc_g0_g1_mraw_nraw,
+                make_tuple(make_pass_through_transform(G0),
+                           make_pass_through_transform(G1),
+                           make_right_pad_transform(MRaw, MPad),
+                           make_pass_through_transform(NRaw)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
+                          GemmSpec == GemmSpecialization::NKPadding)
+        {
+            // pad N, but not M
+            return transform_tensor_descriptor(
+                e_grid_desc_g0_g1_mraw_nraw,
+                make_tuple(make_pass_through_transform(G0),
+                           make_pass_through_transform(G1),
+                           make_pass_through_transform(MRaw),
+                           make_right_pad_transform(NRaw, NPad)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+        }
+        else
+        {
+            // not pad M or N
+            return e_grid_desc_g0_g1_mraw_nraw;
+        }
+    }
+
+    using AGridDesc_M_K       = decltype(MakeAGridDescriptor_M_K(1, 1, 1));
+    using BGridDesc_N_K       = decltype(MakeBGridDescriptor_N_K(1, 1, 1));
+    using EGridDesc_M_N       = decltype(MakeEGridDescriptor_M_N(1, 1, 1, 1));
+    using EGridDesc_G0_G1_M_N = decltype(MakeEGridDescriptor_G0_G1_M_N(1, 1, 1, 1, 1, 1, 1, 1));
+
+    struct ComputePtrOffsetOfStridedBatch
+    {
+        ComputePtrOffsetOfStridedBatch(index_t Batchstride_A,
+                                       index_t Batchstride_B,
+                                       EGridDesc_G0_G1_M_N e_grid_desc_g0_g1_m_n)
+            : Batchstride_A_(Batchstride_A),
+              Batchstride_B_(Batchstride_B),
+              e_grid_desc_g0_g1_m_n_(e_grid_desc_g0_g1_m_n)
+        {
+        }
+
+        __host__ __device__ constexpr long_index_t GetAPtrOffset(index_t g_idx) const
+        {
+            return g_idx * static_cast<long_index_t>(Batchstride_A_);
+        }
+
+        __host__ __device__ constexpr long_index_t GetBPtrOffset(index_t g_idx) const
+        {
+            return g_idx * static_cast<long_index_t>(Batchstride_B_);
+        }
+
+        __host__ __device__ constexpr long_index_t GetCPtrOffset(index_t g_idx) const
+        {
+            const index_t G1 = e_grid_desc_g0_g1_m_n_.GetLength(I1);
+            index_t b0       = g_idx / G1;
+            index_t b1       = g_idx - b0 * G1; // g_idx % G1
+            return e_grid_desc_g0_g1_m_n_.CalculateOffset(make_multi_index(b0, b1, 0, 0));
+        }
+
+        private:
+        index_t Batchstride_A_;
+        index_t Batchstride_B_;
+        EGridDesc_G0_G1_M_N e_grid_desc_g0_g1_m_n_;
+    };
+
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
+        ADataType, // TODO: distinguish A/B datatype
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>, // DsDataType,
+        EDataType,   // EDataType,
+        AElementwiseOperation,
+        BElementwiseOperation,
+        CDEElementwiseOperation,
+        InMemoryDataOperationEnum::Set,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        Tuple<>,
+        EGridDesc_M_N,
+        NumPrefetch,
+        BlockSize,
+        MPerBlock,
+        NPerBlock,
+        KPerBlock,
+        AK1,
+        BK1,
+        MPerXDL,
+        NPerXDL,
+        MXdlPerWave,
+        NXdlPerWave,
+        ABlockTransferThreadClusterLengths_K0_M_K1,
+        ABlockTransferThreadClusterArrangeOrder,
+        ABlockTransferSrcAccessOrder,
+        ABlockTransferSrcVectorDim,
+        ABlockTransferSrcScalarPerVector,
+        ABlockTransferDstScalarPerVector_K1,
+        false, // AThreadTransferSrcResetCoordinateAfterRun,
+        ABlockLdsExtraM,
+        BBlockTransferThreadClusterLengths_K0_N_K1,
+        BBlockTransferThreadClusterArrangeOrder,
+        BBlockTransferSrcAccessOrder,
+        BBlockTransferSrcVectorDim,
+        BBlockTransferSrcScalarPerVector,
+        BBlockTransferDstScalarPerVector_K1,
+        false, // BThreadTransferSrcResetCoordinateAfterRun,
+        BBlockLdsExtraN,
+        CShuffleMXdlPerWavePerShuffle,
+        CShuffleNXdlPerWavePerShuffle,
+        CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
+        CDEBlockTransferScalarPerVector_NPerBlock,
+        LoopSched>;
+
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
+    using EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock = decltype(
+        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(EGridDesc_M_N{}));
+    using Block2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+
+    // Argument
+    struct Argument : public BaseArgument
+    {
+        Argument(const ADataType* p_a_grid,
+                 const BDataType* p_b_grid,
+                 EDataType* p_e_grid,
+                 index_t M,
+                 index_t N,
+                 index_t K,
+                 index_t stride_A,
+                 index_t stride_B,
+                 index_t batch_stride_A,
+                 index_t batch_stride_B,
+                 BatchedGemmEPermuteDesc batched_gemm_e_permute_desc,
+                 index_t BatchCount,
+                 AElementwiseOperation a_element_op,
+                 BElementwiseOperation b_element_op,
+                 CDEElementwiseOperation cde_element_op)
+            : p_a_grid_{p_a_grid},
+              p_b_grid_{p_b_grid},
+              p_e_grid_{p_e_grid},
+              BatchCount_(BatchCount),
+              a_grid_desc_m_k_{DeviceOp::MakeAGridDescriptor_M_K(M, K, stride_A)},
+              b_grid_desc_n_k_{DeviceOp::MakeBGridDescriptor_N_K(K, N, stride_B)},
+              e_grid_desc_m_n_{
+                  DeviceOp::MakeEGridDescriptor_M_N(batched_gemm_e_permute_desc.M_,
+                                                    batched_gemm_e_permute_desc.N_,
+                                                    batched_gemm_e_permute_desc.stride_M_,
+                                                    batched_gemm_e_permute_desc.stride_N_)},
+              a_grid_desc_ak0_m_ak1_{
+                  GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k_)},
+              b_grid_desc_bk0_n_bk1_{
+                  GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k_)},
+              e_grid_desc_mblock_mperblock_nblock_nperblock{},
+              e_grid_desc_g0_g1_m_n_{
+                  DeviceOp::MakeEGridDescriptor_G0_G1_M_N(batched_gemm_e_permute_desc.G0_,
+                                                          batched_gemm_e_permute_desc.G1_,
+                                                          batched_gemm_e_permute_desc.M_,
+                                                          batched_gemm_e_permute_desc.N_,
+                                                          batched_gemm_e_permute_desc.stride_G0_,
+                                                          batched_gemm_e_permute_desc.stride_G1_,
+                                                          batched_gemm_e_permute_desc.stride_M_,
+                                                          batched_gemm_e_permute_desc.stride_N_)},
+              compute_ptr_offset_of_batch_{batch_stride_A, batch_stride_B, e_grid_desc_g0_g1_m_n_},
+              block_2_etile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
+              a_element_op_{a_element_op},
+              b_element_op_{b_element_op},
+              cde_element_op_{cde_element_op}
+        {
+            if(GridwiseGemm::CheckValidity(a_grid_desc_m_k_,
+                                           b_grid_desc_n_k_,
+                                           ck::Tuple<>{},
+                                           e_grid_desc_m_n_,
+                                           block_2_etile_map_))
+            {
+                e_grid_desc_mblock_mperblock_nblock_nperblock =
+                    GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                        e_grid_desc_m_n_);
+            }
+        }
+
+        void Print() const
+        {
+            std::cout << "A[M, K]: " << a_grid_desc_m_k_ << std::endl;
+            std::cout << "B[N, K]: " << b_grid_desc_n_k_ << std::endl;
+            std::cout << "C[M, N]: " << e_grid_desc_m_n_ << std::endl;
+        }
+
+        //  private:
+        // pointers
+        const ADataType* p_a_grid_;
+        const BDataType* p_b_grid_;
+        EDataType* p_e_grid_;
+
+        // batch count
+        index_t BatchCount_;
+
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
+        AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
+        BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
+        EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock e_grid_desc_mblock_mperblock_nblock_nperblock;
+        EGridDesc_G0_G1_M_N e_grid_desc_g0_g1_m_n_;
+
+        // for calculating Batch offset
+        ComputePtrOffsetOfStridedBatch compute_ptr_offset_of_batch_;
+
+        // block-to-e-tile map
+        Block2ETileMap block_2_etile_map_;
+
+        // element-wise op
+        AElementwiseOperation a_element_op_;
+        BElementwiseOperation b_element_op_;
+        CDEElementwiseOperation cde_element_op_;
+    };
+
+    // Invoker
+    struct Invoker : public BaseInvoker
+    {
+        using Argument = DeviceOp::Argument;
+
+        float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
+        {
+            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                            arg.b_grid_desc_n_k_,
+                                            ck::Tuple<>{},
+                                            arg.e_grid_desc_m_n_,
+                                            arg.block_2_etile_map_))
+            {
+                throw std::runtime_error(
+                    "wrong! GridwiseBatchedGemmCPermute_km_kn_m0m1n0n1_xdlops_v2r3 has invalid "
+                    "setting");
+            }
+
+            const index_t grid_size =
+                arg.block_2_etile_map_.CalculateGridSize(arg.e_grid_desc_m_n_) * arg.BatchCount_;
+
+            const auto K =
+                arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) * arg.a_grid_desc_ak0_m_ak1_.GetLength(I2);
+
+            auto launch_kernel = [&](auto has_main_k_block_loop_) {
+                const auto kernel = kernel_batched_gemm_e_permute_xdl<
+                    GridwiseGemm,
+                    ADataType, // TODO: distiguish A/B datatype
+                    EDataType,
+                    remove_reference_t<DeviceOp::AGridDesc_AK0_M_AK1>,
+                    remove_reference_t<DeviceOp::BGridDesc_BK0_N_BK1>,
+                    typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
+                    AElementwiseOperation,
+                    BElementwiseOperation,
+                    CDEElementwiseOperation,
+                    ComputePtrOffsetOfStridedBatch,
+                    remove_reference_t<Block2ETileMap>,
+                    has_main_k_block_loop_>;
+
+                return launch_and_time_kernel(stream_config,
+                                              kernel,
+                                              dim3(grid_size),
+                                              dim3(BlockSize),
+                                              0,
+                                              arg.p_a_grid_,
+                                              arg.p_b_grid_,
+                                              arg.p_e_grid_,
+                                              arg.BatchCount_,
+                                              arg.a_grid_desc_ak0_m_ak1_,
+                                              arg.b_grid_desc_bk0_n_bk1_,
+                                              arg.e_grid_desc_mblock_mperblock_nblock_nperblock,
+                                              arg.a_element_op_,
+                                              arg.b_element_op_,
+                                              arg.cde_element_op_,
+                                              arg.compute_ptr_offset_of_batch_,
+                                              arg.block_2_etile_map_);
+            };
+
+            if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
+            {
+                return launch_kernel(integral_constant<bool, true>{});
+            }
+            else
+            {
+                return launch_kernel(integral_constant<bool, false>{});
+            }
+        }
+
+        // polymorphic
+        float Run(const BaseArgument* p_arg,
+                  const StreamConfig& stream_config = StreamConfig{}) override
+        {
+            return Run(*dynamic_cast<const Argument*>(p_arg), stream_config);
+        }
+    };
+
+    static constexpr bool IsValidCompilationParameter()
+    {
+        // TODO: properly implement this check
+        return true;
+    }
+
+    static bool IsSupportedArgument(const Argument& arg)
+    {
+        return GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                           arg.b_grid_desc_n_k_,
+                                           ck::Tuple<>{},
+                                           arg.e_grid_desc_m_n_,
+                                           arg.block_2_etile_map_);
+    }
+
+    // polymorphic
+    bool IsSupportedArgument(const BaseArgument* p_arg) override
+    {
+        return IsSupportedArgument(*dynamic_cast<const Argument*>(p_arg));
+    }
+
+    static auto MakeArgument(const ADataType* p_a,
+                             const BDataType* p_b,
+                             EDataType* p_e,
+                             index_t M,
+                             index_t N,
+                             index_t K,
+                             index_t stride_A,
+                             index_t stride_B,
+                             index_t batch_stride_A,
+                             index_t batch_stride_B,
+                             BatchedGemmEPermuteDesc batched_gemm_e_permute_desc,
+                             index_t BatchCount,
+                             AElementwiseOperation a_element_op,
+                             BElementwiseOperation b_element_op,
+                             CDEElementwiseOperation cde_element_op)
+    {
+        return Argument{p_a,
+                        p_b,
+                        p_e,
+                        M,
+                        N,
+                        K,
+                        stride_A,
+                        stride_B,
+                        batch_stride_A,
+                        batch_stride_B,
+                        batched_gemm_e_permute_desc,
+                        BatchCount,
+                        a_element_op,
+                        b_element_op,
+                        cde_element_op};
+    }
+
+    static auto MakeInvoker() { return Invoker{}; }
+
+    // polymorphic
+    std::unique_ptr<BaseArgument>
+    MakeArgumentPointer(const void* p_a,
+                        const void* p_b,
+                        void* p_e,
+                        index_t M,
+                        index_t N,
+                        index_t K,
+                        index_t stride_A,
+                        index_t stride_B,
+                        index_t batch_stride_A,
+                        index_t batch_stride_B,
+                        BatchedGemmEPermuteDesc batched_gemm_e_permute_desc,
+                        index_t BatchCount,
+                        AElementwiseOperation a_element_op,
+                        BElementwiseOperation b_element_op,
+                        CDEElementwiseOperation cde_element_op) override
+    {
+        return std::make_unique<Argument>(static_cast<const ADataType*>(p_a),
+                                          static_cast<const BDataType*>(p_b),
+                                          static_cast<EDataType*>(p_e),
+                                          M,
+                                          N,
+                                          K,
+                                          stride_A,
+                                          stride_B,
+                                          batch_stride_A,
+                                          batch_stride_B,
+                                          batched_gemm_e_permute_desc,
+                                          BatchCount,
+                                          a_element_op,
+                                          b_element_op,
+                                          cde_element_op);
+    }
+
+    // polymorphic
+    std::unique_ptr<BaseInvoker> MakeInvokerPointer() override
+    {
+        return std::make_unique<Invoker>(Invoker{});
+    }
+
+    // polymorphic
+    std::string GetTypeString() const override
+    {
+        auto str = std::stringstream();
+
+        // clang-format off
+        str << "DeviceBatchedGemmEPermuteXdl"
+            << "<"
+            << BlockSize << ", "
+            << MPerBlock << ", "
+            << NPerBlock << ", "
+            << KPerBlock
+            << ">";
+        // clang-format on
+
+        return str.str();
+    }
+};
+
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d.hpp
index ca3f574d1e..116e62c009 100644
--- a/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d.hpp
@@ -14,7 +14,8 @@ namespace device {
 
 template <typename ALayout,
           typename BLayout,
-          typename CLayout,
+          typename DsLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
           typename DsDataType,
@@ -26,23 +27,25 @@ struct DeviceBatchedGemmMultiD : public BaseOperator
 {
     static constexpr index_t NumDTensor = DsDataType::Size();
 
+    static_assert(DsLayout::Size() == DsDataType::Size(), "wrong! inconsisiten NumDTensor");
+
     virtual std::unique_ptr<BaseArgument>
     MakeArgumentPointer(const void* p_a,
                         const void* p_b,
-                        std::array<const void*, NumDTensor> p_ds,
-                        void* p_c,
-                        ck::index_t M,
-                        ck::index_t N,
-                        ck::index_t K,
-                        ck::index_t StrideA,
-                        ck::index_t StrideB,
-                        std::array<ck::index_t, NumDTensor> StrideDs,
-                        ck::index_t StrideE,
-                        ck::index_t BatchStrideA,
-                        ck::index_t BatchStrideB,
-                        std::array<ck::index_t, NumDTensor> BatchStrideDs,
-                        ck::index_t BatchStrideE,
-                        ck::index_t Batch,
+                        const std::array<const void*, NumDTensor>& p_ds,
+                        void* p_e,
+                        index_t M,
+                        index_t N,
+                        index_t K,
+                        index_t Batch,
+                        index_t StrideA,
+                        index_t StrideB,
+                        const std::array<ck::index_t, NumDTensor>& StrideDs,
+                        index_t StrideE,
+                        index_t BatchStrideA,
+                        index_t BatchStrideB,
+                        const std::array<ck::index_t, NumDTensor>& BatchStrideDs,
+                        index_t BatchStrideE,
                         AElementwiseOperation a_element_op,
                         BElementwiseOperation b_element_op,
                         CDEElementwiseOperation cde_element_op) = 0;
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d_xdl.hpp
index 1cf3e80c50..4dc170a034 100644
--- a/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d_xdl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d_xdl.hpp
@@ -12,9 +12,10 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/device_batched_gemm_multi_d.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -29,7 +30,7 @@ namespace device {
  * index_t or \p long_index_t. If it returns \p long_index_t, we are not subject to the 2GB
  * limitations.
  *
- * \tparam Block2CTileMap Block2CTileMap::CalculateBottomIndex() takes in id of a workgroup and
+ * \tparam Block2ETileMap Block2ETileMap::CalculateBottomIndex() takes in id of a workgroup and
  * returns the 2D index of the tile that it computes. \see
  * GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3::Run().
  *
@@ -40,45 +41,45 @@ namespace device {
  * DeviceConv3d \endlink uses the same concept, but currently does NOT encapsulate the computing of
  * pointer offset into \p ComputePtrOffsetOfStridedBatch.
  *
- * \note \p Block2CTileMap allows customized mapping between a workgroup and the C-tile it computes.
+ * \note \p Block2ETileMap allows customized mapping between a workgroup and the C-tile it computes.
  * Together with \p ComputePtrOffsetOfBatch, we can reuse GridwiseGemm (and GridwiseGemm fusion ) to
  * realize BatchedGemm and GroupedGemm (and the corresponding GEMM fusion).
  *
  */
 template <typename GridwiseGemm,
-          typename FloatAB,
-          typename FloatDsPointer,
-          typename FloatC,
+          typename ABDataType,
+          typename DsPointer,
+          typename EDataType,
+          typename AElementwiseOperation,
+          typename BElementwiseOperation,
+          typename CDEElementwiseOperation,
           typename AGridDesc_AK0_M_AK1,
           typename BGridDesc_BK0_N_BK1,
           typename DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
           typename EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock,
-          typename AElementwiseOperation,
-          typename BElementwiseOperation,
-          typename CDEElementwiseOperation,
           typename ComputePtrOffsetOfBatch,
-          typename Block2CTileMap,
+          typename Block2ETileMap,
           bool HasMainKBlockLoop>
 __global__ void
 #if CK_USE_LAUNCH_BOUNDS
     __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU)
 #endif
-        kernel_batched_gemm_xdl(const FloatAB* __restrict__ p_a_grid,
-                                const FloatAB* __restrict__ p_b_grid,
-                                FloatDsPointer p_ds_grid,
-                                FloatC* __restrict__ p_e_grid,
+        kernel_batched_gemm_xdl(const ABDataType* __restrict__ p_a_grid,
+                                const ABDataType* __restrict__ p_b_grid,
+                                DsPointer p_ds_grid,
+                                EDataType* __restrict__ p_e_grid,
                                 const index_t batch_count,
+                                const AElementwiseOperation a_element_op,
+                                const BElementwiseOperation b_element_op,
+                                const CDEElementwiseOperation cde_element_op,
                                 const AGridDesc_AK0_M_AK1 a_grid_desc_k0_m_k1,
                                 const BGridDesc_BK0_N_BK1 b_grid_desc_k0_n_k1,
                                 const DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
                                     ds_grid_desc_mblock_mperblock_nblock_nperblock,
                                 const EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock
                                     e_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                const AElementwiseOperation a_element_op,
-                                const BElementwiseOperation b_element_op,
-                                const CDEElementwiseOperation cde_element_op,
                                 const ComputePtrOffsetOfBatch compute_ptr_offset_of_batch,
-                                const Block2CTileMap block_2_ctile_map)
+                                const Block2ETileMap block_2_etile_map)
 {
 
 #if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__))
@@ -97,7 +98,7 @@ __global__ void
 
     __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
 
-    FloatDsPointer p_ds_grid_grp;
+    DsPointer p_ds_grid_grp;
 
     static constexpr index_t NumDTensor =
         DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock::Size();
@@ -117,7 +118,7 @@ __global__ void
                                                   b_grid_desc_k0_n_k1,
                                                   ds_grid_desc_mblock_mperblock_nblock_nperblock,
                                                   e_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                                  block_2_ctile_map);
+                                                  block_2_etile_map);
 #else
     ignore = p_a_grid;
     ignore = p_b_grid;
@@ -132,16 +133,17 @@ __global__ void
     ignore = b_element_op;
     ignore = cde_element_op;
     ignore = compute_ptr_offset_of_batch;
-    ignore = block_2_ctile_map;
+    ignore = block_2_etile_map;
 #endif
 }
 
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename DsLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
-          typename GemmAccDataType,
+          typename AccDataType,
           typename CShuffleDataType,
           typename DsDataType,
           typename EDataType,
@@ -179,18 +181,19 @@ template <typename ALayout,
           typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CDEBlockTransferScalarPerVector_NPerBlock,
           LoopScheduler LoopSched = make_default_loop_scheduler()>
-struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
-                                                                   BLayout,
-                                                                   DELayout,
-                                                                   ADataType,
-                                                                   BDataType,
-                                                                   DsDataType,
-                                                                   EDataType,
-                                                                   AElementwiseOperation,
-                                                                   BElementwiseOperation,
-                                                                   CDEElementwiseOperation>
+struct DeviceBatchedGemmMultiD_Xdl : public DeviceBatchedGemmMultiD<ALayout,
+                                                                    BLayout,
+                                                                    DsLayout,
+                                                                    ELayout,
+                                                                    ADataType,
+                                                                    BDataType,
+                                                                    DsDataType,
+                                                                    EDataType,
+                                                                    AElementwiseOperation,
+                                                                    BElementwiseOperation,
+                                                                    CDEElementwiseOperation>
 {
-    using DeviceOp = DeviceBatchedGemmMultiDXdl;
+    using DeviceOp = DeviceBatchedGemmMultiD_Xdl;
 
     static constexpr index_t NumDTensor = DsDataType::Size();
 
@@ -199,7 +202,10 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
     static constexpr auto I2 = Number<2>{};
     static constexpr auto I3 = Number<3>{};
 
-    static auto MakeAGridDescriptor_AK0_M_AK1(index_t MRaw, index_t KRaw, index_t StrideA)
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
+
+    static auto MakeAGridDescriptor_M_K(index_t MRaw, index_t KRaw, index_t StrideA)
     {
         const auto a_grid_desc_mraw_kraw = [&]() {
             if constexpr(is_same_v<tensor_layout::gemm::RowMajor, ALayout>)
@@ -214,95 +220,10 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both M and K
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(M)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad M, but not K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_right_pad_transform(MRaw, MPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad K, but not M
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k = transform_tensor_descriptor(
-                a_grid_desc_mraw_kraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else
-        {
-            // not pad M or K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
+        return matrix_padder.PadADescriptor_M_K(a_grid_desc_mraw_kraw);
     }
 
-    static auto MakeBGridDescriptor_BK0_N_BK1(index_t KRaw, index_t NRaw, index_t StrideB)
+    static auto MakeBGridDescriptor_N_K(index_t KRaw, index_t NRaw, index_t StrideB)
     {
         const auto b_grid_desc_nraw_kraw = [&]() {
             if constexpr(is_same<tensor_layout::gemm::RowMajor, BLayout>::value)
@@ -317,155 +238,45 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
             }
         }();
 
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto NPad = N - NRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::NKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both N and K
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_right_pad_transform(NRaw, NPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(N)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad N, but not K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_right_pad_transform(NRaw, NPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad K, but not N
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k = transform_tensor_descriptor(
-                b_grid_desc_nraw_kraw,
-                make_tuple(make_pass_through_transform(NRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else
-        {
-            // not pad N or K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
+        return matrix_padder.PadBDescriptor_N_K(b_grid_desc_nraw_kraw);
     }
 
+    template <typename ELay>
     static auto MakeEGridDescriptor_M_N(index_t MRaw, index_t NRaw, index_t StrideE)
     {
-        const auto c_grid_desc_mraw_nraw = [&]() {
-            if constexpr(is_same<tensor_layout::gemm::RowMajor, DELayout>::value)
+        const auto e_grid_desc_mraw_nraw = [&]() {
+            if constexpr(is_same<tensor_layout::gemm::RowMajor, ELay>::value)
             {
                 return make_naive_tensor_descriptor(make_tuple(MRaw, NRaw),
                                                     make_tuple(StrideE, I1));
             }
-            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, DELayout>::value)
+            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, ELay>::value)
             {
                 return make_naive_tensor_descriptor(make_tuple(MRaw, NRaw),
                                                     make_tuple(I1, StrideE));
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto NPad = N - NRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad M and N
-            return transform_tensor_descriptor(c_grid_desc_mraw_nraw,
-                                               make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                          make_right_pad_transform(NRaw, NPad)),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad M, but not N
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(NRaw)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad N, but not M
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(NRaw, NPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else
-        {
-            // not pad M or N
-            return c_grid_desc_mraw_nraw;
-        }
+        return matrix_padder.PadCDescriptor_M_N(e_grid_desc_mraw_nraw);
     }
 
-    using AGridDesc_AK0_M_AK1 = decltype(MakeAGridDescriptor_AK0_M_AK1(1, 1, 1));
-    using BGridDesc_BK0_N_BK1 = decltype(MakeBGridDescriptor_BK0_N_BK1(1, 1, 1));
-    using EGridDesc_M_N       = decltype(MakeEGridDescriptor_M_N(1, 1, 1));
+    static auto MakeDsGridDescriptor_M_N(const std::array<index_t, NumDTensor>& MRaws,
+                                         const std::array<index_t, NumDTensor>& NRaws,
+                                         const std::array<index_t, NumDTensor>& DsStride)
+    {
+        return generate_tuple(
+            [&](auto i) {
+                using DLayout = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+
+                return DeviceOp::MakeEGridDescriptor_M_N<DLayout>(MRaws[i], NRaws[i], DsStride[i]);
+            },
+            Number<NumDTensor>{});
+    }
+
+    using AGridDesc_M_K  = decltype(MakeAGridDescriptor_M_K(1, 1, 1));
+    using BGridDesc_N_K  = decltype(MakeBGridDescriptor_N_K(1, 1, 1));
+    using DsGridDesc_M_N = remove_cvref_t<decltype(MakeDsGridDescriptor_M_N({}, {}, {}))>;
+    using EGridDesc_M_N  = decltype(MakeEGridDescriptor_M_N<ELayout>(1, 1, 1));
 
     struct ComputePtrOffsetOfStridedBatch
     {
@@ -511,9 +322,9 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         index_t BatchStrideE_;
     };
 
-    using GridwiseGemm = GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle<
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
         ADataType, // TODO: distinguish A/B datatype
-        GemmAccDataType,
+        AccDataType,
         CShuffleDataType,
         DsDataType,
         EDataType,
@@ -521,8 +332,9 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         BElementwiseOperation,
         CDEElementwiseOperation,
         InMemoryDataOperationEnum::Set,
-        AGridDesc_AK0_M_AK1,
-        BGridDesc_BK0_N_BK1,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        DsGridDesc_M_N,
         EGridDesc_M_N,
         NumGemmKPrefetchStage,
         BlockSize,
@@ -557,9 +369,12 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         CDEBlockTransferScalarPerVector_NPerBlock,
         LoopSched>;
 
-    using CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock = decltype(
-        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(EGridDesc_M_N{}));
-    using Block2CTileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
+    using Block2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
 
     // Argument
     struct Argument : public BaseArgument
@@ -568,89 +383,112 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
                  const void* p_b_grid,
                  std::array<const void*, NumDTensor> p_ds_grid,
                  void* p_e_grid,
-                 index_t M,
-                 index_t N,
-                 index_t K,
+                 index_t MRaw,
+                 index_t NRaw,
+                 index_t KRaw,
+                 index_t Batch,
                  index_t StrideA,
                  index_t StrideB,
-                 std::array<ck::index_t, NumDTensor> StrideDs,
+                 const std::array<ck::index_t, NumDTensor>& StrideDs,
                  index_t StrideE,
                  index_t BatchStrideA,
                  index_t BatchStrideB,
-                 std::array<ck::index_t, NumDTensor> BatchStrideDs,
+                 const std::array<ck::index_t, NumDTensor>& BatchStrideDs,
                  index_t BatchStrideE,
-                 index_t Batch,
-                 index_t M01,
-                 index_t N01,
                  AElementwiseOperation a_element_op,
                  BElementwiseOperation b_element_op,
                  CDEElementwiseOperation cde_element_op)
             : p_a_grid_{static_cast<const ADataType*>(p_a_grid)},
               p_b_grid_{static_cast<const BDataType*>(p_b_grid)},
-              p_ds_grid_{}, // FIXME
+              p_ds_grid_{},
               p_e_grid_{static_cast<EDataType*>(p_e_grid)},
               Batch_(Batch),
+              a_grid_desc_m_k_{DeviceOp::MakeAGridDescriptor_M_K(MRaw, KRaw, StrideA)},
+              b_grid_desc_n_k_{DeviceOp::MakeBGridDescriptor_N_K(KRaw, NRaw, StrideB)},
+              ds_grid_desc_m_n_{},
+              e_grid_desc_m_n_{DeviceOp::MakeEGridDescriptor_M_N<ELayout>(MRaw, NRaw, StrideE)},
               a_grid_desc_ak0_m_ak1_{
-                  DeviceBatchedGemmMultiDXdl::MakeAGridDescriptor_AK0_M_AK1(M, K, StrideA)},
+                  GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k_)},
               b_grid_desc_bk0_n_bk1_{
-                  DeviceBatchedGemmMultiDXdl::MakeBGridDescriptor_BK0_N_BK1(K, N, StrideB)},
+                  GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k_)},
               ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
-              e_grid_desc_m_n_{DeviceBatchedGemmMultiDXdl::MakeEGridDescriptor_M_N(M, N, StrideE)},
               e_grid_desc_mblock_mperblock_nblock_nperblock_{},
               compute_ptr_offset_of_batch_{BatchStrideA, BatchStrideB, BatchStrideDs, BatchStrideE},
-              block_2_ctile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
-              M01_{M01},
-              N01_{N01},
+              block_2_etile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
               a_element_op_{a_element_op},
               b_element_op_{b_element_op},
               cde_element_op_{cde_element_op}
         {
-            if(GridwiseGemm::CheckValidity(a_grid_desc_ak0_m_ak1_,
-                                           b_grid_desc_bk0_n_bk1_,
+            // populate pointer, desc for Ds
+            static_for<0, NumDTensor, 1>{}([&](auto i) {
+                using DLayout   = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+                using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
+
+                // D pointer
+                p_ds_grid_(i) = static_cast<const DDataType*>(p_ds_grid[i]);
+
+                // D desc
+                ds_grid_desc_m_n_(i) =
+                    DeviceOp::MakeEGridDescriptor_M_N<DLayout>(MRaw, NRaw, StrideDs[i]);
+            });
+
+            // populate desc for Ds/E
+            if(GridwiseGemm::CheckValidity(a_grid_desc_m_k_,
+                                           b_grid_desc_n_k_,
+                                           ds_grid_desc_m_n_,
                                            e_grid_desc_m_n_,
-                                           block_2_ctile_map_))
+                                           block_2_etile_map_))
             {
+                ds_grid_desc_mblock_mperblock_nblock_nperblock_ =
+                    GridwiseGemm::MakeDsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                        ds_grid_desc_m_n_);
+
                 e_grid_desc_mblock_mperblock_nblock_nperblock_ =
                     GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
                         e_grid_desc_m_n_);
-
-                static_for<0, NumDTensor, 1>{}([&](auto i) {
-                    using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
-
-                    p_ds_grid_(i) = static_cast<const DDataType*>(p_ds_grid[i]);
-
-                    const auto d_grid_desc_m_n =
-                        DeviceOp::MakeEGridDescriptor_M_N(M, N, StrideDs[i]);
-
-                    ds_grid_desc_mblock_mperblock_nblock_nperblock_(i) =
-                        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
-                            d_grid_desc_m_n);
-                });
             }
         }
 
+        void Print() const
+        {
+            std::cout << "A[M, K]: " << a_grid_desc_m_k_ << std::endl;
+            std::cout << "B[N, K]: " << b_grid_desc_n_k_ << std::endl;
+            static_for<0, NumDTensor, 1>{}(
+                [&](auto i) { std::cout << "Ds[M, N]: " << ds_grid_desc_m_n_[i] << std::endl; });
+            std::cout << "E[M, N]: " << e_grid_desc_m_n_ << std::endl;
+        }
+
         //  private:
+        // pointers
         const ADataType* p_a_grid_;
         const BDataType* p_b_grid_;
         typename GridwiseGemm::DsGridPointer p_ds_grid_;
         EDataType* p_e_grid_;
+
+        // Batch
         index_t Batch_;
 
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        DsGridDesc_M_N ds_grid_desc_m_n_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
         AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
         BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
-        StaticallyIndexedArray<
-            typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-            NumDTensor>
-            ds_grid_desc_mblock_mperblock_nblock_nperblock_; // FIXME: Ds desc may be of different
-                                                             // type from E
-        EGridDesc_M_N e_grid_desc_m_n_;
+        typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            ds_grid_desc_mblock_mperblock_nblock_nperblock_;
         typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
             e_grid_desc_mblock_mperblock_nblock_nperblock_;
 
+        // for calculating batch offset
         ComputePtrOffsetOfStridedBatch compute_ptr_offset_of_batch_;
-        Block2CTileMap block_2_ctile_map_;
-        index_t M01_;
-        index_t N01_;
+
+        // block-to-e-tile map
+        Block2ETileMap block_2_etile_map_;
+
+        // element-wise op
         AElementwiseOperation a_element_op_;
         BElementwiseOperation b_element_op_;
         CDEElementwiseOperation cde_element_op_;
@@ -659,36 +497,21 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
     // Invoker
     struct Invoker : public BaseInvoker
     {
-        using Argument = DeviceBatchedGemmMultiDXdl::Argument;
+        using Argument = DeviceBatchedGemmMultiD_Xdl::Argument;
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            {
-                std::cout << "arg.a_grid_desc_ak0_m_ak1_{"
-                          << arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) << ", "
-                          << arg.a_grid_desc_ak0_m_ak1_.GetLength(I1) << ", "
-                          << arg.a_grid_desc_ak0_m_ak1_.GetLength(I2) << "}" << std::endl;
-
-                std::cout << "arg.b_grid_desc_bk0_n_bk1_{"
-                          << arg.b_grid_desc_bk0_n_bk1_.GetLength(I0) << ", "
-                          << arg.b_grid_desc_bk0_n_bk1_.GetLength(I1) << ", "
-                          << arg.b_grid_desc_bk0_n_bk1_.GetLength(I2) << "}" << std::endl;
-
-                std::cout << "arg.e_grid_desc_m_n_{" << arg.e_grid_desc_m_n_.GetLength(I0) << ", "
-                          << arg.e_grid_desc_m_n_.GetLength(I1) << "}" << std::endl;
-            }
-
-            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                            arg.b_grid_desc_bk0_n_bk1_,
+            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                            arg.b_grid_desc_n_k_,
+                                            arg.ds_grid_desc_m_n_,
                                             arg.e_grid_desc_m_n_,
-                                            arg.block_2_ctile_map_))
+                                            arg.block_2_etile_map_))
             {
-                throw std::runtime_error(
-                    "wrong! GridwiseBatchedGemm_km_kn_m0m1n0n1_xdlops_v2r3 has invalid setting");
+                throw std::runtime_error("wrong! GridwiseGemm has invalid setting");
             }
 
             const index_t grid_size =
-                arg.block_2_ctile_map_.CalculateGridSize(arg.e_grid_desc_m_n_) * arg.Batch_;
+                arg.block_2_etile_map_.CalculateGridSize(arg.e_grid_desc_m_n_) * arg.Batch_;
 
             const auto K =
                 arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) * arg.a_grid_desc_ak0_m_ak1_.GetLength(I2);
@@ -701,17 +524,15 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
                     ADataType, // TODO: distiguish A/B datatype
                     typename GridwiseGemm::DsGridPointer,
                     EDataType,
-                    DeviceOp::AGridDesc_AK0_M_AK1,
-                    DeviceOp::BGridDesc_BK0_N_BK1,
-                    ck::StaticallyIndexedArray<
-                        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-                        NumDTensor>,
-                    typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     AElementwiseOperation,
                     BElementwiseOperation,
                     CDEElementwiseOperation,
+                    DeviceOp::AGridDesc_AK0_M_AK1,
+                    DeviceOp::BGridDesc_BK0_N_BK1,
+                    typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
+                    typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     ComputePtrOffsetOfStridedBatch,
-                    remove_reference_t<Block2CTileMap>,
+                    Block2ETileMap,
                     has_main_loop>;
 
                 return launch_and_time_kernel(stream_config,
@@ -724,29 +545,25 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
                                               arg.p_ds_grid_,
                                               arg.p_e_grid_,
                                               arg.Batch_,
+                                              arg.a_element_op_,
+                                              arg.b_element_op_,
+                                              arg.cde_element_op_,
                                               arg.a_grid_desc_ak0_m_ak1_,
                                               arg.b_grid_desc_bk0_n_bk1_,
                                               arg.ds_grid_desc_mblock_mperblock_nblock_nperblock_,
                                               arg.e_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                              arg.a_element_op_,
-                                              arg.b_element_op_,
-                                              arg.cde_element_op_,
                                               arg.compute_ptr_offset_of_batch_,
-                                              arg.block_2_ctile_map_);
+                                              arg.block_2_etile_map_);
             };
 
-            float ave_time = 0;
-
             if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
             {
-                ave_time = launch_kernel(integral_constant<bool, true>{});
+                return launch_kernel(integral_constant<bool, true>{});
             }
             else
             {
-                ave_time = launch_kernel(integral_constant<bool, false>{});
+                return launch_kernel(integral_constant<bool, false>{});
             }
-
-            return ave_time;
         }
 
         // polymorphic
@@ -757,18 +574,18 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         }
     };
 
-    static constexpr bool IsValidCompilationParameter()
-    {
-        // TODO: properly implement this check
-        return true;
-    }
-
     static bool IsSupportedArgument(const Argument& arg)
     {
-        return GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                           arg.b_grid_desc_bk0_n_bk1_,
+        if(!(ck::get_device_name() == "gfx908" || ck::get_device_name() == "gfx90a"))
+        {
+            return false;
+        }
+
+        return GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                           arg.b_grid_desc_n_k_,
+                                           arg.ds_grid_desc_m_n_,
                                            arg.e_grid_desc_m_n_,
-                                           arg.block_2_ctile_map_);
+                                           arg.block_2_etile_map_);
     }
 
     // polymorphic
@@ -779,20 +596,20 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
 
     static auto MakeArgument(const void* p_a,
                              const void* p_b,
-                             std::array<const void*, NumDTensor> p_ds,
-                             void* p_c,
+                             const std::array<const void*, NumDTensor>& p_ds,
+                             void* p_e,
                              index_t M,
                              index_t N,
                              index_t K,
+                             index_t Batch,
                              index_t StrideA,
                              index_t StrideB,
-                             std::array<index_t, NumDTensor> StrideDs,
+                             const std::array<index_t, NumDTensor>& StrideDs,
                              index_t StrideE,
                              index_t BatchStrideA,
                              index_t BatchStrideB,
-                             std::array<ck::index_t, NumDTensor> BatchStrideDs,
+                             const std::array<ck::index_t, NumDTensor>& BatchStrideDs,
                              index_t BatchStrideE,
-                             index_t Batch,
                              AElementwiseOperation a_element_op,
                              BElementwiseOperation b_element_op,
                              CDEElementwiseOperation cde_element_op)
@@ -800,10 +617,11 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         return Argument{p_a,
                         p_b,
                         p_ds,
-                        p_c,
+                        p_e,
                         M,
                         N,
                         K,
+                        Batch,
                         StrideA,
                         StrideB,
                         StrideDs,
@@ -812,9 +630,6 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
                         BatchStrideB,
                         BatchStrideDs,
                         BatchStrideE,
-                        Batch,
-                        1,
-                        1,
                         a_element_op,
                         b_element_op,
                         cde_element_op};
@@ -826,20 +641,20 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
     std::unique_ptr<BaseArgument>
     MakeArgumentPointer(const void* p_a,
                         const void* p_b,
-                        std::array<const void*, NumDTensor> p_ds,
-                        void* p_c,
+                        const std::array<const void*, NumDTensor>& p_ds,
+                        void* p_e,
                         index_t M,
                         index_t N,
                         index_t K,
+                        index_t Batch,
                         index_t StrideA,
                         index_t StrideB,
-                        std::array<ck::index_t, NumDTensor> StrideDs,
+                        const std::array<ck::index_t, NumDTensor>& StrideDs,
                         index_t StrideE,
                         index_t BatchStrideA,
                         index_t BatchStrideB,
-                        std::array<ck::index_t, NumDTensor> BatchStrideDs,
+                        const std::array<ck::index_t, NumDTensor>& BatchStrideDs,
                         index_t BatchStrideE,
-                        index_t Batch,
                         AElementwiseOperation a_element_op,
                         BElementwiseOperation b_element_op,
                         CDEElementwiseOperation cde_element_op) override
@@ -847,10 +662,11 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         return std::make_unique<Argument>(p_a,
                                           p_b,
                                           p_ds,
-                                          p_c,
+                                          p_e,
                                           M,
                                           N,
                                           K,
+                                          Batch,
                                           StrideA,
                                           StrideB,
                                           StrideDs,
@@ -859,9 +675,6 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
                                           BatchStrideB,
                                           BatchStrideDs,
                                           BatchStrideE,
-                                          Batch,
-                                          1,
-                                          1,
                                           a_element_op,
                                           b_element_op,
                                           cde_element_op);
@@ -879,7 +692,7 @@ struct DeviceBatchedGemmMultiDXdl : public DeviceBatchedGemmMultiD<ALayout,
         auto str = std::stringstream();
 
         // clang-format off
-        str << "DeviceBatchedGemmMultiDXdl"
+        str << "DeviceBatchedGemmMultiD_Xdl"
             << "<"
             << BlockSize << ", "
             << MPerBlock << ", "
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_reduce_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_reduce_xdl_cshuffle.hpp
index 1486f0ac73..3c5fdbdab0 100644
--- a/include/ck/tensor_operation/gpu/device/device_batched_gemm_reduce_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_reduce_xdl_cshuffle.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_reduce.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_reduce_xdl_cshuffle_v1.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_batched_gemm_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_batched_gemm_xdl.hpp
index ee94290a9d..072588cae7 100644
--- a/include/ck/tensor_operation/gpu/device/device_batched_gemm_xdl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_batched_gemm_xdl.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_batched_gemm.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_binary_elementwise.hpp b/include/ck/tensor_operation/gpu/device/device_binary_elementwise.hpp
index 99be946e92..ef2ab325a7 100644
--- a/include/ck/tensor_operation/gpu/device/device_binary_elementwise.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_binary_elementwise.hpp
@@ -6,8 +6,8 @@
 #include <iostream>
 #include <vector>
 
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 #include "ck/tensor_operation/gpu/device/device_elementwise.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_binary_elementwise_1d.hpp"
diff --git a/include/ck/tensor_operation/gpu/device/device_cgemm_4gemm_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_cgemm_4gemm_xdl_cshuffle.hpp
index ac6b23479c..4277499f99 100644
--- a/include/ck/tensor_operation/gpu/device/device_cgemm_4gemm_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_cgemm_4gemm_xdl_cshuffle.hpp
@@ -16,8 +16,8 @@
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v1.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_binary_elementwise_1d.hpp"
 #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d.hpp b/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d.hpp
index fa0f07d379..dbc525c099 100644
--- a/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d.hpp
@@ -43,14 +43,14 @@ struct DeviceContractionMultipleD : public BaseOperator
                         const void* p_b,
                         std::array<const void*, NumDTensor> p_ds,
                         void* p_e,
-                        std::vector<index_t> a_ms_ks_lengths,
-                        std::vector<index_t> a_ms_ks_strides,
-                        std::vector<index_t> b_ns_ks_lengths,
-                        std::vector<index_t> b_ns_ks_strides,
-                        std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_lengths,
-                        std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_strides,
-                        std::vector<index_t> e_ms_ns_lengths,
-                        std::vector<index_t> e_ms_ns_strides,
+                        const std::vector<index_t>& a_ms_ns_lengths,
+                        const std::vector<index_t>& a_ms_ks_strides,
+                        const std::vector<index_t>& b_ns_ks_lengths,
+                        const std::vector<index_t>& b_ns_ks_strides,
+                        const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_lengths,
+                        const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_strides,
+                        const std::vector<index_t>& e_ms_ns_lengths,
+                        const std::vector<index_t>& e_ms_ns_strides,
                         AElementwiseOperation a_element_op,
                         BElementwiseOperation b_element_op,
                         CDEElementwiseOperation cde_element_op) = 0;
diff --git a/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d_xdl_cshuffle.hpp
index b130290fbe..b1c2545ff0 100644
--- a/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_contraction_multiple_d_xdl_cshuffle.hpp
@@ -12,9 +12,10 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/device_contraction_multiple_d.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 
@@ -106,7 +107,7 @@ template <index_t NumDimM,
           index_t NumDimK,
           typename ADataType,
           typename BDataType,
-          typename GemmAccDataType,
+          typename AccDataType,
           typename CShuffleDataType,
           typename DsDataType,
           typename EDataType,
@@ -165,9 +166,12 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
     static constexpr auto I2 = Number<2>{};
     static constexpr auto I3 = Number<3>{};
 
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
+
     // Assume: A[M0, M1, M2, ..., K0, K1, K2, ...]
-    static auto MakeAGridDescriptor_AK0_M_AK1(const std::vector<index_t>& a_ms_ks_lengths_vec,
-                                              const std::vector<index_t>& a_ms_ks_strides_vec)
+    static auto MakeAGridDescriptor_M_K(const std::vector<index_t>& a_ms_ks_lengths_vec,
+                                        const std::vector<index_t>& a_ms_ks_strides_vec)
     {
         assert(a_ms_ks_lengths_vec.size() == NumDimM + NumDimK &&
                a_ms_ks_strides_vec.size() == NumDimM + NumDimK);
@@ -203,100 +207,12 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
             make_tuple(mDimIds, kDimIds),
             make_tuple(Sequence<0>{}, Sequence<1>{}));
 
-        const auto MRaw = a_grid_desc_mraw_kraw.GetLength(I0);
-        const auto KRaw = a_grid_desc_mraw_kraw.GetLength(I1);
-
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both M and K
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(M)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad M, but not K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_right_pad_transform(MRaw, MPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad K, but not M
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k = transform_tensor_descriptor(
-                a_grid_desc_mraw_kraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else
-        {
-            // not pad M or K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
+        return matrix_padder.PadADescriptor_M_K(a_grid_desc_mraw_kraw);
     }
 
     // Assume: B[N0, N1, N2, ..., K0, K1, K2, ...]
-    static auto MakeBGridDescriptor_BK0_N_BK1(const std::vector<index_t>& b_ns_ks_lengths_vec,
-                                              const std::vector<index_t>& b_ns_ks_strides_vec)
+    static auto MakeBGridDescriptor_N_K(const std::vector<index_t>& b_ns_ks_lengths_vec,
+                                        const std::vector<index_t>& b_ns_ks_strides_vec)
     {
         assert(b_ns_ks_lengths_vec.size() == NumDimN + NumDimK &&
                b_ns_ks_strides_vec.size() == NumDimN + NumDimK);
@@ -332,95 +248,7 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
             make_tuple(nDimIds, kDimIds),
             make_tuple(Sequence<0>{}, Sequence<1>{}));
 
-        const auto NRaw = b_grid_desc_nraw_kraw.GetLength(I0);
-        const auto KRaw = b_grid_desc_nraw_kraw.GetLength(I1);
-
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto NPad = N - NRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::NKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both N and K
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_right_pad_transform(NRaw, NPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(N)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad N, but not K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_right_pad_transform(NRaw, NPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad K, but not N
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k = transform_tensor_descriptor(
-                b_grid_desc_nraw_kraw,
-                make_tuple(make_pass_through_transform(NRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else
-        {
-            // not pad N or K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
+        return matrix_padder.PadBDescriptor_N_K(b_grid_desc_nraw_kraw);
     }
 
     // assume E[M0, M1, M2, ..., N0, N1, N2...]
@@ -461,63 +289,30 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
             make_tuple(mDimIds, nDimIds),
             make_tuple(Sequence<0>{}, Sequence<1>{}));
 
-        const auto MRaw = e_grid_desc_mraw_nraw.GetLength(I0);
-        const auto NRaw = e_grid_desc_mraw_nraw.GetLength(I1);
-
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto NPad = N - NRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad M and N
-            return transform_tensor_descriptor(e_grid_desc_mraw_nraw,
-                                               make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                          make_right_pad_transform(NRaw, NPad)),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad M, but not N
-            return transform_tensor_descriptor(
-                e_grid_desc_mraw_nraw,
-                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(NRaw)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad N, but not M
-            return transform_tensor_descriptor(
-                e_grid_desc_mraw_nraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(NRaw, NPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else
-        {
-            // not pad M or N
-            return e_grid_desc_mraw_nraw;
-        }
+        return matrix_padder.PadCDescriptor_M_N(e_grid_desc_mraw_nraw);
     }
 
-    using AGridDesc_AK0_M_AK1 =
-        decltype(MakeAGridDescriptor_AK0_M_AK1(std::vector<index_t>{}, std::vector<index_t>{}));
-    using BGridDesc_BK0_N_BK1 =
-        decltype(MakeBGridDescriptor_BK0_N_BK1(std::vector<index_t>{}, std::vector<index_t>{}));
-    using EGridDesc_M_N =
-        decltype(MakeEGridDescriptor_M_N(std::vector<index_t>{}, std::vector<index_t>{}));
+    static auto MakeDsGridDescriptor_M_N(
+        const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_lengths_vec,
+        const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_strides_vec)
+    {
+        return generate_tuple(
+            [&](auto i) {
+                return DeviceOp::MakeEGridDescriptor_M_N(ds_ms_ns_lengths_vec[i],
+                                                         ds_ms_ns_strides_vec[i]);
+            },
+            Number<NumDTensor>{});
+    }
+
+    using AGridDesc_M_K  = decltype(MakeAGridDescriptor_M_K({}, {}));
+    using BGridDesc_N_K  = decltype(MakeBGridDescriptor_N_K({}, {}));
+    using DsGridDesc_M_N = remove_cvref_t<decltype(MakeDsGridDescriptor_M_N({{}}, {{}}))>;
+    using EGridDesc_M_N  = decltype(MakeEGridDescriptor_M_N({}, {}));
 
     // GridwiseGemm
-    using GridwiseGemm = GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle<
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
         ADataType, // TODO: distinguish A/B datatype
-        GemmAccDataType,
+        AccDataType,
         CShuffleDataType,
         DsDataType,
         EDataType,
@@ -525,8 +320,9 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
         BElementwiseOperation,
         CDEElementwiseOperation,
         InMemoryDataOperationEnum::Set,
-        AGridDesc_AK0_M_AK1,
-        BGridDesc_BK0_N_BK1,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        DsGridDesc_M_N,
         EGridDesc_M_N,
         NumGemmKPrefetchStage,
         BlockSize,
@@ -561,6 +357,13 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
         CDEBlockTransferScalarPerVector_NPerBlock,
         LoopSched>;
 
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
+    using Block2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+
     // Argument
     struct Argument : public BaseArgument
     {
@@ -568,27 +371,30 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
                  const void* p_b_grid,
                  std::array<const void*, NumDTensor> p_ds_grid,
                  void* p_e_grid,
-                 std::vector<index_t> a_ms_ns_lengths,
-                 std::vector<index_t> a_ms_ks_strides,
-                 std::vector<index_t> b_ns_ks_lengths,
-                 std::vector<index_t> b_ns_ks_strides,
-                 std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_lengths,
-                 std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_strides,
-                 std::vector<index_t> e_ms_ns_lengths,
-                 std::vector<index_t> e_ms_ns_strides,
+                 const std::vector<index_t>& a_ms_ns_lengths,
+                 const std::vector<index_t>& a_ms_ks_strides,
+                 const std::vector<index_t>& b_ns_ks_lengths,
+                 const std::vector<index_t>& b_ns_ks_strides,
+                 const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_lengths,
+                 const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_strides,
+                 const std::vector<index_t>& e_ms_ns_lengths,
+                 const std::vector<index_t>& e_ms_ns_strides,
                  AElementwiseOperation a_element_op,
                  BElementwiseOperation b_element_op,
                  CDEElementwiseOperation cde_element_op)
             : p_a_grid_{static_cast<const ADataType*>(p_a_grid)},
               p_b_grid_{static_cast<const BDataType*>(p_b_grid)},
-              p_ds_grid_{}, // FIXME
+              p_ds_grid_{},
               p_e_grid_{static_cast<EDataType*>(p_e_grid)},
-              a_grid_desc_ak0_m_ak1_{
-                  DeviceOp::MakeAGridDescriptor_AK0_M_AK1(a_ms_ns_lengths, a_ms_ks_strides)},
-              b_grid_desc_bk0_n_bk1_{
-                  DeviceOp::MakeBGridDescriptor_BK0_N_BK1(b_ns_ks_lengths, b_ns_ks_strides)},
-              ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
+              a_grid_desc_m_k_{DeviceOp::MakeAGridDescriptor_M_K(a_ms_ns_lengths, a_ms_ks_strides)},
+              b_grid_desc_n_k_{DeviceOp::MakeBGridDescriptor_N_K(b_ns_ks_lengths, b_ns_ks_strides)},
+              ds_grid_desc_m_n_{},
               e_grid_desc_m_n_{DeviceOp::MakeEGridDescriptor_M_N(e_ms_ns_lengths, e_ms_ns_strides)},
+              a_grid_desc_ak0_m_ak1_{
+                  GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k_)},
+              b_grid_desc_bk0_n_bk1_{
+                  GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k_)},
+              ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
               e_grid_desc_mblock_mperblock_nblock_nperblock_{},
               block_2_etile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
               a_element_op_{a_element_op},
@@ -601,8 +407,22 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
               ds_nz_stride_{},
               e_nz_stride_{}
         {
-            if(GridwiseGemm::CheckValidity(a_grid_desc_ak0_m_ak1_,
-                                           b_grid_desc_bk0_n_bk1_,
+            // populate pointer, batch stride, desc for Ds
+            static_for<0, NumDTensor, 1>{}([&](auto i) {
+                using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
+
+                // D pointer
+                p_ds_grid_(i) = static_cast<const DDataType*>(p_ds_grid[i]);
+
+                // D desc
+                ds_grid_desc_m_n_(i) =
+                    DeviceOp::MakeEGridDescriptor_M_N(ds_ms_ns_lengths[i], ds_ms_ns_strides[i]);
+            });
+
+            // populate desc for Ds/E
+            if(GridwiseGemm::CheckValidity(a_grid_desc_m_k_,
+                                           b_grid_desc_n_k_,
+                                           ds_grid_desc_m_n_,
                                            e_grid_desc_m_n_,
                                            block_2_etile_map_))
             {
@@ -610,18 +430,9 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
                     GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
                         e_grid_desc_m_n_);
 
-                static_for<0, NumDTensor, 1>{}([&](auto i) {
-                    using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
-
-                    p_ds_grid_(i) = static_cast<const DDataType*>(p_ds_grid[i]);
-
-                    const auto d_grid_desc_m_n =
-                        DeviceOp::MakeEGridDescriptor_M_N(ds_ms_ns_lengths[i], ds_ms_ns_strides[i]);
-
-                    ds_grid_desc_mblock_mperblock_nblock_nperblock_(i) =
-                        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
-                            d_grid_desc_m_n);
-                });
+                ds_grid_desc_mblock_mperblock_nblock_nperblock_ =
+                    GridwiseGemm::MakeDsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                        ds_grid_desc_m_n_);
             }
 
             // for sanity check of vector memory access
@@ -639,6 +450,15 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
             e_nz_stride_ = e_ms_ns_strides[NumDimM + NumDimN - 1];
         }
 
+        void Print() const
+        {
+            std::cout << "A[M, K]: " << a_grid_desc_m_k_ << std::endl;
+            std::cout << "B[N, K]: " << b_grid_desc_n_k_ << std::endl;
+            static_for<0, NumDTensor, 1>{}(
+                [&](auto i) { std::cout << "Ds[M, N]: " << ds_grid_desc_m_n_[i] << std::endl; });
+            std::cout << "E[M, N]: " << e_grid_desc_m_n_ << std::endl;
+        }
+
         //  private:
         // pointers
         const ADataType* p_a_grid_;
@@ -646,20 +466,22 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
         typename GridwiseGemm::DsGridPointer p_ds_grid_;
         EDataType* p_e_grid_;
 
-        // tensor descriptors
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        DsGridDesc_M_N ds_grid_desc_m_n_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
         AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
         BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
-        StaticallyIndexedArray<
-            typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-            NumDTensor>
-            ds_grid_desc_mblock_mperblock_nblock_nperblock_; // FIXME: Ds desc may be of different
-                                                             // type from E
-        EGridDesc_M_N e_grid_desc_m_n_;
+        typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            ds_grid_desc_mblock_mperblock_nblock_nperblock_;
         typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
             e_grid_desc_mblock_mperblock_nblock_nperblock_;
 
         // block-to-e-tile map
-        typename GridwiseGemm::DefaultBlock2ETileMap block_2_etile_map_;
+        Block2ETileMap block_2_etile_map_;
 
         // element-wise op
         AElementwiseOperation a_element_op_;
@@ -684,29 +506,14 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-#if 0
-            {
-                std::cout << "arg.a_grid_desc_ak0_m_ak1_{"
-                          << arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) << ", "
-                          << arg.a_grid_desc_ak0_m_ak1_.GetLength(I1) << ", "
-                          << arg.a_grid_desc_ak0_m_ak1_.GetLength(I2) << "}" << std::endl;
-
-                std::cout << "arg.b_grid_desc_bk0_n_bk1_{"
-                          << arg.b_grid_desc_bk0_n_bk1_.GetLength(I0) << ", "
-                          << arg.b_grid_desc_bk0_n_bk1_.GetLength(I1) << ", "
-                          << arg.b_grid_desc_bk0_n_bk1_.GetLength(I2) << "}" << std::endl;
-
-                std::cout << "arg.e_grid_desc_m_n_{ " << arg.e_grid_desc_m_n_.GetLength(I0) << ", "
-                          << arg.e_grid_desc_m_n_.GetLength(I1) << "}" << std::endl;
-            }
-#endif
-
-            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                            arg.b_grid_desc_bk0_n_bk1_,
+            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                            arg.b_grid_desc_n_k_,
+                                            arg.ds_grid_desc_m_n_,
                                             arg.e_grid_desc_m_n_,
                                             arg.block_2_etile_map_))
             {
-                throw std::runtime_error("wrong! GridwiseGemm has invalid setting");
+                throw std::runtime_error(
+                    "wrong! GridwiseGemmMultipleD_xdl_cshuffle has invalid setting");
             }
 
             const index_t grid_size =
@@ -728,9 +535,7 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
                     CDEElementwiseOperation,
                     DeviceOp::AGridDesc_AK0_M_AK1,
                     DeviceOp::BGridDesc_BK0_N_BK1,
-                    ck::StaticallyIndexedArray<
-                        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-                        NumDTensor>,
+                    typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     typename GridwiseGemm::DefaultBlock2ETileMap,
                     has_main_loop>;
@@ -754,18 +559,14 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
                                               arg.block_2_etile_map_);
             };
 
-            float ave_time = 0;
-
             if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
             {
-                ave_time = launch_kernel(integral_constant<bool, true>{});
+                return launch_kernel(integral_constant<bool, true>{});
             }
             else
             {
-                ave_time = launch_kernel(integral_constant<bool, false>{});
+                return launch_kernel(integral_constant<bool, false>{});
             }
-
-            return ave_time;
         }
 
         // polymorphic
@@ -776,12 +577,6 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
         }
     };
 
-    static constexpr bool IsValidCompilationParameter()
-    {
-        // TODO: properly implement this check
-        return true;
-    }
-
     static bool IsSupportedArgument(const Argument& arg)
     {
         if(!(ck::get_device_name() == "gfx908" || ck::get_device_name() == "gfx90a"))
@@ -789,8 +584,9 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
             return false;
         }
 
-        if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                        arg.b_grid_desc_bk0_n_bk1_,
+        if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                        arg.b_grid_desc_n_k_,
+                                        arg.ds_grid_desc_m_n_,
                                         arg.e_grid_desc_m_n_,
                                         arg.block_2_etile_map_))
         {
@@ -878,14 +674,14 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
                              const void* p_b,
                              std::array<const void*, NumDTensor> p_ds,
                              void* p_e,
-                             std::vector<index_t> a_ms_ns_lengths,
-                             std::vector<index_t> a_ms_ks_strides,
-                             std::vector<index_t> b_ns_ks_lengths,
-                             std::vector<index_t> b_ns_ks_strides,
-                             std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_lengths,
-                             std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_strides,
-                             std::vector<index_t> e_ms_ns_lengths,
-                             std::vector<index_t> e_ms_ns_strides,
+                             const std::vector<index_t>& a_ms_ns_lengths,
+                             const std::vector<index_t>& a_ms_ks_strides,
+                             const std::vector<index_t>& b_ns_ks_lengths,
+                             const std::vector<index_t>& b_ns_ks_strides,
+                             const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_lengths,
+                             const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_strides,
+                             const std::vector<index_t>& e_ms_ns_lengths,
+                             const std::vector<index_t>& e_ms_ns_strides,
                              AElementwiseOperation a_element_op,
                              BElementwiseOperation b_element_op,
                              CDEElementwiseOperation cde_element_op)
@@ -915,14 +711,14 @@ struct DeviceContractionMultipleD_Xdl_CShuffle
                         const void* p_b,
                         std::array<const void*, NumDTensor> p_ds,
                         void* p_e,
-                        std::vector<index_t> a_ms_ns_lengths,
-                        std::vector<index_t> a_ms_ks_strides,
-                        std::vector<index_t> b_ns_ks_lengths,
-                        std::vector<index_t> b_ns_ks_strides,
-                        std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_lengths,
-                        std::array<std::vector<index_t>, NumDTensor> ds_ms_ns_strides,
-                        std::vector<index_t> e_ms_ns_lengths,
-                        std::vector<index_t> e_ms_ns_strides,
+                        const std::vector<index_t>& a_ms_ns_lengths,
+                        const std::vector<index_t>& a_ms_ks_strides,
+                        const std::vector<index_t>& b_ns_ks_lengths,
+                        const std::vector<index_t>& b_ns_ks_strides,
+                        const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_lengths,
+                        const std::array<std::vector<index_t>, NumDTensor>& ds_ms_ns_strides,
+                        const std::vector<index_t>& e_ms_ns_lengths,
+                        const std::vector<index_t>& e_ms_ns_strides,
                         AElementwiseOperation a_element_op,
                         BElementwiseOperation b_element_op,
                         CDEElementwiseOperation cde_element_op) override
diff --git a/include/ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
index 31b2ca05e6..9e860f6c40 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
@@ -10,12 +10,12 @@
 #include "ck/tensor_description/tensor_descriptor.hpp"
 #include "ck/tensor_description/tensor_descriptor_helper.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv_backward_weight.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_backward_weight_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_bwd_weight.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_unary_elementwise_1d.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -57,7 +57,14 @@ template <typename InDataType,
           typename CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CBlockTransferScalarPerVector_NWaveNPerXdl>
 struct DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-    : public DeviceConvBwdWeight<InElementwiseOperation,
+    : public DeviceConvBwdWeight<2,
+                                 ck::tensor_layout::convolution::NHWC,
+                                 ck::tensor_layout::convolution::KYXC,
+                                 ck::tensor_layout::convolution::NHWK,
+                                 InDataType,
+                                 WeiDataType,
+                                 OutDataType,
+                                 InElementwiseOperation,
                                  WeiElementwiseOperation,
                                  OutElementwiseOperation>
 {
diff --git a/include/ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp
index 37ef8db332..fa7c4fb3f4 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -55,7 +55,14 @@ template <typename InDataType,
           ck::index_t CThreadTransferSrcDstVectorDim,
           ck::index_t CThreadTransferDstScalarPerVector>
 struct DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-    : public DeviceConvBwdData<InElementwiseOperation,
+    : public DeviceConvBwdData<2,
+                               ck::tensor_layout::convolution::NHWC,
+                               ck::tensor_layout::convolution::KYXC,
+                               ck::tensor_layout::convolution::NHWK,
+                               InDataType,
+                               WeiDataType,
+                               OutDataType,
+                               InElementwiseOperation,
                                WeiElementwiseOperation,
                                OutElementwiseOperation>
 {
diff --git a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp
index 5b880b1fd6..4749665c4f 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_conv_fwd_bias_activation_add.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v3r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp
index bab9898785..bafbfe4d70 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp
@@ -14,8 +14,8 @@
 #include "ck/tensor_operation/gpu/device/device_conv_fwd_bias_activation.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v3r2.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
index 0e7d9cd4a8..6a6d24bf6c 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v3r1.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -58,7 +58,16 @@ template <
     typename CBlockTransferClusterLengths_MBlock_MXdlPerWave_MWaveMPerXdl_NBlock_NXdlPerWave_NWaveNPerXdl,
     index_t CBlockTransferScalarPerVector_NWaveNPerXdl>
 struct DeviceConv2dFwdXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-    : public DeviceConvFwd<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>
+    : public DeviceConvFwd<2,
+                           ck::tensor_layout::convolution::NHWC,
+                           ck::tensor_layout::convolution::KYXC,
+                           ck::tensor_layout::convolution::NHWK,
+                           InDataType,
+                           WeiDataType,
+                           OutDataType,
+                           InElementwiseOperation,
+                           WeiElementwiseOperation,
+                           OutElementwiseOperation>
 {
     using DeviceOp = DeviceConv2dFwdXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K;
 
diff --git a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp
index 84166d6f5f..5821e06b2c 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -55,7 +55,16 @@ template <typename InDataType,
           ck::index_t CThreadTransferSrcDstVectorDim,
           ck::index_t CThreadTransferDstScalarPerVector>
 struct DeviceConv2dFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-    : public DeviceConvFwd<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>
+    : public DeviceConvFwd<2,
+                           ck::tensor_layout::convolution::NHWC,
+                           ck::tensor_layout::convolution::KYXC,
+                           ck::tensor_layout::convolution::NHWK,
+                           InDataType,
+                           WeiDataType,
+                           OutDataType,
+                           InElementwiseOperation,
+                           WeiElementwiseOperation,
+                           OutElementwiseOperation>
 {
     using DeviceOp = DeviceConv2dFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K;
 
diff --git a/include/ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp b/include/ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp
index 83c19703b8..82054a3c94 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp
@@ -4,16 +4,21 @@
 #pragma once
 
 #include <vector>
-#include <iostream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 namespace ck {
 namespace tensor_operation {
 namespace device {
 
-template <typename InElementwiseOperation,
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementwiseOperation,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation>
 struct DeviceConvBwdData : public BaseOperator
@@ -39,12 +44,6 @@ struct DeviceConvBwdData : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename InElementwiseOperation,
-          typename WeiElementwiseOperation,
-          typename OutElementwiseOperation>
-using DeviceConvBwdDataPtr = std::unique_ptr<
-    DeviceConvBwdData<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_conv_backward_weight.hpp b/include/ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp
similarity index 82%
rename from include/ck/tensor_operation/gpu/device/device_conv_backward_weight.hpp
rename to include/ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp
index f171202530..91d2203d13 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv_backward_weight.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp
@@ -4,7 +4,6 @@
 #pragma once
 
 #include <vector>
-#include <iostream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 
@@ -12,7 +11,14 @@ namespace ck {
 namespace tensor_operation {
 namespace device {
 
-template <typename InElementwiseOperation,
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementwiseOperation,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation>
 struct DeviceConvBwdWeight : public BaseOperator
@@ -39,12 +45,6 @@ struct DeviceConvBwdWeight : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename InElementwiseOperation,
-          typename WeiElementwiseOperation,
-          typename OutElementwiseOperation>
-using DeviceConvBwdWeightPtr = std::unique_ptr<
-    DeviceConvBwdWeight<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_conv_fwd.hpp b/include/ck/tensor_operation/gpu/device/device_conv_fwd.hpp
index 5a3fb60d3b..4b9881088d 100644
--- a/include/ck/tensor_operation/gpu/device/device_conv_fwd.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_conv_fwd.hpp
@@ -3,7 +3,6 @@
 
 #pragma once
 
-#include <iostream>
 #include <vector>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
@@ -12,7 +11,14 @@ namespace ck {
 namespace tensor_operation {
 namespace device {
 
-template <typename InElementwiseOperation,
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType,
+          typename InElementwiseOperation,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation>
 struct DeviceConvFwd : public BaseOperator
@@ -38,12 +44,6 @@ struct DeviceConvFwd : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename InElementwiseOperation,
-          typename WeiElementwiseOperation,
-          typename OutElementwiseOperation>
-using DeviceConvFwdPtr = std::unique_ptr<
-    DeviceConvFwd<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp b/include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp
similarity index 97%
rename from include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp
rename to include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp
index a5970c8f13..e10e374b06 100644
--- a/include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp
@@ -13,15 +13,16 @@
 #include "ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_backward_data_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
 namespace device {
 
 // out[N, Ho, Wo, K] = in[N, Hi, Wi, C] * wei[K, Y, X, C]
-template <typename InDataType,
+template <ck::index_t NDimSpatial,
+          typename InDataType,
           typename WeiDataType,
           typename OutDataType,
           typename AccDataType,
@@ -29,7 +30,6 @@ template <typename InDataType,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation,
           ConvolutionBackwardDataSpecialization ConvBackwardDataSpecialization,
-          ck::index_t NumDimSpatial,
           ck::index_t BlockSize,
           ck::index_t MPerBlock,
           ck::index_t NPerBlock,
@@ -55,12 +55,29 @@ template <typename InDataType,
           bool BBlockLdsAddExtraN,
           ck::index_t CThreadTransferSrcDstVectorDim,
           ck::index_t CThreadTransferDstScalarPerVector>
-struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K
-    : public DeviceConvBwdData<InElementwiseOperation,
-                               WeiElementwiseOperation,
-                               OutElementwiseOperation>
+struct DeviceConvNdBwdDataNwcKxcNwk_Xdl
+    : public DeviceConvBwdData<
+          NDimSpatial,
+          ck::tuple_element_t<NDimSpatial - 1,
+                              ck::Tuple<ck::tensor_layout::convolution::NWC,
+                                        ck::tensor_layout::convolution::NHWC,
+                                        ck::tensor_layout::convolution::NDHWC>>,
+          ck::tuple_element_t<NDimSpatial - 1,
+                              ck::Tuple<ck::tensor_layout::convolution::KXC,
+                                        ck::tensor_layout::convolution::KYXC,
+                                        ck::tensor_layout::convolution::KZYXC>>,
+          ck::tuple_element_t<NDimSpatial - 1,
+                              ck::Tuple<ck::tensor_layout::convolution::NWK,
+                                        ck::tensor_layout::convolution::NHWK,
+                                        ck::tensor_layout::convolution::NDHWK>>,
+          InDataType,
+          WeiDataType,
+          OutDataType,
+          InElementwiseOperation,
+          WeiElementwiseOperation,
+          OutElementwiseOperation>
 {
-    using DeviceOp = DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K;
+    using DeviceOp = DeviceConvNdBwdDataNwcKxcNwk_Xdl;
 
     using ADataType = OutDataType;
     using BDataType = WeiDataType;
@@ -950,7 +967,7 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
                                                                   {0, 0, 0});
     }
 
-    using ABCGridDescs = decltype(GetABCGridDesc<NumDimSpatial>());
+    using ABCGridDescs = decltype(GetABCGridDesc<NDimSpatial>());
 
     using AGridDesc_K0_M_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I0])>;
     using BGridDesc_K0_N_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I1])>;
@@ -1037,7 +1054,7 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
               input_left_pads_{input_left_pads},
               input_right_pads_{input_right_pads}
         {
-            CreateABCDesc<NumDimSpatial>();
+            CreateABCDesc<NDimSpatial>();
         }
 
         template <ck::index_t NDim, typename ck::enable_if<NDim == 1, bool>::type = false>
@@ -1060,7 +1077,7 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
                 }
 
                 const auto descs =
-                    DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NumDimSpatial>(
+                    DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NDimSpatial>(
                         Conv_N_,
                         Conv_K_,
                         Conv_C_,
@@ -1118,7 +1135,7 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
                     }
 
                     const auto descs =
-                        DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NumDimSpatial>(
+                        DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NDimSpatial>(
                             Conv_N_,
                             Conv_K_,
                             Conv_C_,
@@ -1186,18 +1203,18 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
                         }
 
                         const auto descs =
-                            DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<
-                                NumDimSpatial>(Conv_N_,
-                                               Conv_K_,
-                                               Conv_C_,
-                                               input_spatial_lengths_,
-                                               filter_spatial_lengths_,
-                                               output_spatial_lengths_,
-                                               conv_filter_strides_,
-                                               conv_filter_dilations_,
-                                               input_left_pads_,
-                                               input_right_pads_,
-                                               {i_ztilde, i_ytilde, i_xtilde});
+                            DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NDimSpatial>(
+                                Conv_N_,
+                                Conv_K_,
+                                Conv_C_,
+                                input_spatial_lengths_,
+                                filter_spatial_lengths_,
+                                output_spatial_lengths_,
+                                conv_filter_strides_,
+                                conv_filter_dilations_,
+                                input_left_pads_,
+                                input_right_pads_,
+                                {i_ztilde, i_ytilde, i_xtilde});
                         a_grid_desc_k0_m_k1_container_.push_back(descs[I0]);
                         b_grid_desc_k0_n_k1_container_.push_back(descs[I1]);
                         c_grid_desc_m_n_container_.push_back(descs[I2]);
@@ -1398,7 +1415,7 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
                      ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0)
         {
             // check if it's 1x1, stride=1 pad = 0 conv
-            for(int i = 0; i < NumDimSpatial; i++)
+            for(int i = 0; i < NDimSpatial; i++)
             {
                 if(!(arg.filter_spatial_lengths_[i] == 1 && arg.conv_filter_strides_[i] == 1 &&
                      arg.input_left_pads_[i] == 0 && arg.input_right_pads_[i] == 0))
@@ -1528,7 +1545,7 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
         auto str = std::stringstream();
 
         // clang-format off
-        str << "DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K"
+        str << "DeviceConvNdBwdDataNwcKxcNwk_Xdl"
             << "<"
             << BlockSize << ", "
             << MPerBlock << ", "
diff --git a/include/ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp
similarity index 71%
rename from include/ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
rename to include/ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp
index 32d91269b4..50e6b538bd 100644
--- a/include/ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp
@@ -10,19 +10,20 @@
 #include "ck/tensor_description/tensor_descriptor.hpp"
 #include "ck/tensor_description/tensor_descriptor_helper.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv_backward_weight.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp"
 #include "ck/tensor_operation/gpu/device/convolution_backward_weight_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_bwd_weight.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_unary_elementwise_1d.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
 namespace device {
 
 // out[N, Ho, Wo, K] = in[N, Hi, Wi, C] * wei[K, Y, X, C]
-template <typename InDataType,
+template <ck::index_t NDimSpatial,
+          typename InDataType,
           typename WeiDataType,
           typename OutDataType,
           typename AccDataType,
@@ -30,7 +31,6 @@ template <typename InDataType,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation,
           ConvolutionBackwardWeightSpecialization ConvBackwardWeightSpecialization,
-          ck::index_t NumDimSpatial,
           ck::index_t BlockSize,
           ck::index_t MPerBlock,
           ck::index_t NPerBlock,
@@ -58,13 +58,29 @@ template <typename InDataType,
           index_t CShuffleNXdlPerWavePerShuffle,
           typename CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CBlockTransferScalarPerVector_NWaveNPerXdl>
-struct DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-    : public DeviceConvBwdWeight<InElementwiseOperation,
-                                 WeiElementwiseOperation,
-                                 OutElementwiseOperation>
+struct DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle
+    : public DeviceConvBwdWeight<
+          NDimSpatial,
+          ck::tuple_element_t<NDimSpatial - 1,
+                              ck::Tuple<ck::tensor_layout::convolution::NWC,
+                                        ck::tensor_layout::convolution::NHWC,
+                                        ck::tensor_layout::convolution::NDHWC>>,
+          ck::tuple_element_t<NDimSpatial - 1,
+                              ck::Tuple<ck::tensor_layout::convolution::KXC,
+                                        ck::tensor_layout::convolution::KYXC,
+                                        ck::tensor_layout::convolution::KZYXC>>,
+          ck::tuple_element_t<NDimSpatial - 1,
+                              ck::Tuple<ck::tensor_layout::convolution::NWK,
+                                        ck::tensor_layout::convolution::NHWK,
+                                        ck::tensor_layout::convolution::NDHWK>>,
+          InDataType,
+          WeiDataType,
+          OutDataType,
+          InElementwiseOperation,
+          WeiElementwiseOperation,
+          OutElementwiseOperation>
 {
-    using DeviceOp =
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K;
+    using DeviceOp = DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle;
 
     using ADataType = OutDataType;
     using BDataType = InDataType;
@@ -675,125 +691,19 @@ struct DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
             return PadDescriptor_M0_1d(desc, gridSize, blockSize);
     }
 
-    using TypeConvertFp32ToBf16Functor =
-        ck::tensor_operation::element_wise::UnaryTypeConvert<ck::bhalf_t, float>;
-    using GridDesc_M0      = decltype(MakeDescriptor_M0<1>({1}, {1}, 1, 1));
-    using GridwiseUEltwise = GridwiseUnaryElementwise_1D<AccDataType,
-                                                         InDataType,
-                                                         GridDesc_M0,
-                                                         TypeConvertFp32ToBf16Functor,
-                                                         4>;
+    using GridDesc_M0 = decltype(MakeDescriptor_M0<1>({1}, {1}, 1, 1));
 
-    using ABCGridDescs = decltype(GetABCGridDesc<NumDimSpatial>());
+    using ABCGridDescs = decltype(GetABCGridDesc<NDimSpatial>());
 
     using AGridDesc_K0_M_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I0])>;
     using BGridDesc_K0_N_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I1])>;
     using CGridDesc_M_N     = remove_cvref_t<decltype(ABCGridDescs{}[I2])>;
 
-    // GridwiseGemm
     using GridwiseGemm = GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_bwd_weight<
         BlockSize,
         ADataType, // TODO: distinguish A/B datatype
         AccDataType,
         CDataType,
-        InMemoryDataOperationEnum::Set,
-        AGridDesc_K0_M_K1,
-        BGridDesc_K0_N_K1,
-        CGridDesc_M_N,
-        AElementwiseOperation,
-        BElementwiseOperation,
-        CElementwiseOperation,
-        MPerBlock,
-        NPerBlock,
-        K0PerBlock,
-        MPerXdl,
-        NPerXdl,
-        K1,
-        MXdlPerWave,
-        NXdlPerWave,
-        ABlockTransferThreadClusterLengths_K0_M_K1,
-        ABlockTransferThreadClusterArrangeOrder,
-        ABlockTransferSrcAccessOrder,
-        ABlockTransferSrcVectorDim,
-        ABlockTransferSrcScalarPerVector,
-        ABlockTransferDstScalarPerVector_K1,
-        false, // AThreadTransferSrcResetCoordinateAfterRun,
-        ABlockLdsAddExtraM,
-        ABlockLdsM1PerBlock,
-        ABlockLdsM0PerBlock,
-        ABlockLdsM1Padding,
-        BBlockTransferThreadClusterLengths_K0_N_K1,
-        BBlockTransferThreadClusterArrangeOrder,
-        BBlockTransferSrcAccessOrder,
-        BBlockTransferSrcVectorDim,
-        BBlockTransferSrcScalarPerVector,
-        BBlockTransferDstScalarPerVector_K1,
-        false, // BThreadTransferSrcResetCoordinateAfterRun,
-        BBlockLdsAddExtraN,
-        BBlockLdsN1PerBlock,
-        BBlockLdsN0PerBlock,
-        BBlockLdsN1Padding,
-        CShuffleMXdlPerWavePerShuffle,
-        CShuffleNXdlPerWavePerShuffle,
-        CBlockTransferScalarPerVector_NWaveNPerXdl,
-        CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
-        true,
-        true>;
-
-    using GridwiseGemmAtomicAdd = GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_bwd_weight<
-        BlockSize,
-        ADataType, // TODO: distinguish A/B datatype
-        AccDataType,
-        CDataType,
-        InMemoryDataOperationEnum::AtomicAdd,
-        AGridDesc_K0_M_K1,
-        BGridDesc_K0_N_K1,
-        CGridDesc_M_N,
-        AElementwiseOperation,
-        BElementwiseOperation,
-        CElementwiseOperation,
-        MPerBlock,
-        NPerBlock,
-        K0PerBlock,
-        MPerXdl,
-        NPerXdl,
-        K1,
-        MXdlPerWave,
-        NXdlPerWave,
-        ABlockTransferThreadClusterLengths_K0_M_K1,
-        ABlockTransferThreadClusterArrangeOrder,
-        ABlockTransferSrcAccessOrder,
-        ABlockTransferSrcVectorDim,
-        ABlockTransferSrcScalarPerVector,
-        ABlockTransferDstScalarPerVector_K1,
-        false, // AThreadTransferSrcResetCoordinateAfterRun,
-        ABlockLdsAddExtraM,
-        ABlockLdsM1PerBlock,
-        ABlockLdsM0PerBlock,
-        ABlockLdsM1Padding,
-        BBlockTransferThreadClusterLengths_K0_N_K1,
-        BBlockTransferThreadClusterArrangeOrder,
-        BBlockTransferSrcAccessOrder,
-        BBlockTransferSrcVectorDim,
-        BBlockTransferSrcScalarPerVector,
-        BBlockTransferDstScalarPerVector_K1,
-        false, // BThreadTransferSrcResetCoordinateAfterRun,
-        BBlockLdsAddExtraN,
-        BBlockLdsN1PerBlock,
-        BBlockLdsN0PerBlock,
-        BBlockLdsN1Padding,
-        CShuffleMXdlPerWavePerShuffle,
-        CShuffleNXdlPerWavePerShuffle,
-        CBlockTransferScalarPerVector_NWaveNPerXdl,
-        CBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
-        true,
-        true>;
-
-    using GridwiseGemmAtomicAddFloatBf16Splitk = GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_bwd_weight<
-        BlockSize,
-        ADataType, // TODO: distinguish A/B datatype
-        AccDataType,
-        AccDataType,
         InMemoryDataOperationEnum::AtomicAdd,
         AGridDesc_K0_M_K1,
         BGridDesc_K0_N_K1,
@@ -890,7 +800,7 @@ struct DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
               k_batch_{split_k}
         {
             const auto descs =
-                DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NumDimSpatial>(
+                DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N<NDimSpatial>(
                     N,
                     K,
                     C,
@@ -980,268 +890,55 @@ struct DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
                 throw std::runtime_error(
                     "wrong! GridwiseGemm_km_kn_m0m1n0n1_xdlops_v3r1 has invalid setting");
             }
-            const auto kbatch = arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I0);
+
             const index_t grid_size =
                 arg.block_2_ctile_map_.CalculateGridSize(arg.c_grid_desc_m_n_);
 
             const auto K0 = arg.a_grid_desc_kbatch_k0_m_k1_.GetLength(I1);
 
-            float ave_time = 0;
-
             const bool has_main_k0_block_loop = GridwiseGemm::CalculateHasMainK0BlockLoop(K0);
 
-            const auto run_conv = [&](const auto& kernel) {
-                hipGetErrorString(hipMemset(
-                    arg.p_c_grid_,
-                    0,
-                    arg.c_grid_desc_mblock_mperblock_nblock_nperblock_.GetElementSpaceSize() *
-                        sizeof(CDataType)));
-                float elapsed_time =
-                    launch_and_time_kernel(stream_config,
-                                           kernel,
-                                           dim3(grid_size),
-                                           dim3(BlockSize),
-                                           0,
-                                           arg.p_a_grid_,
-                                           arg.p_b_grid_,
-                                           arg.p_c_grid_,
-                                           arg.a_grid_desc_kbatch_k0_m_k1_,
-                                           arg.b_grid_desc_kbatch_k0_n_k1_,
-                                           arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                           arg.a_element_op_,
-                                           arg.b_element_op_,
-                                           arg.c_element_op_,
-                                           arg.block_2_ctile_map_);
+            auto launch_kernel = [&](auto has_main_k_block_loop) {
+                constexpr bool has_main_loop = has_main_k_block_loop.value;
 
-                hipGetErrorString(hipMemset(
-                    arg.p_c_grid_,
-                    0,
-                    arg.c_grid_desc_mblock_mperblock_nblock_nperblock_.GetElementSpaceSize() *
-                        sizeof(CDataType)));
+                const auto kernel = kernel_gemm_xdlops_bwd_weight<
+                    GridwiseGemm,
+                    ADataType, // TODO: distiguish A/B datatype
+                    CDataType,
+                    remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
+                    remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
+                    remove_reference_t<DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
+                    OutElementwiseOperation,
+                    InElementwiseOperation,
+                    WeiElementwiseOperation,
+                    remove_reference_t<DeviceOp::Block2CTileMap>,
+                    has_main_loop>;
 
-                launch_and_time_kernel(StreamConfig{nullptr, false},
-                                       kernel,
-                                       dim3(grid_size),
-                                       dim3(BlockSize),
-                                       0,
-                                       arg.p_a_grid_,
-                                       arg.p_b_grid_,
-                                       arg.p_c_grid_,
-                                       arg.a_grid_desc_kbatch_k0_m_k1_,
-                                       arg.b_grid_desc_kbatch_k0_n_k1_,
-                                       arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                       arg.a_element_op_,
-                                       arg.b_element_op_,
-                                       arg.c_element_op_,
-                                       arg.block_2_ctile_map_);
-
-                return elapsed_time;
+                return launch_and_time_kernel(stream_config,
+                                              kernel,
+                                              dim3(grid_size),
+                                              dim3(BlockSize),
+                                              0,
+                                              arg.p_a_grid_,
+                                              arg.p_b_grid_,
+                                              arg.p_c_grid_,
+                                              arg.a_grid_desc_kbatch_k0_m_k1_,
+                                              arg.b_grid_desc_kbatch_k0_n_k1_,
+                                              arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
+                                              arg.a_element_op_,
+                                              arg.b_element_op_,
+                                              arg.c_element_op_,
+                                              arg.block_2_ctile_map_);
             };
 
-            // run kernel for bf16 with splitk
-            const auto run_bf16_splitk = [&](const auto& kernel) {
-                hipGetErrorString(hipMemset(
-                    arg.p_workspace_,
-                    0,
-                    arg.c_grid_desc_mblock_mperblock_nblock_nperblock_.GetElementSpaceSize() *
-                        sizeof(AccDataType)));
-
-                float elapsed_time =
-                    launch_and_time_kernel(stream_config,
-                                           kernel,
-                                           dim3(grid_size),
-                                           dim3(BlockSize),
-                                           0,
-                                           arg.p_a_grid_,
-                                           arg.p_b_grid_,
-                                           static_cast<AccDataType*>(arg.p_workspace_),
-                                           arg.a_grid_desc_kbatch_k0_m_k1_,
-                                           arg.b_grid_desc_kbatch_k0_n_k1_,
-                                           arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                           arg.a_element_op_,
-                                           arg.b_element_op_,
-                                           arg.c_element_op_,
-                                           arg.block_2_ctile_map_);
-
-                hipGetErrorString(hipMemset(
-                    arg.p_workspace_,
-                    0,
-                    arg.c_grid_desc_mblock_mperblock_nblock_nperblock_.GetElementSpaceSize() *
-                        sizeof(AccDataType)));
-
-                launch_and_time_kernel(StreamConfig{nullptr, false},
-                                       kernel,
-                                       dim3(grid_size),
-                                       dim3(BlockSize),
-                                       0,
-                                       arg.p_a_grid_,
-                                       arg.p_b_grid_,
-                                       static_cast<AccDataType*>(arg.p_workspace_),
-                                       arg.a_grid_desc_kbatch_k0_m_k1_,
-                                       arg.b_grid_desc_kbatch_k0_n_k1_,
-                                       arg.c_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                       arg.a_element_op_,
-                                       arg.b_element_op_,
-                                       arg.c_element_op_,
-                                       arg.block_2_ctile_map_);
-
-                return elapsed_time;
-            };
-
-            // kernel for type conversion
-            std::vector<std::size_t> filter_dims{static_cast<std::size_t>(arg.Conv_K_),
-                                                 static_cast<std::size_t>(arg.Conv_C_)};
-
-            filter_dims.insert(std::end(filter_dims),
-                               std::begin(arg.filter_spatial_lengths_),
-                               std::end(arg.filter_spatial_lengths_));
-
-            int tensor_size =
-                std::accumulate(filter_dims.begin(), filter_dims.end(), 1, std::multiplies<int>{});
-
-            const index_t type_convert_grid_size = GridwiseUEltwise::CalculateGridSize(tensor_size);
-            GridDesc_M0 a_grid_desc_m0_ =
-                MakeDescriptor_M0<1>({tensor_size}, {1}, type_convert_grid_size, 256);
-            GridDesc_M0 b_grid_desc_m0_ =
-                MakeDescriptor_M0<1>({tensor_size}, {1}, type_convert_grid_size, 256);
-
-            if(!GridwiseUEltwise::CheckValidity(a_grid_desc_m0_, b_grid_desc_m0_))
+            if(has_main_k0_block_loop)
             {
-                throw std::runtime_error("wrong! GridwiseUnaryElementwise_1D has invalid setting");
-            }
-
-            // run kernel for type conversion
-            void* p_c_grid_tmp_            = static_cast<void*>(arg.p_c_grid_);
-            InDataType* p_c_grid_tmp_bf16_ = static_cast<InDataType*>(p_c_grid_tmp_);
-            const auto run_type_convert    = [&](const auto& kernel) {
-                float elapsed_time =
-                    launch_and_time_kernel(stream_config,
-                                           kernel,
-                                           dim3(type_convert_grid_size),
-                                           dim3(256),
-                                           0,
-                                           static_cast<AccDataType*>(arg.p_workspace_),
-                                           p_c_grid_tmp_bf16_,
-                                           a_grid_desc_m0_,
-                                           b_grid_desc_m0_,
-                                           TypeConvertFp32ToBf16Functor{});
-                return elapsed_time;
-            };
-
-            if constexpr(std::is_same<InDataType, ck::bhalf_t>::value)
-            {
-                auto launch_kernel = [&](auto has_main_k_block_loop) {
-                    constexpr bool has_main_loop = has_main_k_block_loop.value;
-
-                    if(kbatch == 1)
-                    {
-                        const auto kernel = kernel_gemm_xdlops_bwd_weight<
-                            GridwiseGemm,
-                            ADataType, // TODO: distiguish A/B datatype
-                            CDataType,
-                            remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
-                            remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
-                            remove_reference_t<
-                                DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
-                            OutElementwiseOperation,
-                            InElementwiseOperation,
-                            WeiElementwiseOperation,
-                            remove_reference_t<DeviceOp::Block2CTileMap>,
-                            has_main_loop>;
-
-                        return run_conv(kernel);
-                    }
-                    else
-                    {
-                        const auto kernel_type_convert =
-                            kernel_unary_elementwise_1d<GridwiseUEltwise,
-                                                        AccDataType,
-                                                        InDataType,
-                                                        GridDesc_M0,
-                                                        TypeConvertFp32ToBf16Functor>;
-
-                        const auto kernel_conv = kernel_gemm_xdlops_bwd_weight<
-                            GridwiseGemmAtomicAddFloatBf16Splitk,
-                            ADataType, // TODO: distiguish A/B datatype
-                            AccDataType,
-                            remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
-                            remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
-                            remove_reference_t<
-                                DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
-                            OutElementwiseOperation,
-                            InElementwiseOperation,
-                            WeiElementwiseOperation,
-                            remove_reference_t<DeviceOp::Block2CTileMap>,
-                            has_main_loop>;
-
-                        float elapsed_time = 0;
-                        elapsed_time += run_bf16_splitk(kernel_conv);
-                        elapsed_time += run_type_convert(kernel_type_convert);
-                        return elapsed_time;
-                    }
-                };
-                if(has_main_k0_block_loop)
-                {
-                    ave_time = launch_kernel(integral_constant<bool, true>{});
-                }
-                else
-                {
-                    ave_time = launch_kernel(integral_constant<bool, false>{});
-                }
+                return launch_kernel(integral_constant<bool, true>{});
             }
             else
             {
-                auto launch_kernel = [&](auto has_main_k_block_loop) {
-                    constexpr bool has_main_loop = has_main_k_block_loop.value;
-
-                    if(kbatch == 1)
-                    {
-                        const auto kernel = kernel_gemm_xdlops_bwd_weight<
-                            GridwiseGemm,
-                            ADataType, // TODO: distiguish A/B datatype
-                            CDataType,
-                            remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
-                            remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
-                            remove_reference_t<
-                                DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
-                            OutElementwiseOperation,
-                            InElementwiseOperation,
-                            WeiElementwiseOperation,
-                            remove_reference_t<DeviceOp::Block2CTileMap>,
-                            has_main_loop>;
-
-                        return run_conv(kernel);
-                    }
-                    else
-                    {
-                        const auto kernel = kernel_gemm_xdlops_bwd_weight<
-                            GridwiseGemmAtomicAdd,
-                            ADataType, // TODO: distiguish A/B datatype
-                            CDataType,
-                            remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
-                            remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
-                            remove_reference_t<
-                                DeviceOp::CGridDesc_MBlock_MPerBlock_NBlock_NPerBlock>,
-                            OutElementwiseOperation,
-                            InElementwiseOperation,
-                            WeiElementwiseOperation,
-                            remove_reference_t<DeviceOp::Block2CTileMap>,
-                            has_main_loop>;
-
-                        return run_conv(kernel);
-                    }
-                };
-                if(has_main_k0_block_loop)
-                {
-                    ave_time = launch_kernel(integral_constant<bool, true>{});
-                }
-                else
-                {
-                    ave_time = launch_kernel(integral_constant<bool, false>{});
-                }
+                return launch_kernel(integral_constant<bool, false>{});
             }
-
-            return ave_time;
         }
 
         float Run(const BaseArgument* p_arg,
@@ -1263,7 +960,7 @@ struct DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
                      ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0)
         {
             // check if it's 1x1, stride=1 pad = 0 conv
-            for(int i = 0; i < NumDimSpatial; i++)
+            for(int i = 0; i < NDimSpatial; i++)
             {
                 if(!(arg.filter_spatial_lengths_[i] == 1 && arg.conv_filter_strides_[i] == 1 &&
                      arg.input_left_pads_[i] == 0 && arg.input_right_pads_[i] == 0))
@@ -1390,74 +1087,18 @@ struct DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
         auto str = std::stringstream();
 
         // clang-format off
-        str << "DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K"
+        str << "DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle"
             << "<"
             << BlockSize << ", "
             << MPerBlock << ", "
             << NPerBlock << ", "
-            << K0PerBlock
+            << K0PerBlock << ", "
+            << getConvBackwardWeightSpecializationString(ConvBackwardWeightSpecialization)
             << ">";
-        if constexpr(ConvBackwardWeightSpecialization ==
-                     ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0){
-
-            str << " Filter1x1Stride1Pad0";
-        }
-
         // clang-format on
 
         return str.str();
     }
-
-    template <ck::index_t NDim, typename ck::enable_if<NDim == 1, bool>::type = false>
-    static size_t GetWorkSpaceSize(const Argument& arg)
-    {
-        size_t WorkSpaceSize = 0;
-        if(arg.k_batch_ > 1)
-        {
-            if constexpr(std::is_same<InDataType, ck::bhalf_t>::value)
-            {
-                WorkSpaceSize =
-                    arg.Conv_K_ * arg.Conv_C_ * arg.filter_spatial_lengths_[0] * sizeof(float);
-            }
-        }
-        return WorkSpaceSize;
-    }
-
-    template <ck::index_t NDim, typename ck::enable_if<NDim == 2, bool>::type = false>
-    static size_t GetWorkSpaceSize(const Argument& arg)
-    {
-        size_t WorkSpaceSize = 0;
-        if(arg.k_batch_ > 1)
-        {
-            if constexpr(std::is_same<InDataType, ck::bhalf_t>::value)
-            {
-                WorkSpaceSize = arg.Conv_K_ * arg.Conv_C_ * arg.filter_spatial_lengths_[0] *
-                                arg.filter_spatial_lengths_[1] * sizeof(float);
-            }
-        }
-        return WorkSpaceSize;
-    }
-
-    template <ck::index_t NDim, typename ck::enable_if<NDim == 3, bool>::type = false>
-    static size_t GetWorkSpaceSize(const Argument& arg)
-    {
-        size_t WorkSpaceSize = 0;
-        if(arg.k_batch_ > 1)
-        {
-            if constexpr(std::is_same<InDataType, ck::bhalf_t>::value)
-            {
-                WorkSpaceSize = arg.Conv_K_ * arg.Conv_C_ * arg.filter_spatial_lengths_[0] *
-                                arg.filter_spatial_lengths_[1] * arg.filter_spatial_lengths_[2] *
-                                sizeof(float);
-            }
-        }
-        return WorkSpaceSize;
-    }
-
-    size_t GetWorkSpaceSize(const BaseArgument* p_arg) const override final
-    {
-        return GetWorkSpaceSize<NumDimSpatial>(*dynamic_cast<const Argument*>(p_arg));
-    }
 };
 
 } // namespace device
diff --git a/include/ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp b/include/ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp
deleted file mode 100644
index 78f0f02898..0000000000
--- a/include/ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp
+++ /dev/null
@@ -1,1046 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#pragma once
-
-#include <functional>
-#include <iostream>
-#include <iterator>
-#include <numeric>
-#include <sstream>
-
-#include "ck/utility/common_header.hpp"
-#include "ck/tensor_description/tensor_descriptor.hpp"
-#include "ck/tensor_description/tensor_descriptor_helper.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
-#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
-#include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-
-//
-// @brief      Device Convolution operation.
-//
-// Supports:
-//  @li         Inputs with up to 3 spatial dimentions
-//  @li         Input tensor in NHWC data format
-//  @li         Weight tensor in KYXC data format
-//  @li         Output tensor in NHWK data format
-//
-// 1D:
-// out[N, Wo, K] = in[N, Wi, C] * wei[K, X, C]
-// 2D:
-// out[N, Ho, Wo, K] = in[N, Hi, Wi, C] * wei[K, Y, X, C]
-// 3D:
-// out[N, Do, Ho, Wo, K] = in[N, Di, Hi, Wi, C] * wei[K, Z, Y, X, C]
-//
-template <typename InDataType,
-          typename WeiDataType,
-          typename OutDataType,
-          typename AccDataType,
-          typename InElementwiseOperation,
-          typename WeiElementwiseOperation,
-          typename OutElementwiseOperation,
-          ConvolutionForwardSpecialization ConvForwardSpecialization,
-          ck::index_t NumDimSpatial,
-          ck::index_t BlockSize,
-          ck::index_t MPerBlock,
-          ck::index_t NPerBlock,
-          ck::index_t K0PerBlock,
-          ck::index_t K1,
-          ck::index_t MPerXDL,
-          ck::index_t NPerXDL,
-          ck::index_t MXdlPerWave,
-          ck::index_t NXdlPerWave,
-          typename ABlockTransferThreadClusterLengths_K0_M_K1,
-          typename ABlockTransferThreadClusterArrangeOrder,
-          typename ABlockTransferSrcAccessOrder,
-          ck::index_t ABlockTransferSrcVectorDim,
-          ck::index_t ABlockTransferSrcScalarPerVector,
-          ck::index_t ABlockTransferDstScalarPerVector_K1,
-          bool ABlockLdsAddExtraM,
-          typename BBlockTransferThreadClusterLengths_K0_N_K1,
-          typename BBlockTransferThreadClusterArrangeOrder,
-          typename BBlockTransferSrcAccessOrder,
-          ck::index_t BBlockTransferSrcVectorDim,
-          ck::index_t BBlockTransferSrcScalarPerVector,
-          ck::index_t BBlockTransferDstScalarPerVector_K1,
-          bool BBlockLdsAddExtraN,
-          ck::index_t CThreadTransferSrcDstVectorDim,
-          ck::index_t CThreadTransferDstScalarPerVector>
-struct DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-    : public DeviceConvFwd<InElementwiseOperation, WeiElementwiseOperation, OutElementwiseOperation>
-{
-    using DeviceOp = DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K;
-
-    using ADataType = InDataType;
-    using BDataType = WeiDataType;
-    using CDataType = OutDataType;
-
-    // TODO make A/B datatype different
-    using ABDataType = InDataType;
-
-    static constexpr index_t NDimSpatial = NumDimSpatial;
-
-    static constexpr auto I0 = Number<0>{};
-    static constexpr auto I1 = Number<1>{};
-    static constexpr auto I2 = Number<2>{};
-    static constexpr auto I3 = Number<3>{};
-
-    static constexpr auto K1Number     = Number<K1>{};
-    static constexpr auto GemmK1Number = K1Number;
-
-    static auto GetWeightTensorDescriptor(ck::index_t gemm_n, ck::index_t gemm_k)
-    {
-        const ck::index_t gemm_k0 = gemm_k / GemmK1Number;
-        const auto wei_k_yxc_grid_desc =
-            make_naive_tensor_descriptor_packed(make_tuple(gemm_n, gemm_k));
-
-        // wei_gemmk0_gemmn_gemmk1_grid_desc
-        return transform_tensor_descriptor(
-            wei_k_yxc_grid_desc,
-            make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                       make_pass_through_transform(gemm_n)),
-            make_tuple(Sequence<1>{}, Sequence<0>{}),
-            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-    }
-
-    static auto
-    GetOutputTensorDescriptor(ck::index_t gemm_m, ck::index_t gemm_n, ck::index_t gemm_m_pad)
-    {
-        const auto out_gemmmraw_gemmn_grid_desc =
-            make_naive_tensor_descriptor_packed(make_tuple(gemm_m, gemm_n));
-
-        // out_gemmm_gemmn_grid_desc
-        return transform_tensor_descriptor(out_gemmmraw_gemmn_grid_desc,
-                                           make_tuple(make_right_pad_transform(gemm_m, gemm_m_pad),
-                                                      make_pass_through_transform(gemm_n)),
-                                           make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                           make_tuple(Sequence<0>{}, Sequence<1>{}));
-    }
-
-    template <ck::index_t NDim, typename std::enable_if<NDim == 1, bool>::type = false>
-    static auto GetInputTensorDescriptor(ck::index_t N,
-                                         ck::index_t C,
-                                         ck::index_t gemm_m,
-                                         ck::index_t gemm_k,
-                                         ck::index_t gemm_m_pad,
-                                         const std::vector<ck::index_t>& input_spatial_lengths,
-                                         const std::vector<ck::index_t>& filter_spatial_lengths,
-                                         const std::vector<ck::index_t>& output_spatial_lengths,
-                                         const std::vector<ck::index_t>& conv_filter_strides,
-                                         const std::vector<ck::index_t>& conv_filter_dilations,
-                                         const std::vector<ck::index_t>& input_left_pads,
-                                         const std::vector<ck::index_t>& input_right_pads)
-    {
-        const ck::index_t gemm_k0 = gemm_k / GemmK1Number;
-        const index_t Wi          = input_spatial_lengths[0];
-        const index_t Wo          = output_spatial_lengths[0];
-        const index_t ConvStrideW = conv_filter_strides[0];
-
-        if constexpr(ConvForwardSpecialization ==
-                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
-        {
-            const auto in_gemmmraw_gemmk_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(gemm_m, gemm_k));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmmraw_gemmk_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_right_pad_transform(gemm_m, gemm_m_pad)),
-                make_tuple(Sequence<1>{}, Sequence<0>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-        }
-        else if constexpr(ConvForwardSpecialization ==
-                          ConvolutionForwardSpecialization::Filter1x1Pad0)
-        {
-            const auto in_n_wi_c_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(N, Wi, C));
-
-            const auto in_n_wo_c_grid_desc = transform_tensor_descriptor(
-                in_n_wi_c_grid_desc,
-                make_tuple(make_pass_through_transform(N),
-                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
-                           make_pass_through_transform(C)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-
-            const auto in_gemmk0_gemmmraw_gemmk1_grid_desc = transform_tensor_descriptor(
-                in_n_wo_c_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_merge_transform(make_tuple(N, Wo))),
-                make_tuple(Sequence<2>{}, Sequence<0, 1>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmk0_gemmmraw_gemmk1_grid_desc,
-                make_tuple(make_pass_through_transform(gemm_k0),
-                           make_right_pad_transform(gemm_m, gemm_m_pad),
-                           make_pass_through_transform(GemmK1Number)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-        }
-        else
-        {
-            const index_t X             = filter_spatial_lengths[0];
-            const index_t ConvDilationW = conv_filter_dilations[0];
-            const index_t InLeftPadW    = input_left_pads[0];
-            const index_t InRightPadW   = input_right_pads[0];
-
-            const auto in_n_wi_c_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(N, Wi, C));
-
-            const auto in_n_wip_c_grid_desc = transform_tensor_descriptor(
-                in_n_wi_c_grid_desc,
-                make_tuple(make_pass_through_transform(N),
-                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
-                           make_pass_through_transform(C)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-
-            const auto in_n_x_wo_c_grid_desc = transform_tensor_descriptor(
-                in_n_wip_c_grid_desc,
-                make_tuple(
-                    make_pass_through_transform(N),
-                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
-                    make_pass_through_transform(C)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3>{}));
-
-            const auto in_gemmk_gemmmraw_grid_desc =
-                transform_tensor_descriptor(in_n_x_wo_c_grid_desc,
-                                            make_tuple(make_merge_transform(make_tuple(X, C)),
-                                                       make_merge_transform(make_tuple(N, Wo))),
-                                            make_tuple(Sequence<1, 3>{}, Sequence<0, 2>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto in_gemmk0_gemmmraw_gemmk1_grid_desc = transform_tensor_descriptor(
-                in_gemmk_gemmmraw_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_pass_through_transform(gemm_m)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmk0_gemmmraw_gemmk1_grid_desc,
-                make_tuple(make_pass_through_transform(gemm_k0),
-                           make_right_pad_transform(gemm_m, gemm_m_pad),
-                           make_pass_through_transform(GemmK1Number)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-        }
-    }
-
-    template <ck::index_t NDim, typename std::enable_if<NDim == 2, bool>::type = false>
-    static auto GetInputTensorDescriptor(ck::index_t N,
-                                         ck::index_t C,
-                                         ck::index_t gemm_m,
-                                         ck::index_t gemm_k,
-                                         ck::index_t gemm_m_pad,
-                                         const std::vector<ck::index_t>& input_spatial_lengths,
-                                         const std::vector<ck::index_t>& filter_spatial_lengths,
-                                         const std::vector<ck::index_t>& output_spatial_lengths,
-                                         const std::vector<ck::index_t>& conv_filter_strides,
-                                         const std::vector<ck::index_t>& conv_filter_dilations,
-                                         const std::vector<ck::index_t>& input_left_pads,
-                                         const std::vector<ck::index_t>& input_right_pads)
-    {
-        const ck::index_t gemm_k0 = gemm_k / GemmK1Number;
-        const index_t Hi          = input_spatial_lengths[0];
-        const index_t Wi          = input_spatial_lengths[1];
-
-        const index_t Ho = output_spatial_lengths[0];
-        const index_t Wo = output_spatial_lengths[1];
-
-        const index_t ConvStrideH = conv_filter_strides[0];
-        const index_t ConvStrideW = conv_filter_strides[1];
-
-        if constexpr(ConvForwardSpecialization ==
-                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
-        {
-            const auto in_gemmmraw_gemmk_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(gemm_m, gemm_k));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmmraw_gemmk_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_right_pad_transform(gemm_m, gemm_m_pad)),
-                make_tuple(Sequence<1>{}, Sequence<0>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-        }
-        else if constexpr(ConvForwardSpecialization ==
-                          ConvolutionForwardSpecialization::Filter1x1Pad0)
-        {
-            const auto in_n_hi_wi_c_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(N, Hi, Wi, C));
-
-            const auto in_n_ho_wo_c_grid_desc = transform_tensor_descriptor(
-                in_n_hi_wi_c_grid_desc,
-                make_tuple(make_pass_through_transform(N),
-                           make_embed_transform(make_tuple(Ho), make_tuple(ConvStrideH)),
-                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
-                           make_pass_through_transform(C)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
-
-            const auto in_gemmk0_gemmmraw_gemmk1_grid_desc = transform_tensor_descriptor(
-                in_n_ho_wo_c_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_merge_transform(make_tuple(N, Ho, Wo))),
-                make_tuple(Sequence<3>{}, Sequence<0, 1, 2>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmk0_gemmmraw_gemmk1_grid_desc,
-                make_tuple(make_pass_through_transform(gemm_k0),
-                           make_right_pad_transform(gemm_m, gemm_m_pad),
-                           make_pass_through_transform(GemmK1Number)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-        }
-        else
-        {
-            const index_t Y = filter_spatial_lengths[0];
-            const index_t X = filter_spatial_lengths[1];
-
-            const index_t ConvDilationH = conv_filter_dilations[0];
-            const index_t ConvDilationW = conv_filter_dilations[1];
-
-            const index_t InLeftPadH = input_left_pads[0];
-            const index_t InLeftPadW = input_left_pads[1];
-
-            const index_t InRightPadH = input_right_pads[0];
-            const index_t InRightPadW = input_right_pads[1];
-
-            const auto in_n_hi_wi_c_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(N, Hi, Wi, C));
-
-            const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
-                in_n_hi_wi_c_grid_desc,
-                make_tuple(make_pass_through_transform(N),
-                           make_pad_transform(Hi, InLeftPadH, InRightPadH),
-                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
-                           make_pass_through_transform(C)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
-
-            const auto in_n_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
-                in_n_hip_wip_c_grid_desc,
-                make_tuple(
-                    make_pass_through_transform(N),
-                    make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
-                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
-                    make_pass_through_transform(C)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
-                make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3, 4>{}, Sequence<5>{}));
-
-            const auto in_gemmk_gemmmraw_grid_desc =
-                transform_tensor_descriptor(in_n_y_ho_x_wo_c_grid_desc,
-                                            make_tuple(make_merge_transform(make_tuple(Y, X, C)),
-                                                       make_merge_transform(make_tuple(N, Ho, Wo))),
-                                            make_tuple(Sequence<1, 3, 5>{}, Sequence<0, 2, 4>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto in_gemmk0_gemmmraw_gemmk1_grid_desc = transform_tensor_descriptor(
-                in_gemmk_gemmmraw_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_pass_through_transform(gemm_m)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmk0_gemmmraw_gemmk1_grid_desc,
-                make_tuple(make_pass_through_transform(gemm_k0),
-                           make_right_pad_transform(gemm_m, gemm_m_pad),
-                           make_pass_through_transform(GemmK1Number)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-        }
-    }
-
-    template <ck::index_t NDim, typename std::enable_if<NDim == 3, bool>::type = false>
-    static auto GetInputTensorDescriptor(ck::index_t N,
-                                         ck::index_t C,
-                                         ck::index_t gemm_m,
-                                         ck::index_t gemm_k,
-                                         ck::index_t gemm_m_pad,
-                                         const std::vector<ck::index_t>& input_spatial_lengths,
-                                         const std::vector<ck::index_t>& filter_spatial_lengths,
-                                         const std::vector<ck::index_t>& output_spatial_lengths,
-                                         const std::vector<ck::index_t>& conv_filter_strides,
-                                         const std::vector<ck::index_t>& conv_filter_dilations,
-                                         const std::vector<ck::index_t>& input_left_pads,
-                                         const std::vector<ck::index_t>& input_right_pads)
-    {
-        const ck::index_t gemm_k0 = gemm_k / GemmK1Number;
-        const index_t Di          = input_spatial_lengths[0];
-        const index_t Hi          = input_spatial_lengths[1];
-        const index_t Wi          = input_spatial_lengths[2];
-
-        const index_t Do = output_spatial_lengths[0];
-        const index_t Ho = output_spatial_lengths[1];
-        const index_t Wo = output_spatial_lengths[2];
-
-        const index_t ConvStrideD = conv_filter_strides[0];
-        const index_t ConvStrideH = conv_filter_strides[1];
-        const index_t ConvStrideW = conv_filter_strides[2];
-
-        if constexpr(ConvForwardSpecialization ==
-                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
-        {
-            const auto in_gemmmraw_gemmk_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(gemm_m, gemm_k));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmmraw_gemmk_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_right_pad_transform(gemm_m, gemm_m_pad)),
-                make_tuple(Sequence<1>{}, Sequence<0>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-        }
-        else if constexpr(ConvForwardSpecialization ==
-                          ConvolutionForwardSpecialization::Filter1x1Pad0)
-        {
-            const auto in_n_di_hi_wi_c_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(N, Di, Hi, Wi, C));
-
-            const auto in_n_do_ho_wo_c_grid_desc = transform_tensor_descriptor(
-                in_n_di_hi_wi_c_grid_desc,
-                make_tuple(make_pass_through_transform(N),
-                           make_embed_transform(make_tuple(Do), make_tuple(ConvStrideD)),
-                           make_embed_transform(make_tuple(Ho), make_tuple(ConvStrideH)),
-                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
-                           make_pass_through_transform(C)),
-                make_tuple(
-                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
-                make_tuple(
-                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}));
-
-            const auto in_gemmk0_gemmmraw_gemmk1_grid_desc = transform_tensor_descriptor(
-                in_n_do_ho_wo_c_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_merge_transform(make_tuple(N, Do, Ho, Wo))),
-                make_tuple(Sequence<4>{}, Sequence<0, 1, 2, 3>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmk0_gemmmraw_gemmk1_grid_desc,
-                make_tuple(make_pass_through_transform(gemm_k0),
-                           make_right_pad_transform(gemm_m, gemm_m_pad),
-                           make_pass_through_transform(GemmK1Number)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-        }
-        else
-        {
-            const index_t Z = filter_spatial_lengths[0];
-            const index_t Y = filter_spatial_lengths[1];
-            const index_t X = filter_spatial_lengths[2];
-
-            const index_t ConvDilationD = conv_filter_dilations[0];
-            const index_t ConvDilationH = conv_filter_dilations[1];
-            const index_t ConvDilationW = conv_filter_dilations[2];
-
-            const index_t InLeftPadD = input_left_pads[0];
-            const index_t InLeftPadH = input_left_pads[1];
-            const index_t InLeftPadW = input_left_pads[2];
-
-            const index_t InRightPadD = input_right_pads[0];
-            const index_t InRightPadH = input_right_pads[1];
-            const index_t InRightPadW = input_right_pads[2];
-
-            const auto in_n_di_hi_wi_c_grid_desc =
-                make_naive_tensor_descriptor_packed(make_tuple(N, Di, Hi, Wi, C));
-
-            const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
-                in_n_di_hi_wi_c_grid_desc,
-                make_tuple(make_pass_through_transform(N),
-                           make_pad_transform(Di, InLeftPadD, InRightPadD),
-                           make_pad_transform(Hi, InLeftPadH, InRightPadH),
-                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
-                           make_pass_through_transform(C)),
-                make_tuple(
-                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
-                make_tuple(
-                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}));
-
-            const auto in_n_z_do_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
-                in_n_hip_wip_c_grid_desc,
-                make_tuple(
-                    make_pass_through_transform(N),
-                    make_embed_transform(make_tuple(Z, Do), make_tuple(ConvDilationD, ConvStrideD)),
-                    make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
-                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
-                    make_pass_through_transform(C)),
-                make_tuple(
-                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
-                make_tuple(Sequence<0>{},
-                           Sequence<1, 2>{},
-                           Sequence<3, 4>{},
-                           Sequence<5, 6>{},
-                           Sequence<7>{}));
-
-            const auto in_gemmk_gemmmraw_grid_desc = transform_tensor_descriptor(
-                in_n_z_do_y_ho_x_wo_c_grid_desc,
-                make_tuple(make_merge_transform(make_tuple(Z, Y, X, C)),
-                           make_merge_transform(make_tuple(N, Do, Ho, Wo))),
-                make_tuple(Sequence<1, 3, 5, 7>{}, Sequence<0, 2, 4, 6>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto in_gemmk0_gemmmraw_gemmk1_grid_desc = transform_tensor_descriptor(
-                in_gemmk_gemmmraw_grid_desc,
-                make_tuple(make_unmerge_transform(make_tuple(gemm_k0, GemmK1Number)),
-                           make_pass_through_transform(gemm_m)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            // in_gemmk0_gemmm_gemmk1_grid_desc
-            return transform_tensor_descriptor(
-                in_gemmk0_gemmmraw_gemmk1_grid_desc,
-                make_tuple(make_pass_through_transform(gemm_k0),
-                           make_right_pad_transform(gemm_m, gemm_m_pad),
-                           make_pass_through_transform(GemmK1Number)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
-        }
-    }
-
-    static index_t GetGemmMRaw(ck::index_t N,
-                               const std::vector<ck::index_t>& output_spatial_lengths)
-    {
-        return N * std::accumulate(std::begin(output_spatial_lengths),
-                                   std::end(output_spatial_lengths),
-                                   1,
-                                   std::multiplies<ck::index_t>());
-    }
-
-    static index_t GetGemmK(ck::index_t C, const std::vector<ck::index_t>& filter_spatial_lengths)
-    {
-        return C * std::accumulate(std::begin(filter_spatial_lengths),
-                                   std::end(filter_spatial_lengths),
-                                   1,
-                                   std::multiplies<ck::index_t>());
-    }
-
-    static auto
-    MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(ck::index_t N,
-                                                    ck::index_t K,
-                                                    ck::index_t C,
-                                                    std::vector<ck::index_t> input_spatial_lengths,
-                                                    std::vector<ck::index_t> filter_spatial_lengths,
-                                                    std::vector<ck::index_t> output_spatial_lengths,
-                                                    std::vector<ck::index_t> conv_filter_strides,
-                                                    std::vector<ck::index_t> conv_filter_dilations,
-                                                    std::vector<ck::index_t> input_left_pads,
-                                                    std::vector<ck::index_t> input_right_pads)
-    {
-        using namespace ck;
-
-        const index_t GemmMRaw = GetGemmMRaw(N, output_spatial_lengths);
-        const index_t GemmN    = K;
-        const index_t GemmK    = GetGemmK(C, filter_spatial_lengths);
-
-        const auto GemmMPad = math::integer_least_multiple(GemmMRaw, MPerBlock) - GemmMRaw;
-
-        assert(GemmK % GemmK1Number == 0);
-
-        // C = A^T*B
-        // A:
-        const auto in_gemmk0_gemmm_gemmk1_grid_desc =
-            GetInputTensorDescriptor<NumDimSpatial>(N,
-                                                    C,
-                                                    GemmMRaw,
-                                                    GemmK,
-                                                    GemmMPad,
-                                                    input_spatial_lengths,
-                                                    filter_spatial_lengths,
-                                                    output_spatial_lengths,
-                                                    conv_filter_strides,
-                                                    conv_filter_dilations,
-                                                    input_left_pads,
-                                                    input_right_pads);
-        // B:
-        const auto wei_gemmk0_gemmn_gemmk1_grid_desc = GetWeightTensorDescriptor(GemmN, GemmK);
-        // C:
-        const auto out_gemmm_gemmn_grid_desc = GetOutputTensorDescriptor(GemmMRaw, GemmN, GemmMPad);
-
-        return make_tuple(in_gemmk0_gemmm_gemmk1_grid_desc,
-                          wei_gemmk0_gemmn_gemmk1_grid_desc,
-                          out_gemmm_gemmn_grid_desc);
-    }
-
-    template <ck::index_t NDim, typename std::enable_if<NDim == 1, bool>::type = false>
-    static auto GetABCGridDesc()
-    {
-        return MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(
-            1, 1, 1, {1}, {1}, {1}, {1}, {1}, {1}, {1});
-    }
-
-    template <ck::index_t NDim, typename std::enable_if<NDim == 2, bool>::type = false>
-    static auto GetABCGridDesc()
-    {
-        return MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(
-            1, 1, 1, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1});
-    }
-
-    template <ck::index_t NDim, typename std::enable_if<NDim == 3, bool>::type = false>
-    static auto GetABCGridDesc()
-    {
-        return MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(
-            1, 1, 1, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1});
-    }
-
-    using ABCGridDescs = decltype(GetABCGridDesc<NumDimSpatial>());
-
-    using AGridDesc_K0_M_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I0])>;
-    using BGridDesc_K0_N_K1 = remove_cvref_t<decltype(ABCGridDescs{}[I1])>;
-    using CGridDesc_M_N     = remove_cvref_t<decltype(ABCGridDescs{}[I2])>;
-
-    using Block2CTileMap = BlockToCTileMap_M00_N0_M01<MPerBlock, NPerBlock, CGridDesc_M_N>;
-
-    // GridwiseGemm
-    using GridwiseGemm = GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3<
-        BlockSize,
-        ABDataType, // TODO: distinguish A/B datatype
-        AccDataType,
-        CDataType,
-        InMemoryDataOperationEnum::Set,
-        AGridDesc_K0_M_K1,
-        BGridDesc_K0_N_K1,
-        CGridDesc_M_N,
-        InElementwiseOperation,
-        WeiElementwiseOperation,
-        OutElementwiseOperation,
-        MPerBlock,
-        NPerBlock,
-        K0PerBlock,
-        MPerXDL,
-        NPerXDL,
-        K1,
-        MXdlPerWave,
-        NXdlPerWave,
-        ABlockTransferThreadClusterLengths_K0_M_K1,
-        Sequence<1, 0, 2>, // ABlockTransferThreadClusterArrangeOrder,
-        Sequence<1, 0, 2>, // ABlockTransferSrcAccessOrder,
-        2,                 // ABlockTransferSrcVectorDim,
-        ABlockTransferSrcScalarPerVector,
-        ABlockTransferDstScalarPerVector_K1,
-        false, // AThreadTransferSrcResetCoordinateAfterRun,
-        ABlockLdsAddExtraM,
-        BBlockTransferThreadClusterLengths_K0_N_K1,
-        Sequence<1, 0, 2>, // BBlockTransferThreadClusterArrangeOrder,
-        Sequence<1, 0, 2>, // BBlockTransferSrcAccessOrder,
-        2,                 // BBlockTransferSrcVectorDim,
-        BBlockTransferSrcScalarPerVector,
-        BBlockTransferDstScalarPerVector_K1,
-        false, // BThreadTransferSrcResetCoordinateAfterRun,
-        BBlockLdsAddExtraN,
-        Sequence<2, 3, 0, 1, 7, 5, 4, 6>, // CThreadTransferSrcDstAccessOrder,
-        7,                                // CThreadTransferSrcDstVectorDim,
-        CThreadTransferDstScalarPerVector>;
-
-    // Argument
-    struct Argument : public BaseArgument
-    {
-        Argument(const InDataType* p_in_grid,
-                 const WeiDataType* p_wei_grid,
-                 OutDataType* p_out_grid,
-                 ck::index_t N,
-                 ck::index_t K,
-                 ck::index_t C,
-                 std::vector<ck::index_t> input_spatial_lengths,
-                 std::vector<ck::index_t> filter_spatial_lengths,
-                 std::vector<ck::index_t> output_spatial_lengths,
-                 std::vector<ck::index_t> conv_filter_strides,
-                 std::vector<ck::index_t> conv_filter_dilations,
-                 std::vector<ck::index_t> input_left_pads,
-                 std::vector<ck::index_t> input_right_pads,
-                 InElementwiseOperation in_element_op,
-                 WeiElementwiseOperation wei_element_op,
-                 OutElementwiseOperation out_element_op)
-            : p_a_grid_{p_in_grid},
-              p_b_grid_{p_wei_grid},
-              p_c_grid_{p_out_grid},
-              a_grid_desc_k0_m_k1_{},
-              b_grid_desc_k0_n_k1_{},
-              c_grid_desc_m_n_{},
-              c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_{},
-              block_2_ctile_map_{},
-              in_element_op_{in_element_op},
-              wei_element_op_{wei_element_op},
-              out_element_op_{out_element_op},
-              Conv_N_{N},
-              Conv_K_{K},
-              Conv_C_{C},
-              filter_spatial_lengths_{filter_spatial_lengths},
-              conv_filter_strides_{conv_filter_strides},
-              input_left_pads_{input_left_pads},
-              input_right_pads_{input_right_pads}
-        {
-            const auto descs =
-                DeviceOp::MakeABCGridDescriptor_A_K0_M_K1_B_K0_N_K1_C_M_N(N,
-                                                                          K,
-                                                                          C,
-                                                                          input_spatial_lengths,
-                                                                          filter_spatial_lengths,
-                                                                          output_spatial_lengths,
-                                                                          conv_filter_strides,
-                                                                          conv_filter_dilations,
-                                                                          input_left_pads,
-                                                                          input_right_pads);
-
-            a_grid_desc_k0_m_k1_ = descs[I0];
-            b_grid_desc_k0_n_k1_ = descs[I1];
-            c_grid_desc_m_n_     = descs[I2];
-
-            block_2_ctile_map_ = Block2CTileMap{c_grid_desc_m_n_};
-
-            if(GridwiseGemm::CheckValidity(a_grid_desc_k0_m_k1_,
-                                           b_grid_desc_k0_n_k1_,
-                                           c_grid_desc_m_n_,
-                                           block_2_ctile_map_))
-            {
-                c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_ =
-                    GridwiseGemm::MakeCGridDescriptor_M0_N0_M1_N1_M2_M3_M4_N2(c_grid_desc_m_n_);
-            }
-        }
-
-        //  private:
-        const ADataType* p_a_grid_;
-        const BDataType* p_b_grid_;
-        CDataType* p_c_grid_;
-        AGridDesc_K0_M_K1 a_grid_desc_k0_m_k1_;
-        BGridDesc_K0_N_K1 b_grid_desc_k0_n_k1_;
-        CGridDesc_M_N c_grid_desc_m_n_;
-        typename GridwiseGemm::CGridDesc_M0_N0_M1_N1_M2_M3_M4_N2
-            c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_;
-        Block2CTileMap block_2_ctile_map_;
-        InElementwiseOperation in_element_op_;
-        WeiElementwiseOperation wei_element_op_;
-        OutElementwiseOperation out_element_op_;
-        // for checking IsSupportedArgument()
-        index_t Conv_N_;
-        index_t Conv_K_;
-        index_t Conv_C_;
-        std::vector<index_t> filter_spatial_lengths_;
-        std::vector<index_t> conv_filter_strides_;
-        std::vector<index_t> input_left_pads_;
-        std::vector<index_t> input_right_pads_;
-    };
-
-    // Invoker
-    struct Invoker : public BaseInvoker
-    {
-        using Argument = DeviceOp::Argument;
-
-        float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
-        {
-#if 0
-            {
-                std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
-                          << ", " << arg.a_grid_desc_k0_m_k1_.GetLength(I1) << ", "
-                          << arg.a_grid_desc_k0_m_k1_.GetLength(I2) << "}" << std::endl;
-
-                std::cout << "arg.b_grid_desc_k0_n_k1_{" << arg.b_grid_desc_k0_n_k1_.GetLength(I0)
-                          << ", " << arg.b_grid_desc_k0_n_k1_.GetLength(I1) << ", "
-                          << arg.b_grid_desc_k0_n_k1_.GetLength(I2) << "}" << std::endl;
-
-                std::cout << "arg.c_grid_desc_m_n_{ " << arg.c_grid_desc_m_n_.GetLength(I0) << ", "
-                          << arg.c_grid_desc_m_n_.GetLength(I1) << "}" << std::endl;
-            }
-#endif
-            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_k0_m_k1_,
-                                            arg.b_grid_desc_k0_n_k1_,
-                                            arg.c_grid_desc_m_n_,
-                                            arg.block_2_ctile_map_))
-            {
-                throw std::runtime_error(
-                    "wrong! GridwiseGemm_km_kn_m0m1n0n1_xdlops_v2r3 has invalid setting");
-            }
-
-            const index_t grid_size =
-                arg.block_2_ctile_map_.CalculateGridSize(arg.c_grid_desc_m_n_);
-
-            const auto K =
-                arg.a_grid_desc_k0_m_k1_.GetLength(I0) * arg.a_grid_desc_k0_m_k1_.GetLength(I2);
-
-            float ave_time = 0;
-
-            if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
-            {
-                const auto kernel = kernel_gemm_xdlops_v2r3<
-                    GridwiseGemm,
-                    ADataType, // TODO: distiguish A/B datatype
-                    CDataType,
-                    remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
-                    remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
-                    remove_reference_t<typename GridwiseGemm::CGridDesc_M0_N0_M1_N1_M2_M3_M4_N2>,
-                    InElementwiseOperation,
-                    WeiElementwiseOperation,
-                    OutElementwiseOperation,
-                    Block2CTileMap,
-                    true>;
-
-                ave_time = launch_and_time_kernel(stream_config,
-                                                  kernel,
-                                                  dim3(grid_size),
-                                                  dim3(BlockSize),
-                                                  0,
-                                                  arg.p_a_grid_,
-                                                  arg.p_b_grid_,
-                                                  arg.p_c_grid_,
-                                                  arg.a_grid_desc_k0_m_k1_,
-                                                  arg.b_grid_desc_k0_n_k1_,
-                                                  arg.c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_,
-                                                  arg.in_element_op_,
-                                                  arg.wei_element_op_,
-                                                  arg.out_element_op_,
-                                                  arg.block_2_ctile_map_);
-            }
-            else
-            {
-                const auto kernel = kernel_gemm_xdlops_v2r3<
-                    GridwiseGemm,
-                    ADataType, // TODO: distiguish A/B datatype
-                    CDataType,
-                    remove_reference_t<DeviceOp::AGridDesc_K0_M_K1>,
-                    remove_reference_t<DeviceOp::BGridDesc_K0_N_K1>,
-                    remove_reference_t<typename GridwiseGemm::CGridDesc_M0_N0_M1_N1_M2_M3_M4_N2>,
-                    InElementwiseOperation,
-                    WeiElementwiseOperation,
-                    OutElementwiseOperation,
-                    Block2CTileMap,
-                    false>;
-
-                ave_time = launch_and_time_kernel(stream_config,
-                                                  kernel,
-                                                  dim3(grid_size),
-                                                  dim3(BlockSize),
-                                                  0,
-                                                  arg.p_a_grid_,
-                                                  arg.p_b_grid_,
-                                                  arg.p_c_grid_,
-                                                  arg.a_grid_desc_k0_m_k1_,
-                                                  arg.b_grid_desc_k0_n_k1_,
-                                                  arg.c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2_,
-                                                  arg.in_element_op_,
-                                                  arg.wei_element_op_,
-                                                  arg.out_element_op_,
-                                                  arg.block_2_ctile_map_);
-            }
-
-            return ave_time;
-        }
-
-        float Run(const BaseArgument* p_arg,
-                  const StreamConfig& stream_config = StreamConfig{}) override
-        {
-            return Run(*dynamic_cast<const Argument*>(p_arg), stream_config);
-        }
-    };
-
-    static constexpr bool IsValidCompilationParameter()
-    {
-        // TODO: properly implement this check
-        return true;
-    }
-
-    static bool IsSupportedArgument(const Argument& arg)
-    {
-        if(ck::get_device_name() == "gfx908")
-        {
-            if constexpr(!(is_same_v<AccDataType, float> || is_same_v<AccDataType, float> ||
-                           is_same_v<AccDataType, int32_t>))
-            {
-                return false;
-            }
-        }
-        else if(ck::get_device_name() == "gfx90a")
-        {
-            if constexpr(!(is_same_v<AccDataType, float> || is_same_v<AccDataType, float> ||
-                           is_same_v<AccDataType, int32_t> || is_same_v<AccDataType, double>))
-            {
-                return false;
-            }
-        }
-        else
-        {
-            return false;
-        }
-
-        // Input tensors can't be bigger than 2GB each.
-        constexpr ck::long_index_t GB2 = (ck::long_index_t{1} << 31);
-
-        if(arg.a_grid_desc_k0_m_k1_.GetElementSpaceSize() * sizeof(ADataType) > GB2 ||
-           arg.b_grid_desc_k0_n_k1_.GetElementSpaceSize() * sizeof(BDataType) > GB2 ||
-           arg.c_grid_desc_m_n_.GetElementSpaceSize() * sizeof(CDataType) > GB2)
-        {
-            return false;
-        }
-
-        if constexpr(ConvForwardSpecialization ==
-                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
-        {
-            // check if it's 1x1, stride=1 conv
-            for(ck::index_t i = 0; i < NumDimSpatial; ++i)
-            {
-                if(!(arg.filter_spatial_lengths_[i] == 1 && arg.conv_filter_strides_[i] == 1 &&
-                     arg.input_left_pads_[i] == 0 && arg.input_right_pads_[i] == 0))
-                {
-                    return false;
-                }
-            }
-        }
-        else if constexpr(ConvForwardSpecialization ==
-                          ConvolutionForwardSpecialization::Filter1x1Pad0)
-        {
-            // check if it's 1x1 conv
-            for(ck::index_t i = 0; i < NumDimSpatial; ++i)
-            {
-                if(!(arg.filter_spatial_lengths_[i] == 1 && arg.input_left_pads_[i] == 0 &&
-                     arg.input_right_pads_[i] == 0))
-                {
-                    return false;
-                }
-            }
-        }
-
-        // vector load A/B matrix from global memory
-        if(!(ABlockTransferSrcVectorDim == 2 && BBlockTransferSrcVectorDim == 2 &&
-             arg.Conv_C_ % ABlockTransferSrcScalarPerVector == 0 &&
-             arg.Conv_C_ % BBlockTransferSrcScalarPerVector == 0))
-        {
-            return false;
-        }
-
-        // vector store C matrix into global memory
-        if(!(arg.Conv_K_ % CThreadTransferDstScalarPerVector == 0))
-        {
-            return false;
-        }
-
-        // Gridwise GEMM size
-        return GridwiseGemm::CheckValidity(arg.a_grid_desc_k0_m_k1_,
-                                           arg.b_grid_desc_k0_n_k1_,
-                                           arg.c_grid_desc_m_n_,
-                                           arg.block_2_ctile_map_);
-    }
-
-    bool IsSupportedArgument(const BaseArgument* p_arg) override
-    {
-        return IsSupportedArgument(*dynamic_cast<const Argument*>(p_arg));
-    }
-
-    static auto MakeArgument(const InDataType* p_in_grid,
-                             const WeiDataType* p_wei_grid,
-                             OutDataType* p_out_grid,
-                             ck::index_t N,
-                             ck::index_t K,
-                             ck::index_t C,
-                             std::vector<ck::index_t> input_spatial_lengths,
-                             std::vector<ck::index_t> filter_spatial_lengths,
-                             std::vector<ck::index_t> output_spatial_lengths,
-                             std::vector<ck::index_t> conv_filter_strides,
-                             std::vector<ck::index_t> conv_filter_dilations,
-                             std::vector<ck::index_t> input_left_pads,
-                             std::vector<ck::index_t> input_right_pads,
-                             InElementwiseOperation in_element_op,
-                             WeiElementwiseOperation wei_element_op,
-                             OutElementwiseOperation out_element_op)
-    {
-        return Argument{p_in_grid,
-                        p_wei_grid,
-                        p_out_grid,
-                        N,
-                        K,
-                        C,
-                        input_spatial_lengths,
-                        filter_spatial_lengths,
-                        output_spatial_lengths,
-                        conv_filter_strides,
-                        conv_filter_dilations,
-                        input_left_pads,
-                        input_right_pads,
-                        in_element_op,
-                        wei_element_op,
-                        out_element_op};
-    }
-
-    static auto MakeInvoker() { return Invoker{}; }
-
-    std::unique_ptr<BaseArgument>
-    MakeArgumentPointer(const void* p_in_grid,
-                        const void* p_wei_grid,
-                        void* p_out_grid,
-                        ck::index_t N,
-                        ck::index_t K,
-                        ck::index_t C,
-                        std::vector<ck::index_t> input_spatial_lengths,
-                        std::vector<ck::index_t> filter_spatial_lengths,
-                        std::vector<ck::index_t> output_spatial_lengths,
-                        std::vector<ck::index_t> conv_filter_strides,
-                        std::vector<ck::index_t> conv_filter_dilations,
-                        std::vector<ck::index_t> input_left_pads,
-                        std::vector<ck::index_t> input_right_pads,
-                        InElementwiseOperation in_element_op,
-                        WeiElementwiseOperation wei_element_op,
-                        OutElementwiseOperation out_element_op) override
-    {
-        return std::make_unique<Argument>(static_cast<const InDataType*>(p_in_grid),
-                                          static_cast<const WeiDataType*>(p_wei_grid),
-                                          static_cast<OutDataType*>(p_out_grid),
-                                          N,
-                                          K,
-                                          C,
-                                          input_spatial_lengths,
-                                          filter_spatial_lengths,
-                                          output_spatial_lengths,
-                                          conv_filter_strides,
-                                          conv_filter_dilations,
-                                          input_left_pads,
-                                          input_right_pads,
-                                          in_element_op,
-                                          wei_element_op,
-                                          out_element_op);
-    }
-
-    std::unique_ptr<BaseInvoker> MakeInvokerPointer() override
-    {
-        return std::make_unique<Invoker>(Invoker{});
-    }
-
-    std::string GetTypeString() const override
-    {
-        auto str = std::stringstream();
-
-        // clang-format off
-        str << "DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K"
-            << "<"
-            << BlockSize << ", "
-            << MPerBlock << ", "
-            << NPerBlock << ", "
-            << K0PerBlock << ", "
-            << getConvForwardSpecializationString(ConvForwardSpecialization)
-            << ">";
-        // clang-format on
-
-        return str.str();
-    }
-};
-
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm.hpp b/include/ck/tensor_operation/gpu/device/device_gemm.hpp
index 731309c50b..1781456a5c 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm.hpp
@@ -40,25 +40,6 @@ struct DeviceGemm : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename ALayout,
-          typename BLayout,
-          typename CLayout,
-          typename ADataType,
-          typename BDataType,
-          typename CDataType,
-          typename AElementwiseOperation,
-          typename BElementwiseOperation,
-          typename CElementwiseOperation>
-using DeviceGemmPtr = std::unique_ptr<DeviceGemm<ALayout,
-                                                 BLayout,
-                                                 CLayout,
-                                                 ADataType,
-                                                 BDataType,
-                                                 CDataType,
-                                                 AElementwiseOperation,
-                                                 BElementwiseOperation,
-                                                 CElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp
index 1aa3885523..b9a64e8c4b 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_bias_add_reduce_xdl_cshuffle.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_reduce.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_bias_add_reduce_xdl_cshuffle_v1.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_bias_c_permute.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_bias_e_permute.hpp
similarity index 84%
rename from include/ck/tensor_operation/gpu/device/device_gemm_bias_c_permute.hpp
rename to include/ck/tensor_operation/gpu/device/device_gemm_bias_e_permute.hpp
index bde0d48c15..4c2161eaed 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_bias_c_permute.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_bias_e_permute.hpp
@@ -46,12 +46,6 @@ struct DeviceGemmBiasCPermute : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename AElementwiseOperation,
-          typename BElementwiseOperation,
-          typename CElementwiseOperation>
-using DeviceGemmBiasCPermutePtr = std::unique_ptr<
-    DeviceGemmBiasCPermute<AElementwiseOperation, BElementwiseOperation, CElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_bias_c_permute_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_bias_e_permute_xdl.hpp
similarity index 59%
rename from include/ck/tensor_operation/gpu/device/device_gemm_bias_c_permute_xdl.hpp
rename to include/ck/tensor_operation/gpu/device/device_gemm_bias_e_permute_xdl.hpp
index f74cb0dc84..ffdb8d5894 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_bias_c_permute_xdl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_bias_e_permute_xdl.hpp
@@ -10,11 +10,12 @@
 #include "ck/tensor_description/tensor_descriptor.hpp"
 #include "ck/tensor_description/tensor_descriptor_helper.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_bias_c_permute.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_bias_e_permute.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 
@@ -35,7 +36,7 @@ __global__ void
 #if CK_USE_LAUNCH_BOUNDS
     __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU)
 #endif
-        kernel_gemm_bias_c_permute(const FloatAB* __restrict__ p_a_grid,
+        kernel_gemm_bias_e_permute(const FloatAB* __restrict__ p_a_grid,
                                    const FloatAB* __restrict__ p_b_grid,
                                    FloatDsPointer p_ds_grid,
                                    FloatE* __restrict__ p_e_grid,
@@ -99,7 +100,7 @@ template <typename ALayout,
           typename CDELayout,
           typename ADataType,
           typename BDataType,
-          typename GemmAccDataType,
+          typename AccDataType,
           typename CShuffleDataType,
           typename DDataType,
           typename EDataType,
@@ -124,33 +125,36 @@ template <typename ALayout,
           index_t ABlockTransferSrcVectorDim,
           index_t ABlockTransferSrcScalarPerVector,
           index_t ABlockTransferDstScalarPerVector_AK1,
-          bool ABlockLdsExtraM,
+          index_t ABlockLdsExtraM,
           typename BBlockTransferThreadClusterLengths_BK0_N_BK1,
           typename BBlockTransferThreadClusterArrangeOrder,
           typename BBlockTransferSrcAccessOrder,
           index_t BBlockTransferSrcVectorDim,
           index_t BBlockTransferSrcScalarPerVector,
           index_t BBlockTransferDstScalarPerVector_BK1,
-          bool BBlockLdsExtraN,
+          index_t BBlockLdsExtraN,
           index_t CShuffleMXdlPerWavePerShuffle,
           index_t CShuffleNXdlPerWavePerShuffle,
           typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CDEBlockTransferScalarPerVector_NPerBlock,
           LoopScheduler LoopSched = make_default_loop_scheduler()>
-struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOperation,
+struct DeviceGemmBiasEPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOperation,
                                                                   BElementwiseOperation,
                                                                   CDEElementwiseOperation>
 {
-    using DeviceOp = DeviceGemmBiasCPermute_Xdl;
+    using DeviceOp = DeviceGemmBiasEPermute_Xdl;
 
     static constexpr auto I0 = Number<0>{};
     static constexpr auto I1 = Number<1>{};
     static constexpr auto I2 = Number<2>{};
     static constexpr auto I3 = Number<3>{};
 
-    static constexpr index_t NumDTensor = I1;
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
 
-    static auto MakeAGridDescriptor_AK0_M_AK1(index_t MRaw, index_t KRaw, index_t StrideA)
+    static constexpr index_t NumDTensor = 1;
+
+    static auto MakeAGridDescriptor_M_K(index_t MRaw, index_t KRaw, index_t StrideA)
     {
         const auto a_grid_desc_mraw_kraw = [&]() {
             if constexpr(is_same_v<tensor_layout::gemm::RowMajor, ALayout>)
@@ -165,95 +169,10 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both M and K
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(M)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad M, but not K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_right_pad_transform(MRaw, MPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad K, but not M
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k = transform_tensor_descriptor(
-                a_grid_desc_mraw_kraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else
-        {
-            // not pad M or K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
+        return matrix_padder.PadADescriptor_M_K(a_grid_desc_mraw_kraw);
     }
 
-    static auto MakeBGridDescriptor_BK0_N_BK1(index_t KRaw, index_t NRaw, index_t StrideB)
+    static auto MakeBGridDescriptor_N_K(index_t KRaw, index_t NRaw, index_t StrideB)
     {
         const auto b_grid_desc_nraw_kraw = [&]() {
             if constexpr(is_same<tensor_layout::gemm::RowMajor, BLayout>::value)
@@ -268,92 +187,7 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
             }
         }();
 
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto NPad = N - NRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::NKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both N and K
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_right_pad_transform(NRaw, NPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(N)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad N, but not K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_right_pad_transform(NRaw, NPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad K, but not N
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k = transform_tensor_descriptor(
-                b_grid_desc_nraw_kraw,
-                make_tuple(make_pass_through_transform(NRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else
-        {
-            // not pad N or K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
+        return matrix_padder.PadBDescriptor_N_K(b_grid_desc_nraw_kraw);
     }
 
     static auto MakeEGridDescriptor_M_N(DEGridDesc_M0_M1_M2_N0_N1 d_e_grid_desc)
@@ -370,73 +204,32 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
         index_t stride_N0 = d_e_grid_desc.stride_N0_;
         index_t stride_N1 = d_e_grid_desc.stride_N1_;
 
-        const auto MRaw = M0 * M1 * M2;
-        const auto NRaw = N0 * N1;
-
-        const auto c_grid_desc_mraw_nraw = [&]() {
-            const auto c_grid_desc_m0_m1_m2_n0_n1 = make_naive_tensor_descriptor(
+        const auto e_grid_desc_mraw_nraw = [&]() {
+            const auto e_grid_desc_m0_m1_m2_n0_n1 = make_naive_tensor_descriptor(
                 make_tuple(M0, M1, M2, N0, N1),
                 make_tuple(stride_M0, stride_M1, stride_M2, stride_N0, stride_N1));
 
             return transform_tensor_descriptor(
-                c_grid_desc_m0_m1_m2_n0_n1,
+                e_grid_desc_m0_m1_m2_n0_n1,
                 make_tuple(make_merge_transform(make_tuple(M0, M1, M2)),
                            make_merge_transform(make_tuple(N0, N1))),
                 make_tuple(Sequence<0, 1, 2>{}, Sequence<3, 4>{}),
                 make_tuple(Sequence<0>{}, Sequence<1>{}));
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto NPad = N - NRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad M and N
-            return transform_tensor_descriptor(c_grid_desc_mraw_nraw,
-                                               make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                          make_right_pad_transform(NRaw, NPad)),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad M, but not N
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(NRaw)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad N, but not M
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(NRaw, NPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else
-        {
-            // not pad M or N
-            return c_grid_desc_mraw_nraw;
-        }
+        return matrix_padder.PadCDescriptor_M_N(e_grid_desc_mraw_nraw);
     }
 
-    using AGridDesc_AK0_M_AK1 = decltype(MakeAGridDescriptor_AK0_M_AK1(1, 1, 1));
-    using BGridDesc_BK0_N_BK1 = decltype(MakeBGridDescriptor_BK0_N_BK1(1, 1, 1));
-    using EGridDesc_M_N       = decltype(MakeEGridDescriptor_M_N(DEGridDesc_M0_M1_M2_N0_N1{}));
+    using AGridDesc_M_K = decltype(MakeAGridDescriptor_M_K(1, 1, 1));
+    using BGridDesc_N_K = decltype(MakeBGridDescriptor_N_K(1, 1, 1));
+    using EGridDesc_M_N = decltype(MakeEGridDescriptor_M_N(DEGridDesc_M0_M1_M2_N0_N1{}));
+
+    using DsGridDesc_M_N = Tuple<EGridDesc_M_N>;
 
     // GridwiseGemm
-    using GridwiseGemm = GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle<
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
         ADataType, // TODO: distinguish A/B datatype
-        GemmAccDataType,
+        AccDataType,
         CShuffleDataType,
         ck::Tuple<DDataType>,
         EDataType,
@@ -444,8 +237,9 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
         BElementwiseOperation,
         CDEElementwiseOperation,
         InMemoryDataOperationEnum::Set,
-        AGridDesc_AK0_M_AK1,
-        BGridDesc_BK0_N_BK1,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        DsGridDesc_M_N,
         EGridDesc_M_N,
         NumGemmKPrefetchStage,
         BlockSize,
@@ -480,6 +274,13 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
         CDEBlockTransferScalarPerVector_NPerBlock,
         LoopSched>;
 
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
+    using Block2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+
     // Argument
     struct Argument : public BaseArgument
     {
@@ -499,12 +300,17 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
                  CDEElementwiseOperation cde_element_op)
             : p_a_grid_{static_cast<const ADataType*>(p_a_grid)},
               p_b_grid_{static_cast<const BDataType*>(p_b_grid)},
-              p_ds_grid_{}, // FIXME
+              p_ds_grid_{},
               p_e_grid_{static_cast<EDataType*>(p_e_grid)},
-              a_grid_desc_ak0_m_ak1_{DeviceOp::MakeAGridDescriptor_AK0_M_AK1(MRaw, KRaw, StrideA)},
-              b_grid_desc_bk0_n_bk1_{DeviceOp::MakeBGridDescriptor_BK0_N_BK1(KRaw, NRaw, StrideB)},
-              ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
+              a_grid_desc_m_k_{DeviceOp::MakeAGridDescriptor_M_K(MRaw, KRaw, StrideA)},
+              b_grid_desc_n_k_{DeviceOp::MakeBGridDescriptor_N_K(KRaw, NRaw, StrideB)},
+              ds_grid_desc_m_n_{},
               e_grid_desc_m_n_{DeviceOp::MakeEGridDescriptor_M_N(e_grid_desc)},
+              a_grid_desc_ak0_m_ak1_{
+                  GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k_)},
+              b_grid_desc_bk0_n_bk1_{
+                  GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k_)},
+              ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
               e_grid_desc_mblock_mperblock_nblock_nperblock_{},
               block_2_etile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
               a_element_op_{a_element_op},
@@ -522,8 +328,16 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
                 throw std::runtime_error("wrong! GridwiseGemm has invalid setting");
             }
 
-            if(GridwiseGemm::CheckValidity(a_grid_desc_ak0_m_ak1_,
-                                           b_grid_desc_bk0_n_bk1_,
+            // populate pointer, desc for Ds
+            // D pointer
+            p_ds_grid_(I0) = static_cast<const DDataType*>(p_d_grid);
+
+            // D desc
+            ds_grid_desc_m_n_(I0) = DeviceOp::MakeEGridDescriptor_M_N(d_grid_desc);
+
+            if(GridwiseGemm::CheckValidity(a_grid_desc_m_k_,
+                                           b_grid_desc_n_k_,
+                                           ds_grid_desc_m_n_,
                                            e_grid_desc_m_n_,
                                            block_2_etile_map_))
             {
@@ -531,32 +345,37 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
                     GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
                         e_grid_desc_m_n_);
 
-                p_ds_grid_(I0) = static_cast<const DDataType*>(p_d_grid);
-
-                const auto d_grid_desc_m_n = DeviceOp::MakeEGridDescriptor_M_N(d_grid_desc);
-
                 ds_grid_desc_mblock_mperblock_nblock_nperblock_(I0) =
                     GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
-                        d_grid_desc_m_n);
+                        ds_grid_desc_m_n_[I0]);
             }
         }
 
         //  private:
+        // pointers
         const ADataType* p_a_grid_;
         const BDataType* p_b_grid_;
         typename GridwiseGemm::DsGridPointer p_ds_grid_;
         EDataType* p_e_grid_;
+
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        DsGridDesc_M_N ds_grid_desc_m_n_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
         AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
         BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
-        StaticallyIndexedArray<
-            typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-            NumDTensor>
-            ds_grid_desc_mblock_mperblock_nblock_nperblock_; // FIXME: Ds desc may be of different
-                                                             // type from E
-        EGridDesc_M_N e_grid_desc_m_n_;
+        typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            ds_grid_desc_mblock_mperblock_nblock_nperblock_;
         typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
             e_grid_desc_mblock_mperblock_nblock_nperblock_;
-        typename GridwiseGemm::DefaultBlock2ETileMap block_2_etile_map_;
+
+        // block-to-e-tile map
+        Block2ETileMap block_2_etile_map_;
+
+        // element-wise op
         AElementwiseOperation a_element_op_;
         BElementwiseOperation b_element_op_;
         CDEElementwiseOperation cde_element_op_;
@@ -569,8 +388,9 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                            arg.b_grid_desc_bk0_n_bk1_,
+            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                            arg.b_grid_desc_n_k_,
+                                            arg.ds_grid_desc_m_n_,
                                             arg.e_grid_desc_m_n_,
                                             arg.block_2_etile_map_))
             {
@@ -586,7 +406,7 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
             auto launch_kernel = [&](auto has_main_k_block_loop) {
                 constexpr bool has_main_loop = has_main_k_block_loop.value;
 
-                const auto kernel = kernel_gemm_bias_c_permute<
+                const auto kernel = kernel_gemm_bias_e_permute<
                     GridwiseGemm,
                     ADataType, // TODO: distiguish A/B datatype
                     typename GridwiseGemm::DsGridPointer,
@@ -596,9 +416,7 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
                     CDEElementwiseOperation,
                     DeviceOp::AGridDesc_AK0_M_AK1,
                     DeviceOp::BGridDesc_BK0_N_BK1,
-                    ck::StaticallyIndexedArray<
-                        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-                        NumDTensor>,
+                    typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     typename GridwiseGemm::DefaultBlock2ETileMap,
                     has_main_loop>;
@@ -622,18 +440,14 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
                                               arg.block_2_etile_map_);
             };
 
-            float ave_time = 0;
-
             if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
             {
-                ave_time = launch_kernel(integral_constant<bool, true>{});
+                return launch_kernel(integral_constant<bool, true>{});
             }
             else
             {
-                ave_time = launch_kernel(integral_constant<bool, false>{});
+                return launch_kernel(integral_constant<bool, false>{});
             }
-
-            return ave_time;
         }
 
         // polymorphic
@@ -651,8 +465,9 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
             return false;
         }
 
-        return GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                           arg.b_grid_desc_bk0_n_bk1_,
+        return GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                           arg.b_grid_desc_n_k_,
+                                           arg.ds_grid_desc_m_n_,
                                            arg.e_grid_desc_m_n_,
                                            arg.block_2_etile_map_);
     }
@@ -741,7 +556,7 @@ struct DeviceGemmBiasCPermute_Xdl : public DeviceGemmBiasCPermute<AElementwiseOp
         auto str = std::stringstream();
 
         // clang-format off
-        str << "DeviceGemmBiasCPermute_Xdl"
+        str << "DeviceGemmBiasEPermute_Xdl"
             << "<"
             << BlockSize << ", "
             << MPerBlock << ", "
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_dl.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_dl.hpp
index aca413f8a0..0d93e06418 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_dl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_dl.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_dl_v1r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d.hpp
index 9c0594e38c..d562042534 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d.hpp
@@ -21,7 +21,8 @@ namespace device {
 //   D0, D1, ... and E have the same layout
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename DsLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
           typename DsDataType,
@@ -52,27 +53,6 @@ struct DeviceGemmMultipleD : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename ALayout,
-          typename BLayout,
-          typename DELayout,
-          typename ADataType,
-          typename BDataType,
-          typename DsDataType,
-          typename EDataType,
-          typename AElementwiseOperation,
-          typename BElementwiseOperation,
-          typename CDEElementwiseOperation>
-using DeviceGemmMultipleDPtr = std::unique_ptr<DeviceGemmMultipleD<ALayout,
-                                                                   BLayout,
-                                                                   DELayout,
-                                                                   ADataType,
-                                                                   BDataType,
-                                                                   DsDataType,
-                                                                   EDataType,
-                                                                   AElementwiseOperation,
-                                                                   BElementwiseOperation,
-                                                                   CDEElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp
index 8c7f2c15f0..e81b30ecb1 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp
@@ -12,16 +12,17 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/device_gemm_multiple_d.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 
 template <typename GridwiseGemm,
-          typename FloatAB,
-          typename FloatDsPointer,
-          typename FloatE,
+          typename ABDataType,
+          typename DsPointer,
+          typename EDataType,
           typename AElementwiseOperation,
           typename BElementwiseOperation,
           typename CDEElementwiseOperation,
@@ -35,10 +36,10 @@ __global__ void
 #if CK_USE_LAUNCH_BOUNDS
     __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU)
 #endif
-        kernel_gemm_multiple_d_xdl_cshuffle(const FloatAB* __restrict__ p_a_grid,
-                                            const FloatAB* __restrict__ p_b_grid,
-                                            FloatDsPointer p_ds_grid,
-                                            FloatE* __restrict__ p_e_grid,
+        kernel_gemm_multiple_d_xdl_cshuffle(const ABDataType* __restrict__ p_a_grid,
+                                            const ABDataType* __restrict__ p_b_grid,
+                                            DsPointer p_ds_grid,
+                                            EDataType* __restrict__ p_e_grid,
                                             const AElementwiseOperation a_element_op,
                                             const BElementwiseOperation b_element_op,
                                             const CDEElementwiseOperation cde_element_op,
@@ -89,8 +90,8 @@ namespace tensor_operation {
 namespace device {
 
 // GEMM:
-//   input : A[AK0, M, AK1]
-//   input : B[AK0, N, AK1]
+//   input : A[M, K]
+//   input : B[N, K]
 //   input : D0[M, N], D1[M, N], ...
 //   output : E[M, N]
 //   C = a_op(A) * b_op(B)
@@ -99,10 +100,11 @@ namespace device {
 //   D0, D1, ... and E have the same layout
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename DsLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
-          typename GemmAccDataType,
+          typename AccDataType,
           typename CShuffleDataType,
           typename DsDataType,
           typename EDataType,
@@ -127,14 +129,14 @@ template <typename ALayout,
           index_t ABlockTransferSrcVectorDim,
           index_t ABlockTransferSrcScalarPerVector,
           index_t ABlockTransferDstScalarPerVector_AK1,
-          bool ABlockLdsExtraM,
+          index_t ABlockLdsExtraM,
           typename BBlockTransferThreadClusterLengths_BK0_N_BK1,
           typename BBlockTransferThreadClusterArrangeOrder,
           typename BBlockTransferSrcAccessOrder,
           index_t BBlockTransferSrcVectorDim,
           index_t BBlockTransferSrcScalarPerVector,
           index_t BBlockTransferDstScalarPerVector_BK1,
-          bool BBlockLdsExtraN,
+          index_t BBlockLdsExtraN,
           index_t CShuffleMXdlPerWavePerShuffle,
           index_t CShuffleNXdlPerWavePerShuffle,
           typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
@@ -142,7 +144,8 @@ template <typename ALayout,
           LoopScheduler LoopSched = make_default_loop_scheduler()>
 struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
                                                                      BLayout,
-                                                                     DELayout,
+                                                                     DsLayout,
+                                                                     ELayout,
                                                                      ADataType,
                                                                      BDataType,
                                                                      DsDataType,
@@ -160,7 +163,10 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
     static constexpr auto I2 = Number<2>{};
     static constexpr auto I3 = Number<3>{};
 
-    static auto MakeAGridDescriptor_AK0_M_AK1(index_t MRaw, index_t KRaw, index_t StrideA)
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
+
+    static auto MakeAGridDescriptor_M_K(index_t MRaw, index_t KRaw, index_t StrideA)
     {
         const auto a_grid_desc_mraw_kraw = [&]() {
             if constexpr(is_same_v<tensor_layout::gemm::RowMajor, ALayout>)
@@ -175,95 +181,10 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both M and K
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(M)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad M, but not K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_right_pad_transform(MRaw, MPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad K, but not M
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k = transform_tensor_descriptor(
-                a_grid_desc_mraw_kraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else
-        {
-            // not pad M or K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
+        return matrix_padder.PadADescriptor_M_K(a_grid_desc_mraw_kraw);
     }
 
-    static auto MakeBGridDescriptor_BK0_N_BK1(index_t KRaw, index_t NRaw, index_t StrideB)
+    static auto MakeBGridDescriptor_N_K(index_t KRaw, index_t NRaw, index_t StrideB)
     {
         const auto b_grid_desc_nraw_kraw = [&]() {
             if constexpr(is_same<tensor_layout::gemm::RowMajor, BLayout>::value)
@@ -278,160 +199,50 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
             }
         }();
 
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto NPad = N - NRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::NKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both N and K
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_right_pad_transform(NRaw, NPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(N)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad N, but not K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_right_pad_transform(NRaw, NPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad K, but not N
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k = transform_tensor_descriptor(
-                b_grid_desc_nraw_kraw,
-                make_tuple(make_pass_through_transform(NRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else
-        {
-            // not pad N or K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
+        return matrix_padder.PadBDescriptor_N_K(b_grid_desc_nraw_kraw);
     }
 
+    template <typename ELay>
     static auto MakeEGridDescriptor_M_N(index_t MRaw, index_t NRaw, index_t StrideE)
     {
-        const auto c_grid_desc_mraw_nraw = [&]() {
-            if constexpr(is_same<tensor_layout::gemm::RowMajor, DELayout>::value)
+        const auto e_grid_desc_mraw_nraw = [&]() {
+            if constexpr(is_same<tensor_layout::gemm::RowMajor, ELay>::value)
             {
                 return make_naive_tensor_descriptor(make_tuple(MRaw, NRaw),
                                                     make_tuple(StrideE, I1));
             }
-            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, DELayout>::value)
+            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, ELay>::value)
             {
                 return make_naive_tensor_descriptor(make_tuple(MRaw, NRaw),
                                                     make_tuple(I1, StrideE));
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto NPad = N - NRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad M and N
-            return transform_tensor_descriptor(c_grid_desc_mraw_nraw,
-                                               make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                          make_right_pad_transform(NRaw, NPad)),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad M, but not N
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(NRaw)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad N, but not M
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(NRaw, NPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else
-        {
-            // not pad M or N
-            return c_grid_desc_mraw_nraw;
-        }
+        return matrix_padder.PadCDescriptor_M_N(e_grid_desc_mraw_nraw);
     }
 
-    using AGridDesc_AK0_M_AK1 = decltype(MakeAGridDescriptor_AK0_M_AK1(1, 1, 1));
-    using BGridDesc_BK0_N_BK1 = decltype(MakeBGridDescriptor_BK0_N_BK1(1, 1, 1));
-    using EGridDesc_M_N       = decltype(MakeEGridDescriptor_M_N(1, 1, 1));
+    static auto MakeDsGridDescriptor_M_N(const std::array<index_t, NumDTensor>& MRaws,
+                                         const std::array<index_t, NumDTensor>& NRaws,
+                                         const std::array<index_t, NumDTensor>& DsStride)
+    {
+        return generate_tuple(
+            [&](auto i) {
+                using DLayout = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+
+                return DeviceOp::MakeEGridDescriptor_M_N<DLayout>(MRaws[i], NRaws[i], DsStride[i]);
+            },
+            Number<NumDTensor>{});
+    }
+
+    using AGridDesc_M_K  = decltype(MakeAGridDescriptor_M_K(1, 1, 1));
+    using BGridDesc_N_K  = decltype(MakeBGridDescriptor_N_K(1, 1, 1));
+    using DsGridDesc_M_N = remove_cvref_t<decltype(MakeDsGridDescriptor_M_N({}, {}, {}))>;
+    using EGridDesc_M_N  = decltype(MakeEGridDescriptor_M_N<ELayout>(1, 1, 1));
 
     // GridwiseGemm
-    using GridwiseGemm = GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle<
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
         ADataType, // TODO: distinguish A/B datatype
-        GemmAccDataType,
+        AccDataType,
         CShuffleDataType,
         DsDataType,
         EDataType,
@@ -439,8 +250,9 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
         BElementwiseOperation,
         CDEElementwiseOperation,
         InMemoryDataOperationEnum::Set,
-        AGridDesc_AK0_M_AK1,
-        BGridDesc_BK0_N_BK1,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        DsGridDesc_M_N,
         EGridDesc_M_N,
         NumGemmKPrefetchStage,
         BlockSize,
@@ -475,6 +287,13 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
         CDEBlockTransferScalarPerVector_NPerBlock,
         LoopSched>;
 
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
+    using Block2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+
     // Argument
     struct Argument : public BaseArgument
     {
@@ -494,42 +313,62 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
                  CDEElementwiseOperation cde_element_op)
             : p_a_grid_{static_cast<const ADataType*>(p_a_grid)},
               p_b_grid_{static_cast<const BDataType*>(p_b_grid)},
-              p_ds_grid_{}, // FIXME
+              p_ds_grid_{},
               p_e_grid_{static_cast<EDataType*>(p_e_grid)},
-              a_grid_desc_ak0_m_ak1_{DeviceOp::MakeAGridDescriptor_AK0_M_AK1(MRaw, KRaw, StrideA)},
-              b_grid_desc_bk0_n_bk1_{DeviceOp::MakeBGridDescriptor_BK0_N_BK1(KRaw, NRaw, StrideB)},
+              a_grid_desc_m_k_{DeviceOp::MakeAGridDescriptor_M_K(MRaw, KRaw, StrideA)},
+              b_grid_desc_n_k_{DeviceOp::MakeBGridDescriptor_N_K(KRaw, NRaw, StrideB)},
+              ds_grid_desc_m_n_{},
+              e_grid_desc_m_n_{DeviceOp::MakeEGridDescriptor_M_N<ELayout>(MRaw, NRaw, StrideE)},
+              a_grid_desc_ak0_m_ak1_{
+                  GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k_)},
+              b_grid_desc_bk0_n_bk1_{
+                  GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k_)},
               ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
-              e_grid_desc_m_n_{DeviceOp::MakeEGridDescriptor_M_N(MRaw, NRaw, StrideE)},
               e_grid_desc_mblock_mperblock_nblock_nperblock_{},
               block_2_etile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
               a_element_op_{a_element_op},
               b_element_op_{b_element_op},
               cde_element_op_{cde_element_op}
         {
-            if(GridwiseGemm::CheckValidity(a_grid_desc_ak0_m_ak1_,
-                                           b_grid_desc_bk0_n_bk1_,
+            // populate pointer, desc for Ds
+            static_for<0, NumDTensor, 1>{}([&](auto i) {
+                using DLayout   = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+                using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
+
+                // D pointer
+                p_ds_grid_(i) = static_cast<const DDataType*>(p_ds_grid[i]);
+
+                // D desc
+                ds_grid_desc_m_n_(i) =
+                    DeviceOp::MakeEGridDescriptor_M_N<DLayout>(MRaw, NRaw, StrideDs[i]);
+            });
+
+            // populate desc for Ds/E
+            if(GridwiseGemm::CheckValidity(a_grid_desc_m_k_,
+                                           b_grid_desc_n_k_,
+                                           ds_grid_desc_m_n_,
                                            e_grid_desc_m_n_,
                                            block_2_etile_map_))
             {
+                ds_grid_desc_mblock_mperblock_nblock_nperblock_ =
+                    GridwiseGemm::MakeDsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                        ds_grid_desc_m_n_);
+
                 e_grid_desc_mblock_mperblock_nblock_nperblock_ =
                     GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
                         e_grid_desc_m_n_);
-
-                static_for<0, NumDTensor, 1>{}([&](auto i) {
-                    using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
-
-                    p_ds_grid_(i) = static_cast<const DDataType*>(p_ds_grid[i]);
-
-                    const auto d_grid_desc_m_n =
-                        DeviceOp::MakeEGridDescriptor_M_N(MRaw, NRaw, StrideDs[i]);
-
-                    ds_grid_desc_mblock_mperblock_nblock_nperblock_(i) =
-                        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
-                            d_grid_desc_m_n);
-                });
             }
         }
 
+        void Print() const
+        {
+            std::cout << "A[M, K]: " << a_grid_desc_m_k_ << std::endl;
+            std::cout << "B[N, K]: " << b_grid_desc_n_k_ << std::endl;
+            static_for<0, NumDTensor, 1>{}(
+                [&](auto i) { std::cout << "Ds[M, N]: " << ds_grid_desc_m_n_[i] << std::endl; });
+            std::cout << "E[M, N]: " << e_grid_desc_m_n_ << std::endl;
+        }
+
         //  private:
         // pointers
         const ADataType* p_a_grid_;
@@ -537,20 +376,22 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
         typename GridwiseGemm::DsGridPointer p_ds_grid_;
         EDataType* p_e_grid_;
 
-        // tensor descriptors
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        DsGridDesc_M_N ds_grid_desc_m_n_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
         AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
         BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
-        StaticallyIndexedArray<
-            typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-            NumDTensor>
-            ds_grid_desc_mblock_mperblock_nblock_nperblock_; // FIXME: Ds desc may be of different
-                                                             // type from E
-        EGridDesc_M_N e_grid_desc_m_n_;
+        typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            ds_grid_desc_mblock_mperblock_nblock_nperblock_;
         typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
             e_grid_desc_mblock_mperblock_nblock_nperblock_;
 
         // block-to-e-tile map
-        typename GridwiseGemm::DefaultBlock2ETileMap block_2_etile_map_;
+        Block2ETileMap block_2_etile_map_;
 
         // element-wise op
         AElementwiseOperation a_element_op_;
@@ -565,8 +406,9 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                            arg.b_grid_desc_bk0_n_bk1_,
+            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                            arg.b_grid_desc_n_k_,
+                                            arg.ds_grid_desc_m_n_,
                                             arg.e_grid_desc_m_n_,
                                             arg.block_2_etile_map_))
             {
@@ -592,9 +434,7 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
                     CDEElementwiseOperation,
                     DeviceOp::AGridDesc_AK0_M_AK1,
                     DeviceOp::BGridDesc_BK0_N_BK1,
-                    ck::StaticallyIndexedArray<
-                        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-                        NumDTensor>,
+                    typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
                     typename GridwiseGemm::DefaultBlock2ETileMap,
                     has_main_loop>;
@@ -618,18 +458,14 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
                                               arg.block_2_etile_map_);
             };
 
-            float ave_time = 0;
-
             if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
             {
-                ave_time = launch_kernel(integral_constant<bool, true>{});
+                return launch_kernel(integral_constant<bool, true>{});
             }
             else
             {
-                ave_time = launch_kernel(integral_constant<bool, false>{});
+                return launch_kernel(integral_constant<bool, false>{});
             }
-
-            return ave_time;
         }
 
         // polymorphic
@@ -647,8 +483,9 @@ struct DeviceGemmMultipleD_Xdl_CShuffle : public DeviceGemmMultipleD<ALayout,
             return false;
         }
 
-        return GridwiseGemm::CheckValidity(arg.a_grid_desc_ak0_m_ak1_,
-                                           arg.b_grid_desc_bk0_n_bk1_,
+        return GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                           arg.b_grid_desc_n_k_,
+                                           arg.ds_grid_desc_m_n_,
                                            arg.e_grid_desc_m_n_,
                                            arg.block_2_etile_map_);
     }
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_reduce_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_reduce_xdl_cshuffle.hpp
index 722ae1137b..cf19083954 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_reduce_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_reduce_xdl_cshuffle.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_reduce.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_reduce_xdl_cshuffle_v1.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_xdl.hpp
index 98028e1f28..0f91797ec9 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_xdl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_xdl.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_cshuffle.hpp
index 9c8b189add..b7502ddac8 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_cshuffle.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v1.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_layernorm_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_layernorm_cshuffle.hpp
index b82fcb67f7..875623dcb6 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_layernorm_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_layernorm_cshuffle.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_layernorm_cshuffle_v1.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk.hpp
index 306a73dff1..b5eed11aeb 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_splitk.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r4.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk_c_shuffle.hpp b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk_c_shuffle.hpp
index 52bdacf7db..eb2e521bdb 100644
--- a/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk_c_shuffle.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_gemm_xdl_splitk_c_shuffle.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_gemm_splitk.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r4r2.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp b/include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp
new file mode 100644
index 0000000000..335b8b0b35
--- /dev/null
+++ b/include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp
@@ -0,0 +1,63 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <vector>
+
+#include "ck/tensor_operation/gpu/device/device_base.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+
+// Convolution Forward:
+//   input : input image A[G, N, C, Hi, Wi],
+//   input : weight B[G, K, C, Y, X],
+//   input : D0[G, N, K, Ho, Wo], D1[G, N, K, Ho, Wo], ...
+//   output : output image E[G, N, K, Ho, Wo]
+//   C = a_op(A) * b_op(B)
+//   E = cde_op(C, D0, D1, ...)
+template <index_t NDimSpatial,
+          typename ALayout,
+          typename BLayout,
+          typename DsLayout,
+          typename ELayout,
+          typename ADataType,
+          typename BDataType,
+          typename DsDataType,
+          typename EDataType,
+          typename AElementwiseOperation,
+          typename BElementwiseOperation,
+          typename CDEElementwiseOperation>
+struct DeviceGroupedConvFwdMultipleD : public BaseOperator
+{
+    static constexpr index_t NumDTensor = DsDataType::Size();
+
+    virtual std::unique_ptr<BaseArgument> MakeArgumentPointer(
+        const void* p_a,
+        const void* p_b,
+        const std::array<const void*, NumDTensor>& p_ds,
+        void* p_e,
+        const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+        const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+        const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+        const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_strides,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_lengths,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_strides,
+        const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+        const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_strides,
+        const std::array<index_t, NDimSpatial>& conv_filter_strides,
+        const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+        const std::array<index_t, NDimSpatial>& input_left_pads,
+        const std::array<index_t, NDimSpatial>& input_right_pads,
+        const AElementwiseOperation& a_element_op,
+        const BElementwiseOperation& b_element_op,
+        const CDEElementwiseOperation& cde_element_op) = 0;
+
+    virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
+};
+
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp
new file mode 100644
index 0000000000..936ac25d09
--- /dev/null
+++ b/include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp
@@ -0,0 +1,1813 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <functional>
+#include <iostream>
+#include <iterator>
+#include <numeric>
+#include <sstream>
+
+#include "ck/utility/common_header.hpp"
+#include "ck/tensor_description/tensor_descriptor.hpp"
+#include "ck/tensor_description/tensor_descriptor_helper.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
+#include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
+#include "ck/host_utility/io.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+
+namespace {
+
+template <index_t NumDTensor>
+struct ComputePtrOffsetOfStridedBatch
+{
+    ComputePtrOffsetOfStridedBatch() = default;
+
+    ComputePtrOffsetOfStridedBatch(index_t BatchStrideA,
+                                   index_t BatchStrideB,
+                                   Array<ck::index_t, NumDTensor> BatchStrideDs,
+                                   index_t BatchStrideE)
+        : BatchStrideA_(BatchStrideA),
+          BatchStrideB_(BatchStrideB),
+          BatchStrideDs_(BatchStrideDs),
+          BatchStrideE_(BatchStrideE)
+    {
+    }
+
+    __host__ __device__ constexpr long_index_t GetAPtrOffset(index_t g_idx) const
+    {
+        return g_idx * static_cast<long_index_t>(BatchStrideA_);
+    }
+
+    __host__ __device__ constexpr long_index_t GetBPtrOffset(index_t g_idx) const
+    {
+        return g_idx * static_cast<long_index_t>(BatchStrideB_);
+    }
+
+    __host__ __device__ constexpr auto GetDsPtrOffset(index_t g_idx) const
+    {
+        Array<long_index_t, NumDTensor> ds_offset;
+        static_for<0, NumDTensor, 1>{}(
+            [&](auto i) { ds_offset(i) = g_idx * static_cast<long_index_t>(BatchStrideDs_[i]); });
+        return ds_offset;
+    }
+
+    __host__ __device__ constexpr long_index_t GetEPtrOffset(index_t g_idx) const
+    {
+        return g_idx * static_cast<long_index_t>(BatchStrideE_);
+    }
+
+    index_t BatchStrideA_;
+    index_t BatchStrideB_;
+    Array<ck::index_t, NumDTensor> BatchStrideDs_;
+    index_t BatchStrideE_;
+};
+
+/*
+ * \brief Wrapper function of GridwiseGemm::Run to realize BatchedGEMM.
+ *
+ * \tparam ComputePtrOffsetOfBatch Class that computes the base pointer offsets of A, B, C matrix
+ * given the batch. For example, ComputePtrOffsetOfStridedBatch() computes the offsets of evenly
+ * strided batched, but we can easily extend to other layouts. The returned offset can be either \p
+ * index_t or \p long_index_t. If it returns \p long_index_t, we are not subject to the 2GB
+ * limitations.
+ *
+ * \tparam Block2ETileMap Block2ETileMap::CalculateBottomIndex() takes in id of a workgroup and
+ * returns the 2D index of the tile that it computes. \see
+ * GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3::Run().
+ *
+ * \note Using \p ComputePtrOffsetOfBatch gives us the flexibility that 2 workgroups can compute 2
+ * tiles from different matrices. Keep in mind that these 2 matrices can share the same grid
+ * descriptor (like in BatchedGEMM), or use their own grid descriptors (in GroupedGemm). \link
+ * device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp kernel_gemm_xdlops_v2r3_for_conv3d \endlink for \link
+ * DeviceConv3d \endlink uses the same concept, but currently does NOT encapsulate the computing of
+ * pointer offset into \p ComputePtrOffsetOfStridedBatch.
+ *
+ * \note \p Block2ETileMap allows customized mapping between a workgroup and the C-tile it computes.
+ * Together with \p ComputePtrOffsetOfBatch, we can reuse GridwiseGemm (and GridwiseGemm fusion ) to
+ * realize BatchedGemm and GroupedGemm (and the corresponding GEMM fusion).
+ *
+ */
+template <typename GridwiseGemm,
+          typename ABDataType,
+          typename DsPointer,
+          typename EDataType,
+          typename AElementwiseOperation,
+          typename BElementwiseOperation,
+          typename CDEElementwiseOperation,
+          typename AGridDesc_AK0_M_AK1,
+          typename BGridDesc_BK0_N_BK1,
+          typename DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
+          typename EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock,
+          typename Block2ETileMap,
+          typename ComputePtrOffsetOfBatch,
+          bool HasMainKBlockLoop>
+__global__ void
+#if CK_USE_LAUNCH_BOUNDS
+    __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU)
+#endif
+        kernel_batch_gemm_multiple_d_xdl_cshuffle(
+            const ABDataType* __restrict__ p_a_grid,
+            const ABDataType* __restrict__ p_b_grid,
+            DsPointer p_ds_grid,
+            EDataType* __restrict__ p_e_grid,
+            const AElementwiseOperation a_element_op,
+            const BElementwiseOperation b_element_op,
+            const CDEElementwiseOperation cde_element_op,
+            const index_t batch_count,
+            const AGridDesc_AK0_M_AK1 a_grid_desc_k0_m_k1,
+            const BGridDesc_BK0_N_BK1 b_grid_desc_k0_n_k1,
+            const DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+                ds_grid_desc_mblock_mperblock_nblock_nperblock,
+            const EGridDesc_MBlock_MPerBlock_NBlock_NPerBlock
+                e_grid_desc_mblock_mperblock_nblock_nperblock_,
+            const Block2ETileMap block_2_ctile_map,
+            const ComputePtrOffsetOfBatch compute_ptr_offset_of_batch)
+{
+#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__))
+
+#if 1
+    const index_t num_blocks_per_batch =
+        __builtin_amdgcn_readfirstlane(get_grid_size() / batch_count);
+    const index_t g_idx = __builtin_amdgcn_readfirstlane(get_block_1d_id() / num_blocks_per_batch);
+
+    const long_index_t a_batch_offset = __builtin_amdgcn_readfirstlane(
+        static_cast<long_index_t>(compute_ptr_offset_of_batch.GetAPtrOffset(g_idx)));
+    const long_index_t b_batch_offset = __builtin_amdgcn_readfirstlane(
+        static_cast<long_index_t>(compute_ptr_offset_of_batch.GetBPtrOffset(g_idx)));
+    const long_index_t e_batch_offset = __builtin_amdgcn_readfirstlane(
+        static_cast<long_index_t>(compute_ptr_offset_of_batch.GetEPtrOffset(g_idx)));
+
+    const auto ds_batch_offset = compute_ptr_offset_of_batch.GetDsPtrOffset(g_idx);
+
+    __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
+
+    DsPointer p_ds_grid_grp;
+
+    static constexpr index_t NumDTensor =
+        DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock::Size();
+
+    static_for<0, NumDTensor, 1>{}(
+        [&](auto i) { p_ds_grid_grp(i) = p_ds_grid[i] + ds_batch_offset[i]; });
+
+    GridwiseGemm::template Run<HasMainKBlockLoop>(p_a_grid + a_batch_offset,
+                                                  p_b_grid + b_batch_offset,
+                                                  p_ds_grid_grp,
+                                                  p_e_grid + e_batch_offset,
+                                                  p_shared,
+                                                  a_element_op,
+                                                  b_element_op,
+                                                  cde_element_op,
+                                                  a_grid_desc_k0_m_k1,
+                                                  b_grid_desc_k0_n_k1,
+                                                  ds_grid_desc_mblock_mperblock_nblock_nperblock,
+                                                  e_grid_desc_mblock_mperblock_nblock_nperblock_,
+                                                  block_2_ctile_map);
+#else
+    __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
+
+    GridwiseGemm::template Run<HasMainKBlockLoop>(p_a_grid,
+                                                  p_b_grid,
+                                                  p_ds_grid,
+                                                  p_e_grid,
+                                                  p_shared,
+                                                  a_element_op,
+                                                  b_element_op,
+                                                  cde_element_op,
+                                                  a_grid_desc_k0_m_k1,
+                                                  b_grid_desc_k0_n_k1,
+                                                  ds_grid_desc_mblock_mperblock_nblock_nperblock,
+                                                  e_grid_desc_mblock_mperblock_nblock_nperblock_,
+                                                  block_2_ctile_map);
+#endif
+
+#else
+    ignore = p_a_grid;
+    ignore = p_b_grid;
+    ignore = p_ds_grid;
+    ignore = p_e_grid;
+    ignore = batch_count;
+    ignore = a_grid_desc_k0_m_k1;
+    ignore = b_grid_desc_k0_n_k1;
+    ignore = ds_grid_desc_mblock_mperblock_nblock_nperblock;
+    ignore = e_grid_desc_mblock_mperblock_nblock_nperblock_;
+    ignore = a_element_op;
+    ignore = b_element_op;
+    ignore = cde_element_op;
+    ignore = compute_ptr_offset_of_batch;
+    ignore = block_2_ctile_map;
+#endif
+}
+
+} // namespace
+
+//
+// @brief      Device Convolution operation.
+//
+// Supports:
+//  @li         Forward convolution with up to 3 spatial dimentions
+//  @li         Input tensor in GNWC data format
+//  @li         Weight tensor in GKXC data format
+//  @li         Output tensor in GNWK data format
+//
+// 1D:
+// out[N, Wo, K] = in[N, Wi, C] * wei[K, X, C]
+// 2D:
+// out[N, Ho, Wo, K] = in[N, Hi, Wi, C] * wei[K, Y, X, C]
+// 3D:
+// out[N, Do, Ho, Wo, K] = in[N, Di, Hi, Wi, C] * wei[K, Z, Y, X, C]
+//
+template <index_t NDimSpatial,
+          typename ALayout,
+          typename BLayout,
+          typename DsLayout,
+          typename ELayout,
+          typename ADataType,
+          typename BDataType,
+          typename AccDataType,
+          typename CShuffleDataType,
+          typename DsDataType,
+          typename EDataType,
+          typename AElementwiseOperation,
+          typename BElementwiseOperation,
+          typename CDEElementwiseOperation,
+          ConvolutionForwardSpecialization ConvForwardSpecialization,
+          GemmSpecialization GemmSpec,
+          index_t NumGemmKPrefetchStage,
+          index_t BlockSize,
+          index_t MPerBlock,
+          index_t NPerBlock,
+          index_t KPerBlock,
+          index_t AK1,
+          index_t BK1,
+          index_t MPerXDL,
+          index_t NPerXDL,
+          index_t MXdlPerWave,
+          index_t NXdlPerWave,
+          typename ABlockTransferThreadClusterLengths_AK0_M_AK1,
+          typename ABlockTransferThreadClusterArrangeOrder,
+          typename ABlockTransferSrcAccessOrder,
+          index_t ABlockTransferSrcVectorDim,
+          index_t ABlockTransferSrcScalarPerVector,
+          index_t ABlockTransferDstScalarPerVector_AK1,
+          index_t ABlockLdsExtraM,
+          typename BBlockTransferThreadClusterLengths_BK0_N_BK1,
+          typename BBlockTransferThreadClusterArrangeOrder,
+          typename BBlockTransferSrcAccessOrder,
+          index_t BBlockTransferSrcVectorDim,
+          index_t BBlockTransferSrcScalarPerVector,
+          index_t BBlockTransferDstScalarPerVector_BK1,
+          index_t BBlockLdsExtraN,
+          index_t CShuffleMXdlPerWavePerShuffle,
+          index_t CShuffleNXdlPerWavePerShuffle,
+          typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
+          index_t CDEBlockTransferScalarPerVector_NPerBlock,
+          LoopScheduler LoopSched = make_default_loop_scheduler()>
+struct DeviceGroupedConvFwdMultipleD_Xdl_CShuffle
+    : public DeviceGroupedConvFwdMultipleD<NDimSpatial,
+                                           ALayout,
+                                           BLayout,
+                                           DsLayout,
+                                           ELayout,
+                                           ADataType,
+                                           BDataType,
+                                           DsDataType,
+                                           EDataType,
+                                           AElementwiseOperation,
+                                           BElementwiseOperation,
+                                           CDEElementwiseOperation>
+{
+    using DeviceOp = DeviceGroupedConvFwdMultipleD_Xdl_CShuffle;
+
+    static constexpr index_t NumDTensor = DsDataType::Size();
+
+    static constexpr auto I0 = Number<0>{};
+    static constexpr auto I1 = Number<1>{};
+    static constexpr auto I2 = Number<2>{};
+    static constexpr auto I3 = Number<3>{};
+
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
+
+    template <typename ALay,
+              typename std::enable_if<NDimSpatial == 1 &&
+                                          is_same_v<ALay, tensor_layout::convolution::GNWC>,
+                                      bool>::type = false>
+    static auto
+    MakeAGridDescriptor_M_K(const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* a_g_n_c_wis_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */,
+                            const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                            const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                            const std::array<index_t, NDimSpatial>& input_left_pads,
+                            const std::array<index_t, NDimSpatial>& input_right_pads)
+    {
+        const index_t N = a_g_n_c_wis_lengths[1];
+        const index_t C = a_g_n_c_wis_lengths[2];
+
+        const index_t Wi = a_g_n_c_wis_lengths[3];
+
+        const index_t Wo = e_g_n_k_wos_lengths[3];
+
+        const index_t ConvStrideW = conv_filter_strides[0];
+
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            const index_t NWo = N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                                    e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                                    index_t{1},
+                                                    std::multiplies<index_t>());
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(NWo, C));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            const auto in_n_wi_c_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(N, Wi, C));
+
+            const auto in_n_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
+
+            const auto in_gemmmraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+                in_n_wo_c_grid_desc,
+                make_tuple(make_merge_transform(make_tuple(N, Wo)), make_pass_through_transform(C)),
+                make_tuple(Sequence<0, 1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else
+        {
+            const index_t X             = b_g_k_c_xs_lengths[3];
+            const index_t ConvDilationW = conv_filter_dilations[0];
+            const index_t InLeftPadW    = input_left_pads[0];
+            const index_t InRightPadW   = input_right_pads[0];
+
+            const auto in_n_wi_c_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(N, Wi, C));
+
+            const auto in_n_wip_c_grid_desc = transform_tensor_descriptor(
+                in_n_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
+
+            const auto in_n_x_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_wip_c_grid_desc,
+                make_tuple(
+                    make_pass_through_transform(N),
+                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
+                    make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3>{}));
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                transform_tensor_descriptor(in_n_x_wo_c_grid_desc,
+                                            make_tuple(make_merge_transform(make_tuple(N, Wo)),
+                                                       make_merge_transform(make_tuple(X, C))),
+                                            make_tuple(Sequence<0, 2>{}, Sequence<1, 3>{}),
+                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+    }
+
+    template <typename ALay,
+              typename std::enable_if<NDimSpatial == 2 &&
+                                          is_same_v<ALay, tensor_layout::convolution::GNHWC>,
+                                      bool>::type = false>
+    static auto
+    MakeAGridDescriptor_M_K(const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* a_g_n_c_wis_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */,
+                            const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                            const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                            const std::array<index_t, NDimSpatial>& input_left_pads,
+                            const std::array<index_t, NDimSpatial>& input_right_pads)
+    {
+        const index_t N = a_g_n_c_wis_lengths[1];
+        const index_t C = a_g_n_c_wis_lengths[2];
+
+        const index_t Hi = a_g_n_c_wis_lengths[3];
+        const index_t Wi = a_g_n_c_wis_lengths[4];
+
+        const index_t Ho = e_g_n_k_wos_lengths[3];
+        const index_t Wo = e_g_n_k_wos_lengths[4];
+
+        const index_t ConvStrideH = conv_filter_strides[0];
+        const index_t ConvStrideW = conv_filter_strides[1];
+
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            const index_t NHoWo = N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                                      e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                                      index_t{1},
+                                                      std::multiplies<index_t>());
+
+            const auto in_gemmmraw_gemmkraw_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(NHoWo, C));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            const auto in_n_hi_wi_c_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(N, Hi, Wi, C));
+
+            const auto in_n_ho_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_embed_transform(make_tuple(Ho), make_tuple(ConvStrideH)),
+                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                transform_tensor_descriptor(in_n_ho_wo_c_grid_desc,
+                                            make_tuple(make_merge_transform(make_tuple(N, Ho, Wo)),
+                                                       make_pass_through_transform(C)),
+                                            make_tuple(Sequence<0, 1, 2>{}, Sequence<3>{}),
+                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else
+        {
+            const index_t Y = b_g_k_c_xs_lengths[3];
+            const index_t X = b_g_k_c_xs_lengths[4];
+
+            const index_t ConvDilationH = conv_filter_dilations[0];
+            const index_t ConvDilationW = conv_filter_dilations[1];
+
+            const index_t InLeftPadH = input_left_pads[0];
+            const index_t InLeftPadW = input_left_pads[1];
+
+            const index_t InRightPadH = input_right_pads[0];
+            const index_t InRightPadW = input_right_pads[1];
+
+            const auto in_n_hi_wi_c_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(N, Hi, Wi, C));
+
+            const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
+                in_n_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_pad_transform(Hi, InLeftPadH, InRightPadH),
+                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+
+            const auto in_n_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_hip_wip_c_grid_desc,
+                make_tuple(
+                    make_pass_through_transform(N),
+                    make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
+                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
+                    make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3, 4>{}, Sequence<5>{}));
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                transform_tensor_descriptor(in_n_y_ho_x_wo_c_grid_desc,
+                                            make_tuple(make_merge_transform(make_tuple(N, Ho, Wo)),
+                                                       make_merge_transform(make_tuple(Y, X, C))),
+                                            make_tuple(Sequence<0, 2, 4>{}, Sequence<1, 3, 5>{}),
+                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+    }
+
+    template <typename ALay,
+              typename std::enable_if<NDimSpatial == 3 &&
+                                          is_same_v<ALay, tensor_layout::convolution::GNDHWC>,
+                                      bool>::type = false>
+    static auto
+    MakeAGridDescriptor_M_K(const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* a_g_n_c_wis_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */,
+                            const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                            const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                            const std::array<index_t, NDimSpatial>& input_left_pads,
+                            const std::array<index_t, NDimSpatial>& input_right_pads)
+    {
+        const index_t N = a_g_n_c_wis_lengths[1];
+        const index_t C = a_g_n_c_wis_lengths[2];
+
+        const index_t Di = a_g_n_c_wis_lengths[3];
+        const index_t Hi = a_g_n_c_wis_lengths[4];
+        const index_t Wi = a_g_n_c_wis_lengths[5];
+
+        const index_t Do = e_g_n_k_wos_lengths[3];
+        const index_t Ho = e_g_n_k_wos_lengths[4];
+        const index_t Wo = e_g_n_k_wos_lengths[5];
+
+        const index_t ConvStrideD = conv_filter_strides[0];
+        const index_t ConvStrideH = conv_filter_strides[1];
+        const index_t ConvStrideW = conv_filter_strides[2];
+
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            const index_t NDoHoWo =
+                N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                    e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                    index_t{1},
+                                    std::multiplies<index_t>());
+
+            const auto in_gemmmraw_gemmkraw_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(NDoHoWo, C));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            const auto in_n_di_hi_wi_c_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(N, Di, Hi, Wi, C));
+
+            const auto in_n_do_ho_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_di_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_embed_transform(make_tuple(Do), make_tuple(ConvStrideD)),
+                           make_embed_transform(make_tuple(Ho), make_tuple(ConvStrideH)),
+                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
+                           make_pass_through_transform(C)),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}));
+
+            const auto in_gemmmraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+                in_n_do_ho_wo_c_grid_desc,
+                make_tuple(make_merge_transform(make_tuple(N, Do, Ho, Wo)),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0, 1, 2, 3>{}, Sequence<4>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else
+        {
+            const index_t Z = b_g_k_c_xs_lengths[3];
+            const index_t Y = b_g_k_c_xs_lengths[4];
+            const index_t X = b_g_k_c_xs_lengths[5];
+
+            const index_t ConvDilationD = conv_filter_dilations[0];
+            const index_t ConvDilationH = conv_filter_dilations[1];
+            const index_t ConvDilationW = conv_filter_dilations[2];
+
+            const index_t InLeftPadD = input_left_pads[0];
+            const index_t InLeftPadH = input_left_pads[1];
+            const index_t InLeftPadW = input_left_pads[2];
+
+            const index_t InRightPadD = input_right_pads[0];
+            const index_t InRightPadH = input_right_pads[1];
+            const index_t InRightPadW = input_right_pads[2];
+
+            const auto in_n_di_hi_wi_c_grid_desc =
+                make_naive_tensor_descriptor_packed(make_tuple(N, Di, Hi, Wi, C));
+
+            const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
+                in_n_di_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_pad_transform(Di, InLeftPadD, InRightPadD),
+                           make_pad_transform(Hi, InLeftPadH, InRightPadH),
+                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
+                           make_pass_through_transform(C)),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}));
+
+            const auto in_n_z_do_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_hip_wip_c_grid_desc,
+                make_tuple(
+                    make_pass_through_transform(N),
+                    make_embed_transform(make_tuple(Z, Do), make_tuple(ConvDilationD, ConvStrideD)),
+                    make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
+                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
+                    make_pass_through_transform(C)),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
+                make_tuple(Sequence<0>{},
+                           Sequence<1, 2>{},
+                           Sequence<3, 4>{},
+                           Sequence<5, 6>{},
+                           Sequence<7>{}));
+
+            const auto in_gemmmraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+                in_n_z_do_y_ho_x_wo_c_grid_desc,
+                make_tuple(make_merge_transform(make_tuple(N, Do, Ho, Wo)),
+                           make_merge_transform(make_tuple(Z, Y, X, C))),
+                make_tuple(Sequence<0, 2, 4, 6>{}, Sequence<1, 3, 5, 7>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+    }
+
+    // TODO: implement ck::tensor_layout::convolution that describe packed/strided dimemsion as
+    // properties
+    template <typename ALay,
+              typename std::enable_if<NDimSpatial == 1 &&
+                                          (is_same_v<ALay, tensor_layout::convolution::G_NW_C> ||
+                                           is_same_v<ALay, tensor_layout::convolution::NWGC>),
+                                      bool>::type = false>
+    static auto
+    MakeAGridDescriptor_M_K(const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */,
+                            const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                            const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                            const std::array<index_t, NDimSpatial>& input_left_pads,
+                            const std::array<index_t, NDimSpatial>& input_right_pads)
+    {
+        const index_t N = a_g_n_c_wis_lengths[1];
+        const index_t C = a_g_n_c_wis_lengths[2];
+
+        const index_t Wi = a_g_n_c_wis_lengths[3];
+
+        const index_t Wo = e_g_n_k_wos_lengths[3];
+
+        const index_t ConvStrideW = conv_filter_strides[0];
+
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            const index_t NHoWo = N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                                      e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                                      index_t{1},
+                                                      std::multiplies<index_t>());
+
+            // This is different
+            const index_t WiStride = a_g_n_c_wis_strides[2 + NDimSpatial];
+            const auto CStride     = I1;
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                make_naive_tensor_descriptor(make_tuple(NHoWo, C), make_tuple(WiStride, CStride));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            // This is different
+            const index_t NStride  = a_g_n_c_wis_strides[1];
+            const index_t WiStride = a_g_n_c_wis_strides[3];
+            const auto CStride     = I1;
+
+            const auto in_n_wi_c_grid_desc = make_naive_tensor_descriptor(
+                make_tuple(N, Wi, C), make_tuple(NStride, WiStride, CStride));
+
+            const auto in_n_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
+
+            const auto in_gemmmraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+                in_n_wo_c_grid_desc,
+                make_tuple(make_merge_transform(make_tuple(N, Wo)), make_pass_through_transform(C)),
+                make_tuple(Sequence<0, 1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else
+        {
+            const index_t X             = b_g_k_c_xs_lengths[3];
+            const index_t ConvDilationW = conv_filter_dilations[0];
+            const index_t InLeftPadW    = input_left_pads[0];
+            const index_t InRightPadW   = input_right_pads[0];
+
+            // This is different
+            const index_t NStride  = a_g_n_c_wis_strides[1];
+            const index_t WiStride = a_g_n_c_wis_strides[3];
+            const auto CStride     = I1;
+
+            const auto in_n_wi_c_grid_desc = make_naive_tensor_descriptor(
+                make_tuple(N, Wi, C), make_tuple(NStride, WiStride, CStride));
+
+            const auto in_n_wip_c_grid_desc = transform_tensor_descriptor(
+                in_n_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}));
+
+            const auto in_n_x_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_wip_c_grid_desc,
+                make_tuple(
+                    make_pass_through_transform(N),
+                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
+                    make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}),
+                make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3>{}));
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                transform_tensor_descriptor(in_n_x_wo_c_grid_desc,
+                                            make_tuple(make_merge_transform(make_tuple(N, Wo)),
+                                                       make_merge_transform(make_tuple(X, C))),
+                                            make_tuple(Sequence<0, 2>{}, Sequence<1, 3>{}),
+                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+    }
+
+    template <typename ALay,
+              typename std::enable_if<NDimSpatial == 2 &&
+                                          (is_same_v<ALay, tensor_layout::convolution::G_NHW_C> ||
+                                           is_same_v<ALay, tensor_layout::convolution::NHWGC>),
+                                      bool>::type = false>
+    static auto
+    MakeAGridDescriptor_M_K(const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */,
+                            const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                            const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                            const std::array<index_t, NDimSpatial>& input_left_pads,
+                            const std::array<index_t, NDimSpatial>& input_right_pads)
+    {
+        const index_t N = a_g_n_c_wis_lengths[1];
+        const index_t C = a_g_n_c_wis_lengths[2];
+
+        const index_t Hi = a_g_n_c_wis_lengths[3];
+        const index_t Wi = a_g_n_c_wis_lengths[4];
+
+        const index_t Ho = e_g_n_k_wos_lengths[3];
+        const index_t Wo = e_g_n_k_wos_lengths[4];
+
+        const index_t ConvStrideH = conv_filter_strides[0];
+        const index_t ConvStrideW = conv_filter_strides[1];
+
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            const index_t NHoWo = N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                                      e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                                      index_t{1},
+                                                      std::multiplies<index_t>());
+
+            // This is different
+            const index_t WiStride = a_g_n_c_wis_strides[2 + NDimSpatial];
+            const auto CStride     = I1;
+
+            const auto in_gemmmraw_gemmkraw_grid_desc =
+                make_naive_tensor_descriptor(make_tuple(NHoWo, C), make_tuple(WiStride, CStride));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            // This is different
+            const index_t NStride  = a_g_n_c_wis_strides[1];
+            const index_t HiStride = a_g_n_c_wis_strides[3];
+            const index_t WiStride = a_g_n_c_wis_strides[4];
+            const auto CStride     = I1;
+
+            const auto in_n_hi_wi_c_grid_desc = make_naive_tensor_descriptor(
+                make_tuple(N, Hi, Wi, C), make_tuple(NStride, HiStride, WiStride, CStride));
+
+            const auto in_n_ho_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_embed_transform(make_tuple(Ho), make_tuple(ConvStrideH)),
+                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                transform_tensor_descriptor(in_n_ho_wo_c_grid_desc,
+                                            make_tuple(make_merge_transform(make_tuple(N, Ho, Wo)),
+                                                       make_pass_through_transform(C)),
+                                            make_tuple(Sequence<0, 1, 2>{}, Sequence<3>{}),
+                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else
+        {
+            const index_t Y = b_g_k_c_xs_lengths[3];
+            const index_t X = b_g_k_c_xs_lengths[4];
+
+            const index_t ConvDilationH = conv_filter_dilations[0];
+            const index_t ConvDilationW = conv_filter_dilations[1];
+
+            const index_t InLeftPadH = input_left_pads[0];
+            const index_t InLeftPadW = input_left_pads[1];
+
+            const index_t InRightPadH = input_right_pads[0];
+            const index_t InRightPadW = input_right_pads[1];
+
+            // This is different
+            const index_t NStride  = a_g_n_c_wis_strides[1];
+            const index_t HiStride = a_g_n_c_wis_strides[3];
+            const index_t WiStride = a_g_n_c_wis_strides[4];
+            const auto CStride     = I1;
+
+            const auto in_n_hi_wi_c_grid_desc = make_naive_tensor_descriptor(
+                make_tuple(N, Hi, Wi, C), make_tuple(NStride, HiStride, WiStride, CStride));
+
+            const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
+                in_n_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_pad_transform(Hi, InLeftPadH, InRightPadH),
+                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}));
+
+            const auto in_n_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_hip_wip_c_grid_desc,
+                make_tuple(
+                    make_pass_through_transform(N),
+                    make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
+                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
+                    make_pass_through_transform(C)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}),
+                make_tuple(Sequence<0>{}, Sequence<1, 2>{}, Sequence<3, 4>{}, Sequence<5>{}));
+
+            const auto in_gemmmraw_gemmk_grid_desc =
+                transform_tensor_descriptor(in_n_y_ho_x_wo_c_grid_desc,
+                                            make_tuple(make_merge_transform(make_tuple(N, Ho, Wo)),
+                                                       make_merge_transform(make_tuple(Y, X, C))),
+                                            make_tuple(Sequence<0, 2, 4>{}, Sequence<1, 3, 5>{}),
+                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmk_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+    }
+
+    template <typename ALay,
+              typename std::enable_if<NDimSpatial == 3 &&
+                                          (is_same_v<ALay, tensor_layout::convolution::G_NDHW_C> ||
+                                           is_same_v<ALay, tensor_layout::convolution::NDHWGC>),
+                                      bool>::type = false>
+    static auto
+    MakeAGridDescriptor_M_K(const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */,
+                            const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                            const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                            const std::array<index_t, NDimSpatial>& input_left_pads,
+                            const std::array<index_t, NDimSpatial>& input_right_pads)
+    {
+        const index_t N = a_g_n_c_wis_lengths[1];
+        const index_t C = a_g_n_c_wis_lengths[2];
+
+        const index_t Di = a_g_n_c_wis_lengths[3];
+        const index_t Hi = a_g_n_c_wis_lengths[4];
+        const index_t Wi = a_g_n_c_wis_lengths[5];
+
+        const index_t Do = e_g_n_k_wos_lengths[3];
+        const index_t Ho = e_g_n_k_wos_lengths[4];
+        const index_t Wo = e_g_n_k_wos_lengths[5];
+
+        const index_t ConvStrideD = conv_filter_strides[0];
+        const index_t ConvStrideH = conv_filter_strides[1];
+        const index_t ConvStrideW = conv_filter_strides[2];
+
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            const index_t NDoHoWo =
+                N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                    e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                    index_t{1},
+                                    std::multiplies<index_t>());
+
+            // This is different
+            const index_t WiStride = a_g_n_c_wis_strides[2 + NDimSpatial];
+            const auto CStride     = I1;
+
+            const auto in_gemmmraw_gemmkraw_grid_desc =
+                make_naive_tensor_descriptor(make_tuple(NDoHoWo, C), make_tuple(WiStride, CStride));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            // This is different
+            const index_t NStride  = a_g_n_c_wis_strides[1];
+            const index_t DiStride = a_g_n_c_wis_strides[3];
+            const index_t HiStride = a_g_n_c_wis_strides[4];
+            const index_t WiStride = a_g_n_c_wis_strides[5];
+            const auto CStride     = I1;
+
+            const auto in_n_di_hi_wi_c_grid_desc = make_naive_tensor_descriptor(
+                make_tuple(N, Di, Hi, Wi, C),
+                make_tuple(NStride, DiStride, HiStride, WiStride, CStride));
+
+            const auto in_n_do_ho_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_di_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_embed_transform(make_tuple(Do), make_tuple(ConvStrideD)),
+                           make_embed_transform(make_tuple(Ho), make_tuple(ConvStrideH)),
+                           make_embed_transform(make_tuple(Wo), make_tuple(ConvStrideW)),
+                           make_pass_through_transform(C)),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}));
+
+            const auto in_gemmmraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+                in_n_do_ho_wo_c_grid_desc,
+                make_tuple(make_merge_transform(make_tuple(N, Do, Ho, Wo)),
+                           make_pass_through_transform(C)),
+                make_tuple(Sequence<0, 1, 2, 3>{}, Sequence<4>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+        else
+        {
+            const index_t Z = b_g_k_c_xs_lengths[3];
+            const index_t Y = b_g_k_c_xs_lengths[4];
+            const index_t X = b_g_k_c_xs_lengths[5];
+
+            const index_t ConvDilationD = conv_filter_dilations[0];
+            const index_t ConvDilationH = conv_filter_dilations[1];
+            const index_t ConvDilationW = conv_filter_dilations[2];
+
+            const index_t InLeftPadD = input_left_pads[0];
+            const index_t InLeftPadH = input_left_pads[1];
+            const index_t InLeftPadW = input_left_pads[2];
+
+            const index_t InRightPadD = input_right_pads[0];
+            const index_t InRightPadH = input_right_pads[1];
+            const index_t InRightPadW = input_right_pads[2];
+
+            // This is different
+            const index_t NStride  = a_g_n_c_wis_strides[1];
+            const index_t DiStride = a_g_n_c_wis_strides[3];
+            const index_t HiStride = a_g_n_c_wis_strides[4];
+            const index_t WiStride = a_g_n_c_wis_strides[5];
+            const auto CStride     = I1;
+
+            const auto in_n_di_hi_wi_c_grid_desc = make_naive_tensor_descriptor(
+                make_tuple(N, Di, Hi, Wi, C),
+                make_tuple(NStride, DiStride, HiStride, WiStride, CStride));
+
+            const auto in_n_hip_wip_c_grid_desc = transform_tensor_descriptor(
+                in_n_di_hi_wi_c_grid_desc,
+                make_tuple(make_pass_through_transform(N),
+                           make_pad_transform(Di, InLeftPadD, InRightPadD),
+                           make_pad_transform(Hi, InLeftPadH, InRightPadH),
+                           make_pad_transform(Wi, InLeftPadW, InRightPadW),
+                           make_pass_through_transform(C)),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}));
+
+            const auto in_n_z_do_y_ho_x_wo_c_grid_desc = transform_tensor_descriptor(
+                in_n_hip_wip_c_grid_desc,
+                make_tuple(
+                    make_pass_through_transform(N),
+                    make_embed_transform(make_tuple(Z, Do), make_tuple(ConvDilationD, ConvStrideD)),
+                    make_embed_transform(make_tuple(Y, Ho), make_tuple(ConvDilationH, ConvStrideH)),
+                    make_embed_transform(make_tuple(X, Wo), make_tuple(ConvDilationW, ConvStrideW)),
+                    make_pass_through_transform(C)),
+                make_tuple(
+                    Sequence<0>{}, Sequence<1>{}, Sequence<2>{}, Sequence<3>{}, Sequence<4>{}),
+                make_tuple(Sequence<0>{},
+                           Sequence<1, 2>{},
+                           Sequence<3, 4>{},
+                           Sequence<5, 6>{},
+                           Sequence<7>{}));
+
+            const auto in_gemmmraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+                in_n_z_do_y_ho_x_wo_c_grid_desc,
+                make_tuple(make_merge_transform(make_tuple(N, Do, Ho, Wo)),
+                           make_merge_transform(make_tuple(Z, Y, X, C))),
+                make_tuple(Sequence<0, 2, 4, 6>{}, Sequence<1, 3, 5, 7>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+            const auto in_gemmm_gemmk_grid_desc =
+                matrix_padder.PadADescriptor_M_K(in_gemmmraw_gemmkraw_grid_desc);
+
+            return in_gemmm_gemmk_grid_desc;
+        }
+    }
+
+    template <typename BLay,
+              typename std::enable_if<is_same_v<BLay, tensor_layout::convolution::GKXC> ||
+                                          is_same_v<BLay, tensor_layout::convolution::GKYXC> ||
+                                          is_same_v<BLay, tensor_layout::convolution::GKZYXC>,
+                                      bool>::type = false>
+    static auto
+    MakeBGridDescriptor_N_K(const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* b_g_k_c_xs_strides */)
+    {
+        const index_t K = b_g_k_c_xs_lengths[1];
+        const index_t C = b_g_k_c_xs_lengths[2];
+
+        const index_t YX = std::accumulate(b_g_k_c_xs_lengths.begin() + 3,
+                                           b_g_k_c_xs_lengths.begin() + 3 + NDimSpatial,
+                                           index_t{1},
+                                           std::multiplies<index_t>());
+
+        const auto wei_k_yxc_grid_desc = make_naive_tensor_descriptor_packed(make_tuple(K, YX * C));
+
+        const auto wei_gemmn_gemmk_grid_desc =
+            matrix_padder.PadBDescriptor_N_K(wei_k_yxc_grid_desc);
+
+        return wei_gemmn_gemmk_grid_desc;
+    }
+
+    template <typename BLay,
+              typename std::enable_if<is_same_v<BLay, tensor_layout::convolution::G_K_X_C> ||
+                                          is_same_v<BLay, tensor_layout::convolution::G_K_YX_C> ||
+                                          is_same_v<BLay, tensor_layout::convolution::G_K_ZYX_C> ||
+                                          is_same_v<BLay, tensor_layout::convolution::KXGC> ||
+                                          is_same_v<BLay, tensor_layout::convolution::KYXGC> ||
+                                          is_same_v<BLay, tensor_layout::convolution::KZYXGC>,
+                                      bool>::type = false>
+    static auto
+    MakeBGridDescriptor_N_K(const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_strides)
+    {
+        const index_t K = b_g_k_c_xs_lengths[1];
+        const index_t C = b_g_k_c_xs_lengths[2];
+
+        const index_t YX = std::accumulate(b_g_k_c_xs_lengths.begin() + 3,
+                                           b_g_k_c_xs_lengths.begin() + 3 + NDimSpatial,
+                                           index_t{1},
+                                           std::multiplies<index_t>());
+
+        const index_t KStride = b_g_k_c_xs_strides[1];
+        const index_t XStride = b_g_k_c_xs_strides[2 + NDimSpatial];
+        const auto CStride    = I1;
+
+        const auto wei_k_yx_c_grid_desc = make_naive_tensor_descriptor(
+            make_tuple(K, YX, C), make_tuple(KStride, XStride, CStride));
+
+        const auto wei_gemmnraw_gemmkraw_grid_desc = transform_tensor_descriptor(
+            wei_k_yx_c_grid_desc,
+            make_tuple(make_pass_through_transform(K), make_merge_transform(make_tuple(YX, C))),
+            make_tuple(Sequence<0>{}, Sequence<1, 2>{}),
+            make_tuple(Sequence<0>{}, Sequence<1>{}));
+
+        const auto wei_gemmn_gemmk_grid_desc =
+            matrix_padder.PadBDescriptor_N_K(wei_gemmnraw_gemmkraw_grid_desc);
+
+        return wei_gemmn_gemmk_grid_desc;
+    }
+
+    template <typename ELay,
+              typename std::enable_if<is_same_v<ELay, tensor_layout::convolution::GNWK> ||
+                                          is_same_v<ELay, tensor_layout::convolution::GNHWK> ||
+                                          is_same_v<ELay, tensor_layout::convolution::GNDHWK>,
+                                      bool>::type = false>
+    static auto
+    MakeEGridDescriptor_M_N(const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& /* e_g_n_k_wos_strides */)
+    {
+        const index_t N = e_g_n_k_wos_lengths[1];
+        const index_t K = e_g_n_k_wos_lengths[2];
+
+        const index_t NHoWo = N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                                  e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                                  index_t{1},
+                                                  std::multiplies<index_t>());
+
+        const auto out_gemmmraw_gemmnraw_grid_desc =
+            make_naive_tensor_descriptor_packed(make_tuple(NHoWo, K));
+
+        const auto out_gemmm_gemmn_grid_desc =
+            matrix_padder.PadCDescriptor_M_N(out_gemmmraw_gemmnraw_grid_desc);
+
+        return out_gemmm_gemmn_grid_desc;
+    }
+
+    template <typename ELay,
+              typename std::enable_if<is_same_v<ELay, tensor_layout::convolution::G_NW_K> ||
+                                          is_same_v<ELay, tensor_layout::convolution::G_NHW_K> ||
+                                          is_same_v<ELay, tensor_layout::convolution::G_NDHW_K> ||
+                                          is_same_v<ELay, tensor_layout::convolution::NWGK> ||
+                                          is_same_v<ELay, tensor_layout::convolution::NHWGK> ||
+                                          is_same_v<ELay, tensor_layout::convolution::NDHWGK>,
+                                      bool>::type = false>
+    static auto
+    MakeEGridDescriptor_M_N(const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                            const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_strides)
+    {
+        const index_t N = e_g_n_k_wos_lengths[1];
+        const index_t K = e_g_n_k_wos_lengths[2];
+
+        const auto KStride     = I1;
+        const index_t WoStride = e_g_n_k_wos_strides[NDimSpatial + 2];
+
+        const index_t NHoWo = N * std::accumulate(e_g_n_k_wos_lengths.begin() + 3,
+                                                  e_g_n_k_wos_lengths.begin() + 3 + NDimSpatial,
+                                                  index_t{1},
+                                                  std::multiplies<index_t>());
+
+        const auto out_gemmmraw_gemmnraw_grid_desc =
+            make_naive_tensor_descriptor(make_tuple(NHoWo, K), make_tuple(WoStride, KStride));
+
+        const auto out_gemmm_gemmn_grid_desc =
+            matrix_padder.PadCDescriptor_M_N(out_gemmmraw_gemmnraw_grid_desc);
+
+        return out_gemmm_gemmn_grid_desc;
+    }
+
+    static auto MakeDsGridDescriptor_M_N(
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_lengths,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_strides)
+    {
+        return generate_tuple(
+            [&](auto i) {
+                using DLayout = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+
+                return DeviceOp::MakeEGridDescriptor_M_N<DLayout>(ds_g_n_k_wos_lengths[i],
+                                                                  ds_g_n_k_wos_strides[i]);
+            },
+            Number<NumDTensor>{});
+    }
+
+    using AGridDesc_M_K  = remove_cvref_t<decltype(
+        MakeAGridDescriptor_M_K<ALayout>({}, {}, {}, {}, {}, {}, {}, {}, {}, {}))>;
+    using BGridDesc_N_K  = remove_cvref_t<decltype(MakeBGridDescriptor_N_K<BLayout>({}, {}))>;
+    using DsGridDesc_M_N = remove_cvref_t<decltype(MakeDsGridDescriptor_M_N({}, {}))>;
+    using EGridDesc_M_N  = remove_cvref_t<decltype(MakeEGridDescriptor_M_N<ELayout>({}, {}))>;
+
+    // GridwiseGemm
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
+        ADataType, // TODO: distinguish A/B datatype
+        AccDataType,
+        CShuffleDataType,
+        DsDataType,
+        EDataType,
+        AElementwiseOperation,
+        BElementwiseOperation,
+        CDEElementwiseOperation,
+        InMemoryDataOperationEnum::Set,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        DsGridDesc_M_N,
+        EGridDesc_M_N,
+        NumGemmKPrefetchStage,
+        BlockSize,
+        MPerBlock,
+        NPerBlock,
+        KPerBlock,
+        AK1,
+        BK1,
+        MPerXDL,
+        NPerXDL,
+        MXdlPerWave,
+        NXdlPerWave,
+        ABlockTransferThreadClusterLengths_AK0_M_AK1,
+        ABlockTransferThreadClusterArrangeOrder,
+        ABlockTransferSrcAccessOrder,
+        ABlockTransferSrcVectorDim,
+        ABlockTransferSrcScalarPerVector,
+        ABlockTransferDstScalarPerVector_AK1,
+        false,
+        ABlockLdsExtraM,
+        BBlockTransferThreadClusterLengths_BK0_N_BK1,
+        BBlockTransferThreadClusterArrangeOrder,
+        BBlockTransferSrcAccessOrder,
+        BBlockTransferSrcVectorDim,
+        BBlockTransferSrcScalarPerVector,
+        BBlockTransferDstScalarPerVector_BK1,
+        false,
+        BBlockLdsExtraN,
+        CShuffleMXdlPerWavePerShuffle,
+        CShuffleNXdlPerWavePerShuffle,
+        CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
+        CDEBlockTransferScalarPerVector_NPerBlock,
+        LoopSched>;
+
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
+    using Block2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+
+    // Argument
+    struct Argument : public BaseArgument
+    {
+        Argument(const void* p_a,
+                 const void* p_b,
+                 const std::array<const void*, NumDTensor>& p_ds,
+                 void* p_e,
+                 const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+                 const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+                 const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+                 const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_strides,
+                 const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>&
+                     ds_g_n_k_wos_lengths,
+                 const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>&
+                     ds_g_n_k_wos_strides,
+                 const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+                 const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_strides,
+                 const std::array<index_t, NDimSpatial>& conv_filter_strides,
+                 const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+                 const std::array<index_t, NDimSpatial>& input_left_pads,
+                 const std::array<index_t, NDimSpatial>& input_right_pads,
+                 const AElementwiseOperation& a_element_op,
+                 const BElementwiseOperation& b_element_op,
+                 const CDEElementwiseOperation& cde_element_op)
+            : p_a_grid_{static_cast<const ADataType*>(p_a)},
+              p_b_grid_{static_cast<const BDataType*>(p_b)},
+              p_ds_grid_{},
+              p_e_grid_{static_cast<EDataType*>(p_e)},
+              a_grid_desc_m_k_{DeviceOp::MakeAGridDescriptor_M_K<ALayout>(a_g_n_c_wis_lengths,
+                                                                          a_g_n_c_wis_strides,
+                                                                          b_g_k_c_xs_lengths,
+                                                                          b_g_k_c_xs_strides,
+                                                                          e_g_n_k_wos_lengths,
+                                                                          e_g_n_k_wos_strides,
+                                                                          conv_filter_strides,
+                                                                          conv_filter_dilations,
+                                                                          input_left_pads,
+                                                                          input_right_pads)},
+              b_grid_desc_n_k_{DeviceOp::MakeBGridDescriptor_N_K<BLayout>(b_g_k_c_xs_lengths,
+                                                                          b_g_k_c_xs_strides)},
+              ds_grid_desc_m_n_{},
+              e_grid_desc_m_n_{DeviceOp::MakeEGridDescriptor_M_N<ELayout>(e_g_n_k_wos_lengths,
+                                                                          e_g_n_k_wos_strides)},
+              a_grid_desc_ak0_m_ak1_{
+                  GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k_)},
+              b_grid_desc_bk0_n_bk1_{
+                  GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k_)},
+              ds_grid_desc_mblock_mperblock_nblock_nperblock_{},
+              e_grid_desc_mblock_mperblock_nblock_nperblock_{},
+              block_2_etile_map_{GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n_)},
+              compute_ptr_offset_of_batch_{},
+              a_element_op_{a_element_op},
+              b_element_op_{b_element_op},
+              cde_element_op_{cde_element_op},
+              a_g_n_c_wis_lengths_{a_g_n_c_wis_lengths},
+              a_g_n_c_wis_strides_{a_g_n_c_wis_strides},
+              b_g_k_c_xs_lengths_{b_g_k_c_xs_lengths},
+              b_g_k_c_xs_strides_{b_g_k_c_xs_strides},
+              ds_g_n_k_wos_lengths_{ds_g_n_k_wos_lengths},
+              ds_g_n_k_wos_strides_{ds_g_n_k_wos_strides},
+              e_g_n_k_wos_lengths_{e_g_n_k_wos_lengths},
+              e_g_n_k_wos_strides_{e_g_n_k_wos_strides},
+              conv_filter_strides_{conv_filter_strides},
+              conv_filter_dilations_{conv_filter_dilations},
+              input_left_pads_{input_left_pads},
+              input_right_pads_{input_right_pads}
+        {
+            // A/B/E Batch Stride
+            compute_ptr_offset_of_batch_.BatchStrideA_ = a_g_n_c_wis_strides[0];
+            compute_ptr_offset_of_batch_.BatchStrideB_ = b_g_k_c_xs_strides[0];
+            compute_ptr_offset_of_batch_.BatchStrideE_ = e_g_n_k_wos_strides[0];
+
+            // populate pointer, batch stride, desc for Ds
+            static_for<0, NumDTensor, 1>{}([&](auto i) {
+                using DLayout   = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+                using DDataType = remove_cvref_t<tuple_element_t<i.value, DsDataType>>;
+
+                // D pointer
+                p_ds_grid_(i) = static_cast<const DDataType*>(p_ds[i]);
+
+                // D batch stride
+                compute_ptr_offset_of_batch_.BatchStrideDs_(i) = ds_g_n_k_wos_strides[i][0];
+
+                // D desc
+                ds_grid_desc_m_n_(i) = DeviceOp::MakeEGridDescriptor_M_N<DLayout>(
+                    ds_g_n_k_wos_lengths[i], ds_g_n_k_wos_strides[i]);
+            });
+
+            // populate desc for Ds/E
+            if(GridwiseGemm::CheckValidity(a_grid_desc_m_k_,
+                                           b_grid_desc_n_k_,
+                                           ds_grid_desc_m_n_,
+                                           e_grid_desc_m_n_,
+                                           block_2_etile_map_))
+            {
+                e_grid_desc_mblock_mperblock_nblock_nperblock_ =
+                    GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                        e_grid_desc_m_n_);
+
+                ds_grid_desc_mblock_mperblock_nblock_nperblock_ =
+                    GridwiseGemm::MakeDsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                        ds_grid_desc_m_n_);
+            }
+        }
+
+        void Print() const
+        {
+            std::cout << "A[M, K]: " << a_grid_desc_m_k_ << std::endl;
+            std::cout << "B[N, K]: " << b_grid_desc_n_k_ << std::endl;
+            static_for<0, NumDTensor, 1>{}(
+                [&](auto i) { std::cout << "Ds[M, N]: " << ds_grid_desc_m_n_[i] << std::endl; });
+            std::cout << "E[M, N]: " << e_grid_desc_m_n_ << std::endl;
+        }
+
+        //  private:
+        // pointers
+        const ADataType* p_a_grid_;
+        const BDataType* p_b_grid_;
+        typename GridwiseGemm::DsGridPointer p_ds_grid_;
+        EDataType* p_e_grid_;
+
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        DsGridDesc_M_N ds_grid_desc_m_n_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
+        AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
+        BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
+        typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            ds_grid_desc_mblock_mperblock_nblock_nperblock_;
+        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            e_grid_desc_mblock_mperblock_nblock_nperblock_;
+
+        // block-to-e-tile map
+        Block2ETileMap block_2_etile_map_;
+
+        ComputePtrOffsetOfStridedBatch<NumDTensor> compute_ptr_offset_of_batch_;
+
+        // element-wise op
+        AElementwiseOperation a_element_op_;
+        BElementwiseOperation b_element_op_;
+        CDEElementwiseOperation cde_element_op_;
+
+        // for checking IsSupportedArgument()
+        std::array<index_t, NDimSpatial + 3> a_g_n_c_wis_lengths_;
+        std::array<index_t, NDimSpatial + 3> a_g_n_c_wis_strides_;
+        std::array<index_t, NDimSpatial + 3> b_g_k_c_xs_lengths_;
+        std::array<index_t, NDimSpatial + 3> b_g_k_c_xs_strides_;
+        std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor> ds_g_n_k_wos_lengths_;
+        std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor> ds_g_n_k_wos_strides_;
+        std::array<index_t, NDimSpatial + 3> e_g_n_k_wos_lengths_;
+        std::array<index_t, NDimSpatial + 3> e_g_n_k_wos_strides_;
+        std::array<index_t, NDimSpatial> conv_filter_strides_;
+        std::array<index_t, NDimSpatial> conv_filter_dilations_;
+        std::array<index_t, NDimSpatial> input_left_pads_;
+        std::array<index_t, NDimSpatial> input_right_pads_;
+    };
+
+    // Invoker
+    struct Invoker : public BaseInvoker
+    {
+        using Argument = DeviceOp::Argument;
+
+        float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
+        {
+#if 1
+            arg.Print();
+#endif
+            if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                            arg.b_grid_desc_n_k_,
+                                            arg.ds_grid_desc_m_n_,
+                                            arg.e_grid_desc_m_n_,
+                                            arg.block_2_etile_map_))
+            {
+                throw std::runtime_error(
+                    "wrong! GridwiseGemmMultipleD_xdl_cshuffle has invalid setting");
+            }
+
+            const index_t grid_size =
+                arg.block_2_etile_map_.CalculateGridSize(arg.e_grid_desc_m_n_) *
+                arg.a_g_n_c_wis_lengths_[0]; // Group count
+
+            const auto K =
+                arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) * arg.a_grid_desc_ak0_m_ak1_.GetLength(I2);
+
+            auto launch_kernel = [&](auto has_main_k_block_loop) {
+                constexpr bool has_main_loop = has_main_k_block_loop.value;
+
+                const auto kernel = kernel_batch_gemm_multiple_d_xdl_cshuffle<
+                    GridwiseGemm,
+                    ADataType, // TODO: distiguish A/B datatype
+                    typename GridwiseGemm::DsGridPointer,
+                    EDataType,
+                    AElementwiseOperation,
+                    BElementwiseOperation,
+                    CDEElementwiseOperation,
+                    DeviceOp::AGridDesc_AK0_M_AK1,
+                    DeviceOp::BGridDesc_BK0_N_BK1,
+                    typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
+                    typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
+                    Block2ETileMap,
+                    ComputePtrOffsetOfStridedBatch<NumDTensor>,
+                    has_main_loop>;
+
+                return launch_and_time_kernel(stream_config,
+                                              kernel,
+                                              dim3(grid_size),
+                                              dim3(BlockSize),
+                                              0,
+                                              arg.p_a_grid_,
+                                              arg.p_b_grid_,
+                                              arg.p_ds_grid_,
+                                              arg.p_e_grid_,
+                                              arg.a_element_op_,
+                                              arg.b_element_op_,
+                                              arg.cde_element_op_,
+                                              arg.a_g_n_c_wis_lengths_[0], // Group count
+                                              arg.a_grid_desc_ak0_m_ak1_,
+                                              arg.b_grid_desc_bk0_n_bk1_,
+                                              arg.ds_grid_desc_mblock_mperblock_nblock_nperblock_,
+                                              arg.e_grid_desc_mblock_mperblock_nblock_nperblock_,
+                                              arg.block_2_etile_map_,
+                                              arg.compute_ptr_offset_of_batch_);
+            };
+
+            if(GridwiseGemm::CalculateHasMainKBlockLoop(K))
+            {
+                return launch_kernel(integral_constant<bool, true>{});
+            }
+            else
+            {
+                return launch_kernel(integral_constant<bool, false>{});
+            }
+        }
+
+        float Run(const BaseArgument* p_arg,
+                  const StreamConfig& stream_config = StreamConfig{}) override
+        {
+            return Run(*dynamic_cast<const Argument*>(p_arg), stream_config);
+        }
+    };
+
+    static bool IsSupportedArgument(const Argument& arg)
+    {
+        namespace ctc = tensor_layout::convolution;
+
+        // check device
+        if(get_device_name() == "gfx908")
+        {
+            if constexpr(!(is_same_v<AccDataType, float> || is_same_v<AccDataType, float> ||
+                           is_same_v<AccDataType, int32_t>))
+            {
+                return false;
+            }
+        }
+        else if(get_device_name() == "gfx90a")
+        {
+            if constexpr(!(is_same_v<AccDataType, float> || is_same_v<AccDataType, float> ||
+                           is_same_v<AccDataType, int32_t> || is_same_v<AccDataType, double>))
+            {
+                return false;
+            }
+        }
+        else
+        {
+            return false;
+        }
+
+        // check ConvolutionForwardSpecialization
+        if constexpr(ConvForwardSpecialization ==
+                     ConvolutionForwardSpecialization::Filter1x1Stride1Pad0)
+        {
+            // check if it's 1x1, stride=1 conv
+            for(index_t i = 0; i < NDimSpatial; ++i)
+            {
+                const index_t X          = arg.b_g_k_c_xs_lengths_[i + 2];
+                const index_t ConvStride = arg.conv_filter_strides_[i];
+                const index_t LeftPad    = arg.input_left_pads_[i];
+                const index_t RightPad   = arg.input_right_pads_[i];
+
+                if(!(X == 1 && ConvStride == 1 && LeftPad == 0 && RightPad == 0))
+                {
+                    return false;
+                }
+            }
+        }
+        else if constexpr(ConvForwardSpecialization ==
+                          ConvolutionForwardSpecialization::Filter1x1Pad0)
+        {
+            // check if it's 1x1 conv
+            for(index_t i = 0; i < NDimSpatial; ++i)
+            {
+                const index_t X        = arg.b_g_k_c_xs_lengths_[i + 2];
+                const index_t LeftPad  = arg.input_left_pads_[i];
+                const index_t RightPad = arg.input_right_pads_[i];
+
+                if(!(X == 1 && LeftPad == 0 && RightPad == 0))
+                {
+                    return false;
+                }
+            }
+        }
+
+        // check vector access of A
+        // FIXME: layout
+        if constexpr(is_same_v<ALayout, ctc::G_NW_C> || is_same_v<ALayout, ctc::G_NHW_C> ||
+                     is_same_v<ALayout, ctc::G_NDHW_C> || is_same_v<ALayout, ctc::GNWC> ||
+                     is_same_v<ALayout, ctc::GNHWC> || is_same_v<ALayout, ctc::GNDHWC> ||
+                     is_same_v<ALayout, ctc::NWGC> || is_same_v<ALayout, ctc::NHWGC> ||
+                     is_same_v<ALayout, ctc::NDHWGC>)
+        {
+            const index_t C = arg.a_g_n_c_wis_lengths_[2];
+
+            if(!(ABlockTransferSrcVectorDim == 2 && C % ABlockTransferSrcScalarPerVector == 0))
+            {
+                return false;
+            }
+        }
+        else
+        {
+            return false;
+        }
+
+        // check vector access of B
+        // FIXME: layout
+        if constexpr(is_same_v<BLayout, ctc::G_K_X_C> || is_same_v<BLayout, ctc::G_K_YX_C> ||
+                     is_same_v<BLayout, ctc::G_K_ZYX_C> || is_same_v<BLayout, ctc::GKXC> ||
+                     is_same_v<BLayout, ctc::GKYXC> || is_same_v<BLayout, ctc::GKZYXC> ||
+                     is_same_v<BLayout, ctc::KXGC> || is_same_v<BLayout, ctc::KYXGC> ||
+                     is_same_v<BLayout, ctc::KZYXGC>)
+
+        {
+            const index_t C = arg.b_g_k_c_xs_lengths_[2];
+
+            if(!(BBlockTransferSrcVectorDim == 2 && C % BBlockTransferSrcScalarPerVector == 0))
+            {
+                return false;
+            }
+        }
+        else
+        {
+            return false;
+        }
+
+        //  check vector access of Ds
+        bool valid = true;
+
+        static_for<0, NumDTensor, 1>{}([&](auto i) {
+            using DLayout = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+
+            // FIXME: layout
+            if constexpr(is_same_v<DLayout, ctc::G_NW_K> || is_same_v<DLayout, ctc::G_NHW_K> ||
+                         is_same_v<DLayout, ctc::G_NDHW_K> || is_same_v<DLayout, ctc::GNWK> ||
+                         is_same_v<DLayout, ctc::GNHWK> || is_same_v<DLayout, ctc::GNDHWK> ||
+                         is_same_v<DLayout, ctc::NWGK> || is_same_v<DLayout, ctc::NHWGK> ||
+                         is_same_v<DLayout, ctc::NDHWGK>)
+            {
+                const index_t K = arg.ds_g_n_k_wos_lengths_[i][2];
+
+                if(!(K % CDEBlockTransferScalarPerVector_NPerBlock == 0))
+                {
+                    valid = false;
+                }
+            }
+            else
+            {
+                valid = false;
+            }
+        });
+
+        if(!valid)
+        {
+            return false;
+        }
+
+        // check vector access of E
+        if constexpr(is_same_v<ELayout, ctc::G_NW_K> || is_same_v<ELayout, ctc::G_NHW_K> ||
+                     is_same_v<ELayout, ctc::G_NDHW_K> || is_same_v<ELayout, ctc::GNWK> ||
+                     is_same_v<ELayout, ctc::GNHWK> || is_same_v<ELayout, ctc::GNDHWK> ||
+                     is_same_v<ELayout, ctc::NWGK> || is_same_v<ELayout, ctc::NHWGK> ||
+                     is_same_v<ELayout, ctc::NDHWGK>)
+        {
+            const index_t K = arg.e_g_n_k_wos_lengths_[2];
+
+            if(!(K % CDEBlockTransferScalarPerVector_NPerBlock == 0))
+            {
+                return false;
+            }
+        }
+        else
+        {
+            return false;
+        }
+
+        // check Gridwise GEMM
+        return GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_,
+                                           arg.b_grid_desc_n_k_,
+                                           arg.ds_grid_desc_m_n_,
+                                           arg.e_grid_desc_m_n_,
+                                           arg.block_2_etile_map_);
+    }
+
+    bool IsSupportedArgument(const BaseArgument* p_arg) override
+    {
+        return IsSupportedArgument(*dynamic_cast<const Argument*>(p_arg));
+    }
+
+    static auto MakeArgument(
+        const void* p_a,
+        const void* p_b,
+        const std::array<const void*, NumDTensor>& p_ds,
+        void* p_e,
+        const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+        const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+        const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+        const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_strides,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_lengths,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_strides,
+        const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+        const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_strides,
+        const std::array<index_t, NDimSpatial>& conv_filter_strides,
+        const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+        const std::array<index_t, NDimSpatial>& input_left_pads,
+        const std::array<index_t, NDimSpatial>& input_right_pads,
+        const AElementwiseOperation& a_element_op,
+        const BElementwiseOperation& b_element_op,
+        const CDEElementwiseOperation& cde_element_op)
+    {
+        return Argument{p_a,
+                        p_b,
+                        p_ds,
+                        p_e,
+                        a_g_n_c_wis_lengths,
+                        a_g_n_c_wis_strides,
+                        b_g_k_c_xs_lengths,
+                        b_g_k_c_xs_strides,
+                        ds_g_n_k_wos_lengths,
+                        ds_g_n_k_wos_strides,
+                        e_g_n_k_wos_lengths,
+                        e_g_n_k_wos_strides,
+                        conv_filter_strides,
+                        conv_filter_dilations,
+                        input_left_pads,
+                        input_right_pads,
+                        a_element_op,
+                        b_element_op,
+                        cde_element_op};
+    }
+
+    static auto MakeInvoker() { return Invoker{}; }
+
+    std::unique_ptr<BaseArgument> MakeArgumentPointer(
+        const void* p_a,
+        const void* p_b,
+        const std::array<const void*, NumDTensor>& p_ds,
+        void* p_e,
+        const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths,
+        const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,
+        const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths,
+        const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_strides,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_lengths,
+        const std::array<std::array<index_t, NDimSpatial + 3>, NumDTensor>& ds_g_n_k_wos_strides,
+        const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths,
+        const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_strides,
+        const std::array<index_t, NDimSpatial>& conv_filter_strides,
+        const std::array<index_t, NDimSpatial>& conv_filter_dilations,
+        const std::array<index_t, NDimSpatial>& input_left_pads,
+        const std::array<index_t, NDimSpatial>& input_right_pads,
+        const AElementwiseOperation& a_element_op,
+        const BElementwiseOperation& b_element_op,
+        const CDEElementwiseOperation& cde_element_op) override
+    {
+        return std::make_unique<Argument>(p_a,
+                                          p_b,
+                                          p_ds,
+                                          p_e,
+                                          a_g_n_c_wis_lengths,
+                                          a_g_n_c_wis_strides,
+                                          b_g_k_c_xs_lengths,
+                                          b_g_k_c_xs_strides,
+                                          ds_g_n_k_wos_lengths,
+                                          ds_g_n_k_wos_strides,
+                                          e_g_n_k_wos_lengths,
+                                          e_g_n_k_wos_strides,
+                                          conv_filter_strides,
+                                          conv_filter_dilations,
+                                          input_left_pads,
+                                          input_right_pads,
+                                          a_element_op,
+                                          b_element_op,
+                                          cde_element_op);
+    }
+
+    std::unique_ptr<BaseInvoker> MakeInvokerPointer() override
+    {
+        return std::make_unique<Invoker>(Invoker{});
+    }
+
+    std::string GetTypeString() const override
+    {
+        auto str = std::stringstream();
+
+        // clang-format off
+        str << "DeviceGroupedConvFwdMultipleD_Xdl_CShuffle"
+            << "<"
+            << BlockSize << ", "
+            << MPerBlock << ", "
+            << NPerBlock << ", "
+            << KPerBlock << ", "
+            << getConvForwardSpecializationString(ConvForwardSpecialization)
+            << ">";
+        // clang-format on
+
+        return str.str();
+    }
+};
+
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_grouped_gemm.hpp b/include/ck/tensor_operation/gpu/device/device_grouped_gemm.hpp
index 57398c96a5..181ee4b428 100644
--- a/include/ck/tensor_operation/gpu/device/device_grouped_gemm.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_grouped_gemm.hpp
@@ -18,7 +18,8 @@ struct GemmDesc
 
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename DsLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
           typename DsDataType,
@@ -30,6 +31,8 @@ struct DeviceGroupedGemm : public BaseOperator
 {
     static constexpr index_t NumDTensor = DsDataType::Size();
 
+    static_assert(DsLayout::Size() == DsDataType::Size(), "wrong! inconsisiten NumDTensor");
+
     virtual std::unique_ptr<BaseArgument>
     MakeArgumentPointer(std::vector<const void*>& p_a,
                         std::vector<const void*>& p_b,
@@ -43,27 +46,6 @@ struct DeviceGroupedGemm : public BaseOperator
     virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
 };
 
-template <typename ALayout,
-          typename BLayout,
-          typename DELayout,
-          typename ADataType,
-          typename BDataType,
-          typename DsDataType,
-          typename EDataType,
-          typename AElementwiseOperation,
-          typename BElementwiseOperation,
-          typename CElementwiseOperation>
-using DeviceGroupedGemmPtr = std::unique_ptr<DeviceGroupedGemm<ALayout,
-                                                               BLayout,
-                                                               DELayout,
-                                                               ADataType,
-                                                               BDataType,
-                                                               DsDataType,
-                                                               EDataType,
-                                                               AElementwiseOperation,
-                                                               BElementwiseOperation,
-                                                               CElementwiseOperation>>;
-
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp b/include/ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp
index 642cf01e00..abdfd078cf 100644
--- a/include/ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp
@@ -13,9 +13,10 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/device_grouped_gemm.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -72,11 +73,11 @@ __global__ void
         a_element_op,
         b_element_op,
         c_element_op,
-        gemm_desc_ptr[group_id].a_grid_desc_k0_m_k1_,
-        gemm_desc_ptr[group_id].b_grid_desc_k0_n_k1_,
+        gemm_desc_ptr[group_id].a_grid_desc_ak0_m_ak1_,
+        gemm_desc_ptr[group_id].b_grid_desc_bk0_n_bk1_,
         gemm_desc_ptr[group_id].ds_grid_desc_mblock_mperblock_nblock_nperblock_,
         gemm_desc_ptr[group_id].e_grid_desc_mblock_mperblock_nblock_nperblock_,
-        gemm_desc_ptr[group_id].block_2_ctile_map_);
+        gemm_desc_ptr[group_id].block_2_etile_map_);
 #else
     ignore = gemm_descs_const;
     ignore = group_count;
@@ -88,10 +89,11 @@ __global__ void
 
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename DsLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
-          typename GemmAccDataType,
+          typename AccDataType,
           typename CShuffleDataType,
           typename DsDataType,
           typename EDataType,
@@ -116,37 +118,43 @@ template <typename ALayout,
           ck::index_t ABlockTransferSrcVectorDim,
           ck::index_t ABlockTransferSrcScalarPerVector,
           ck::index_t ABlockTransferDstScalarPerVector_K1,
-          bool ABlockLdsAddExtraM,
+          bool ABlockLdsExtraM,
           typename BBlockTransferThreadClusterLengths_K0_N_K1,
           typename BBlockTransferThreadClusterArrangeOrder,
           typename BBlockTransferSrcAccessOrder,
           ck::index_t BBlockTransferSrcVectorDim,
           ck::index_t BBlockTransferSrcScalarPerVector,
           ck::index_t BBlockTransferDstScalarPerVector_K1,
-          bool BBlockLdsAddExtraN,
+          bool BBlockLdsExtraN,
           index_t CShuffleMXdlPerWavePerShuffle,
           index_t CShuffleNXdlPerWavePerShuffle,
           typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CDEBlockTransferScalarPerVector_NPerBlock,
           LoopScheduler LoopSched = make_default_loop_scheduler()>
-struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
-                                                       BLayout,
-                                                       DELayout,
-                                                       ADataType,
-                                                       BDataType,
-                                                       DsDataType,
-                                                       EDataType,
-                                                       AElementwiseOperation,
-                                                       BElementwiseOperation,
-                                                       CDEElementwiseOperation>
+struct DeviceGroupedGemm_Xdl : public DeviceGroupedGemm<ALayout,
+                                                        BLayout,
+                                                        DsLayout,
+                                                        ELayout,
+                                                        ADataType,
+                                                        BDataType,
+                                                        DsDataType,
+                                                        EDataType,
+                                                        AElementwiseOperation,
+                                                        BElementwiseOperation,
+                                                        CDEElementwiseOperation>
 {
+    using DeviceOp = DeviceGroupedGemm_Xdl;
+
     static constexpr index_t NumDTensor = DsDataType::Size();
 
     static constexpr auto I0 = Number<0>{};
     static constexpr auto I1 = Number<1>{};
     static constexpr auto I2 = Number<2>{};
 
-    static auto MakeAGridDescriptor_AK0_M_AK1(index_t MRaw, index_t KRaw, index_t StrideA)
+    static constexpr auto matrix_padder =
+        MatrixPadder<GemmSpec, index_t, index_t, index_t>{MPerBlock, NPerBlock, KPerBlock};
+
+    static auto MakeAGridDescriptor_M_K(index_t MRaw, index_t KRaw, index_t StrideA)
     {
         const auto a_grid_desc_mraw_kraw = [&]() {
             if constexpr(is_same_v<tensor_layout::gemm::RowMajor, ALayout>)
@@ -161,95 +169,10 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both M and K
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(M)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad M, but not K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_right_pad_transform(MRaw, MPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad K, but not M
-            assert(K % AK1 == 0);
-
-            const auto AK0 = K / AK1;
-
-            const auto a_grid_desc_m_k = transform_tensor_descriptor(
-                a_grid_desc_mraw_kraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_m_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
-        else
-        {
-            // not pad M or K
-            assert(KRaw % AK1 == 0);
-
-            const auto AK0 = KRaw / AK1;
-
-            const auto a_grid_desc_ak0_m_ak1 =
-                transform_tensor_descriptor(a_grid_desc_mraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
-                                                       make_pass_through_transform(MRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return a_grid_desc_ak0_m_ak1;
-        }
+        return matrix_padder.PadADescriptor_M_K(a_grid_desc_mraw_kraw);
     }
 
-    static auto MakeBGridDescriptor_BK0_N_BK1(index_t KRaw, index_t NRaw, index_t StrideB)
+    static auto MakeBGridDescriptor_N_K(index_t KRaw, index_t NRaw, index_t StrideB)
     {
         const auto b_grid_desc_nraw_kraw = [&]() {
             if constexpr(is_same<tensor_layout::gemm::RowMajor, BLayout>::value)
@@ -264,160 +187,50 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
             }
         }();
 
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-        const auto K = math::integer_divide_ceil(KRaw, KPerBlock) * KPerBlock;
-
-        const auto NPad = N - NRaw;
-        const auto KPad = K - KRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::NKPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad both N and K
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_right_pad_transform(NRaw, NPad),
-                                                       make_right_pad_transform(KRaw, KPad)),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                            make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(N)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::MNPadding)
-        {
-            // pad N, but not K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_right_pad_transform(NRaw, NPad)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad K, but not N
-            assert(K % BK1 == 0);
-
-            const auto BK0 = K / BK1;
-
-            const auto b_grid_desc_n_k = transform_tensor_descriptor(
-                b_grid_desc_nraw_kraw,
-                make_tuple(make_pass_through_transform(NRaw), make_right_pad_transform(KRaw, KPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_n_k,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
-        else
-        {
-            // not pad N or K
-            assert(KRaw % BK1 == 0);
-
-            const auto BK0 = KRaw / BK1;
-
-            const auto b_grid_desc_bk0_n_bk1 =
-                transform_tensor_descriptor(b_grid_desc_nraw_kraw,
-                                            make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
-                                                       make_pass_through_transform(NRaw)),
-                                            make_tuple(Sequence<1>{}, Sequence<0>{}),
-                                            make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
-
-            return b_grid_desc_bk0_n_bk1;
-        }
+        return matrix_padder.PadBDescriptor_N_K(b_grid_desc_nraw_kraw);
     }
 
+    template <typename ELay>
     static auto MakeEGridDescriptor_M_N(index_t MRaw, index_t NRaw, index_t StrideE)
     {
-        const auto c_grid_desc_mraw_nraw = [&]() {
-            if constexpr(is_same<tensor_layout::gemm::RowMajor, DELayout>::value)
+        const auto e_grid_desc_mraw_nraw = [&]() {
+            if constexpr(is_same<tensor_layout::gemm::RowMajor, ELay>::value)
             {
                 return make_naive_tensor_descriptor(make_tuple(MRaw, NRaw),
                                                     make_tuple(StrideE, I1));
             }
-            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, DELayout>::value)
+            else if constexpr(is_same<tensor_layout::gemm::ColumnMajor, ELay>::value)
             {
                 return make_naive_tensor_descriptor(make_tuple(MRaw, NRaw),
                                                     make_tuple(I1, StrideE));
             }
         }();
 
-        const auto M = math::integer_divide_ceil(MRaw, MPerBlock) * MPerBlock;
-        const auto N = math::integer_divide_ceil(NRaw, NPerBlock) * NPerBlock;
-
-        const auto MPad = M - MRaw;
-        const auto NPad = N - NRaw;
-
-        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
-                     GemmSpec == GemmSpecialization::MNKPadding)
-        {
-            // pad M and N
-            return transform_tensor_descriptor(c_grid_desc_mraw_nraw,
-                                               make_tuple(make_right_pad_transform(MRaw, MPad),
-                                                          make_right_pad_transform(NRaw, NPad)),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
-                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
-                          GemmSpec == GemmSpecialization::MKPadding)
-        {
-            // pad M, but not N
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(NRaw)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
-                          GemmSpec == GemmSpecialization::NKPadding)
-        {
-            // pad N, but not M
-            return transform_tensor_descriptor(
-                c_grid_desc_mraw_nraw,
-                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(NRaw, NPad)),
-                make_tuple(Sequence<0>{}, Sequence<1>{}),
-                make_tuple(Sequence<0>{}, Sequence<1>{}));
-        }
-        else
-        {
-            // not pad M or N
-            return c_grid_desc_mraw_nraw;
-        }
+        return matrix_padder.PadCDescriptor_M_N(e_grid_desc_mraw_nraw);
     }
 
-    using AGridDesc_AK0_M_AK1 = decltype(MakeAGridDescriptor_AK0_M_AK1(1, 1, 1));
-    using BGridDesc_BK0_N_BK1 = decltype(MakeBGridDescriptor_BK0_N_BK1(1, 1, 1));
-    using EGridDesc_M_N       = decltype(MakeEGridDescriptor_M_N(1, 1, 1));
+    static auto MakeDsGridDescriptor_M_N(const std::array<index_t, NumDTensor>& MRaws,
+                                         const std::array<index_t, NumDTensor>& NRaws,
+                                         const std::array<index_t, NumDTensor>& DsStride)
+    {
+        return generate_tuple(
+            [&](auto i) {
+                using DLayout = remove_cvref_t<tuple_element_t<i.value, DsLayout>>;
+
+                return DeviceOp::MakeEGridDescriptor_M_N<DLayout>(MRaws[i], NRaws[i], DsStride[i]);
+            },
+            Number<NumDTensor>{});
+    }
+
+    using AGridDesc_M_K  = decltype(MakeAGridDescriptor_M_K(1, 1, 1));
+    using BGridDesc_N_K  = decltype(MakeBGridDescriptor_N_K(1, 1, 1));
+    using DsGridDesc_M_N = remove_cvref_t<decltype(MakeDsGridDescriptor_M_N({}, {}, {}))>;
+    using EGridDesc_M_N  = decltype(MakeEGridDescriptor_M_N<ELayout>(1, 1, 1));
 
     // GridwiseGemm
-    using GridwiseGemm = GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle<
+    using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
         ADataType, // TODO: distinguish A/B datatype
-        GemmAccDataType,
+        AccDataType,
         CShuffleDataType,
         DsDataType,
         EDataType,
@@ -425,8 +238,9 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
         BElementwiseOperation,
         CDEElementwiseOperation,
         InMemoryDataOperationEnum::Set,
-        AGridDesc_AK0_M_AK1,
-        BGridDesc_BK0_N_BK1,
+        AGridDesc_M_K,
+        BGridDesc_N_K,
+        DsGridDesc_M_N,
         EGridDesc_M_N,
         NumPrefetch, // NumGemmKPrefetchStage
         BlockSize,
@@ -446,7 +260,7 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
         ABlockTransferSrcScalarPerVector,
         ABlockTransferDstScalarPerVector_K1,
         false, // AThreadTransferSrcResetCoordinateAfterRun,
-        ABlockLdsAddExtraM,
+        ABlockLdsExtraM,
         BBlockTransferThreadClusterLengths_K0_N_K1,
         BBlockTransferThreadClusterArrangeOrder,
         BBlockTransferSrcAccessOrder,
@@ -454,76 +268,87 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
         BBlockTransferSrcScalarPerVector,
         BBlockTransferDstScalarPerVector_K1,
         false, // BThreadTransferSrcResetCoordinateAfterRun,
-        BBlockLdsAddExtraN,
+        BBlockLdsExtraN,
         CShuffleMXdlPerWavePerShuffle,
         CShuffleNXdlPerWavePerShuffle,
         CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
         CDEBlockTransferScalarPerVector_NPerBlock,
         LoopSched>;
 
+    using AGridDesc_AK0_M_AK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using BGridDesc_BK0_N_BK1 = remove_cvref_t<decltype(
+        GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+
     struct GroupedGemmBlock2ETileMap
     {
-        using UnderlyingBlock2CTileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+        using UnderlyingBlock2ETileMap = typename GridwiseGemm::DefaultBlock2ETileMap;
+
         static_assert(
             std::is_same<decltype(GridwiseGemm::MakeDefaultBlock2ETileMap(EGridDesc_M_N{})),
                          typename GridwiseGemm::DefaultBlock2ETileMap>::value,
             "Wrong! Should be the same type name");
+
         GroupedGemmBlock2ETileMap()
         {
-            block_2_ctile_map_ = GridwiseGemm::MakeDefaultBlock2ETileMap(EGridDesc_M_N{});
+            block_2_etile_map_ = GridwiseGemm::MakeDefaultBlock2ETileMap(EGridDesc_M_N{});
             BlockStart_        = -1;
         }
 
-        GroupedGemmBlock2ETileMap(const EGridDesc_M_N& c_grid_desc_m_n, ck::index_t BlockStart)
+        GroupedGemmBlock2ETileMap(const EGridDesc_M_N& e_grid_desc_m_n, ck::index_t BlockStart)
         {
-            block_2_ctile_map_ = GridwiseGemm::MakeDefaultBlock2ETileMap(c_grid_desc_m_n);
+            block_2_etile_map_ = GridwiseGemm::MakeDefaultBlock2ETileMap(e_grid_desc_m_n);
             BlockStart_        = BlockStart;
         }
 
         template <typename TopIdx>
         __host__ __device__ constexpr auto CalculateBottomIndex(const TopIdx& idx_top) const
         {
-            return block_2_ctile_map_.CalculateBottomIndex(
+            return block_2_etile_map_.CalculateBottomIndex(
                 make_multi_index(idx_top[I0] - BlockStart_));
         }
 
+        // it's actually E-Tile
         template <typename CTileIdx, typename CTileDim>
         __host__ __device__ bool ValidCTileIndex(const CTileIdx& c_tile_idx,
                                                  const CTileDim& c_tile_dim) const
         {
-            return block_2_ctile_map_.ValidCTileIndex(c_tile_idx, c_tile_dim);
+            return block_2_etile_map_.ValidCTileIndex(c_tile_idx, c_tile_dim);
         }
 
-        __host__ bool CheckValidity(const EGridDesc_M_N& c_grid_desc_m_n) const
+        __host__ bool CheckValidity(const EGridDesc_M_N& e_grid_desc_m_n) const
         {
-            return block_2_ctile_map_.CheckValidity(c_grid_desc_m_n);
+            return block_2_etile_map_.CheckValidity(e_grid_desc_m_n);
         }
 
-        typename GridwiseGemm::DefaultBlock2ETileMap block_2_ctile_map_;
+        typename GridwiseGemm::DefaultBlock2ETileMap block_2_etile_map_;
         ck::index_t BlockStart_;
     };
 
     struct GemmBiasTransKernelArg
     {
-        AGridDesc_AK0_M_AK1 a_grid_desc_k0_m_k1_;
-        BGridDesc_BK0_N_BK1 b_grid_desc_k0_n_k1_;
-        EGridDesc_M_N e_grid_desc_m_n_;
-
-        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
-            e_grid_desc_mblock_mperblock_nblock_nperblock_;
-
-        StaticallyIndexedArray<
-            typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-            NumDTensor>
-            ds_grid_desc_mblock_mperblock_nblock_nperblock_; // FIXME: Ds desc may be of different
-
-        GroupedGemmBlock2ETileMap block_2_ctile_map_;
-
+        // pointers
         const ADataType* a_ptr_;
         const BDataType* b_ptr_;
         typename GridwiseGemm::DsGridPointer ds_ptr_;
         EDataType* e_ptr_;
 
+        // tensor descriptors for problem definiton
+        AGridDesc_M_K a_grid_desc_m_k_;
+        BGridDesc_N_K b_grid_desc_n_k_;
+        DsGridDesc_M_N ds_grid_desc_m_n_;
+        EGridDesc_M_N e_grid_desc_m_n_;
+
+        // tensor descriptors for block/thread-wise copy
+        AGridDesc_AK0_M_AK1 a_grid_desc_ak0_m_ak1_;
+        BGridDesc_BK0_N_BK1 b_grid_desc_bk0_n_bk1_;
+        typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            ds_grid_desc_mblock_mperblock_nblock_nperblock_;
+        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+            e_grid_desc_mblock_mperblock_nblock_nperblock_;
+
+        // block-to-e-tile map
+        GroupedGemmBlock2ETileMap block_2_etile_map_;
         ck::index_t BlockStart_, BlockEnd_;
     };
 
@@ -563,66 +388,85 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
                 const index_t StrideB = gemm_descs[i].stride_B_;
                 const index_t StrideC = gemm_descs[i].stride_C_;
 
-                const auto a_grid_desc_k0_m_k1_ =
-                    DeviceGroupedGemmXdl::MakeAGridDescriptor_AK0_M_AK1(M, K, StrideA);
-                const auto b_grid_desc_k0_n_k1_ =
-                    DeviceGroupedGemmXdl::MakeBGridDescriptor_BK0_N_BK1(K, N, StrideB);
+                // pointer
+                typename GridwiseGemm::DsGridPointer p_ds_grid{};
 
-                const auto e_grid_desc_m_n_ =
-                    DeviceGroupedGemmXdl::MakeEGridDescriptor_M_N(M, N, StrideC);
+                static_for<0, NumDTensor, 1>{}([&](auto j) {
+                    using DDataType = remove_cvref_t<tuple_element_t<j.value, DsDataType>>;
+
+                    p_ds_grid(j) = static_cast<const DDataType*>(p_Ds[i][j]);
+                });
+
+                // tensor descriptors for problem definiton
+                const auto a_grid_desc_m_k = DeviceOp::MakeAGridDescriptor_M_K(M, K, StrideA);
+                const auto b_grid_desc_n_k = DeviceOp::MakeBGridDescriptor_N_K(K, N, StrideB);
+
+                DsGridDesc_M_N ds_grid_desc_m_n;
+
+                static_for<0, NumDTensor, 1>{}([&](auto j) {
+                    using DLayout = remove_cvref_t<tuple_element_t<j.value, DsLayout>>;
+
+                    ds_grid_desc_m_n(j) = DeviceOp::MakeEGridDescriptor_M_N<DLayout>(
+                        M, N, gemm_descs[i].stride_Ds_[j]);
+                });
+
+                const auto e_grid_desc_m_n =
+                    DeviceOp::MakeEGridDescriptor_M_N<ELayout>(M, N, StrideC);
+
+                // tensor descriptors for block/thread-wise copy
+                const auto a_grid_desc_ak0_m_ak1 =
+                    GridwiseGemm::MakeDefaultAGridDescriptor_AK0_M_AK1(a_grid_desc_m_k);
+
+                const auto b_grid_desc_bk0_n_bk1 =
+                    GridwiseGemm::MakeDefaultBGridDescriptor_BK0_N_BK1(b_grid_desc_n_k);
 
                 const index_t grid_size_grp =
-                    GroupedGemmBlock2ETileMap(e_grid_desc_m_n_, 0)
-                        .block_2_ctile_map_.CalculateGridSize(e_grid_desc_m_n_);
+                    GroupedGemmBlock2ETileMap(e_grid_desc_m_n, 0)
+                        .block_2_etile_map_.CalculateGridSize(e_grid_desc_m_n);
 
                 const index_t BlockStart = grid_size_;
                 const index_t BlockEnd   = grid_size_ + grid_size_grp;
 
                 grid_size_ += grid_size_grp;
 
-                const auto block_2_ctile_map_ =
-                    GroupedGemmBlock2ETileMap(e_grid_desc_m_n_, BlockStart);
+                // block-to-e-tile map
+                const auto block_2_etile_map =
+                    GroupedGemmBlock2ETileMap(e_grid_desc_m_n, BlockStart);
 
-                if(GridwiseGemm::CheckValidity(a_grid_desc_k0_m_k1_,
-                                               b_grid_desc_k0_n_k1_,
-                                               e_grid_desc_m_n_,
-                                               block_2_ctile_map_))
+                if(GridwiseGemm::CheckValidity(a_grid_desc_m_k,
+                                               b_grid_desc_n_k,
+                                               ds_grid_desc_m_n,
+                                               e_grid_desc_m_n,
+                                               block_2_etile_map))
                 {
-                    auto e_grid_desc_mblock_mperblock_nblock_nperblock_ =
-                        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
-                            e_grid_desc_m_n_);
-                    StaticallyIndexedArray<
-                        typename GridwiseGemm::EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-                        NumDTensor>
-                        ds_grid_desc_mblock_mperblock_nblock_nperblock_; // FIXME: Ds desc may be of
-                                                                         // different
-
-                    typename GridwiseGemm::DsGridPointer p_ds_grid_{};
+                    // tensor descriptors for block/thread-wise copy
+                    typename GridwiseGemm::DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock
+                        ds_grid_desc_mblock_mperblock_nblock_nperblock;
 
                     static_for<0, NumDTensor, 1>{}([&](auto j) {
-                        using DDataType = remove_cvref_t<tuple_element_t<j.value, DsDataType>>;
-
-                        p_ds_grid_(j) = static_cast<const DDataType*>(p_Ds[i][j]);
-
-                        const auto d_grid_desc_m_n = DeviceGroupedGemmXdl::MakeEGridDescriptor_M_N(
-                            M, N, gemm_descs[i].stride_Ds_[j]);
-
-                        ds_grid_desc_mblock_mperblock_nblock_nperblock_(j) =
+                        ds_grid_desc_mblock_mperblock_nblock_nperblock(j) =
                             GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
-                                d_grid_desc_m_n);
+                                ds_grid_desc_m_n[j]);
                     });
 
+                    const auto e_grid_desc_mblock_mperblock_nblock_nperblock =
+                        GridwiseGemm::MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+                            e_grid_desc_m_n);
+
                     gemm_desc_kernel_arg_.push_back(
-                        GemmBiasTransKernelArg{a_grid_desc_k0_m_k1_,
-                                               b_grid_desc_k0_n_k1_,
-                                               e_grid_desc_m_n_,
-                                               e_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                               ds_grid_desc_mblock_mperblock_nblock_nperblock_,
-                                               block_2_ctile_map_,
-                                               static_cast<const ADataType*>(p_As[i]),
+                        GemmBiasTransKernelArg{static_cast<const ADataType*>(p_As[i]),
                                                static_cast<const BDataType*>(p_Bs[i]),
-                                               p_ds_grid_,
+                                               p_ds_grid,
                                                static_cast<EDataType*>(p_Es[i]),
+                                               a_grid_desc_m_k,
+                                               b_grid_desc_n_k,
+                                               ds_grid_desc_m_n,
+                                               e_grid_desc_m_n,
+                                               a_grid_desc_ak0_m_ak1,
+                                               b_grid_desc_bk0_n_bk1,
+                                               ds_grid_desc_mblock_mperblock_nblock_nperblock,
+                                               e_grid_desc_mblock_mperblock_nblock_nperblock,
+                                               block_2_etile_map,
                                                BlockStart,
                                                BlockEnd});
                 }
@@ -643,7 +487,7 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
     // Invoker
     struct Invoker : public BaseInvoker
     {
-        using Argument = DeviceGroupedGemmXdl::Argument;
+        using Argument = DeviceOp::Argument;
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
@@ -651,32 +495,39 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
 
             for(std::size_t i = 0; i < arg.gemm_desc_kernel_arg_.size(); i++)
             {
-                std::cout << "group: " << i << " arg.a_grid_desc_k0_m_k1_{"
-                          << arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_.GetLength(I0) << ", "
-                          << arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_.GetLength(I1) << ", "
-                          << arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_.GetLength(I2) << "}";
+                std::cout << "group: " << i << " arg.a_grid_desc_ak0_m_ak1_{"
+                          << arg.gemm_desc_kernel_arg_[i].a_grid_desc_ak0_m_ak1_.GetLength(I0)
+                          << ", "
+                          << arg.gemm_desc_kernel_arg_[i].a_grid_desc_ak0_m_ak1_.GetLength(I1)
+                          << ", "
+                          << arg.gemm_desc_kernel_arg_[i].a_grid_desc_ak0_m_ak1_.GetLength(I2)
+                          << "}";
 
-                std::cout << ", arg.b_grid_desc_k0_n_k1_{"
-                          << arg.gemm_desc_kernel_arg_[i].b_grid_desc_k0_n_k1_.GetLength(I0) << ", "
-                          << arg.gemm_desc_kernel_arg_[i].b_grid_desc_k0_n_k1_.GetLength(I1) << ", "
-                          << arg.gemm_desc_kernel_arg_[i].b_grid_desc_k0_n_k1_.GetLength(I2) << "}";
+                std::cout << ", arg.b_grid_desc_bk0_n_bk1_{"
+                          << arg.gemm_desc_kernel_arg_[i].b_grid_desc_bk0_n_bk1_.GetLength(I0)
+                          << ", "
+                          << arg.gemm_desc_kernel_arg_[i].b_grid_desc_bk0_n_bk1_.GetLength(I1)
+                          << ", "
+                          << arg.gemm_desc_kernel_arg_[i].b_grid_desc_bk0_n_bk1_.GetLength(I2)
+                          << "}";
 
                 std::cout << ", arg.e_grid_desc_m_n_{ "
                           << arg.gemm_desc_kernel_arg_[i].e_grid_desc_m_n_.GetLength(I0) << ", "
                           << arg.gemm_desc_kernel_arg_[i].e_grid_desc_m_n_.GetLength(I1) << "}"
                           << std::endl;
 
-                if(!GridwiseGemm::CheckValidity(arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_,
-                                                arg.gemm_desc_kernel_arg_[i].b_grid_desc_k0_n_k1_,
+                if(!GridwiseGemm::CheckValidity(arg.gemm_desc_kernel_arg_[i].a_grid_desc_m_k_,
+                                                arg.gemm_desc_kernel_arg_[i].b_grid_desc_n_k_,
+                                                arg.gemm_desc_kernel_arg_[i].ds_grid_desc_m_n_,
                                                 arg.gemm_desc_kernel_arg_[i].e_grid_desc_m_n_,
-                                                arg.gemm_desc_kernel_arg_[i].block_2_ctile_map_))
+                                                arg.gemm_desc_kernel_arg_[i].block_2_etile_map_))
                 {
                     throw std::runtime_error(
                         "wrong! GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3 has invalid setting");
                 }
 
-                const auto K = arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_.GetLength(I0) *
-                               arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_.GetLength(I2);
+                const auto K = arg.gemm_desc_kernel_arg_[i].a_grid_desc_ak0_m_ak1_.GetLength(I0) *
+                               arg.gemm_desc_kernel_arg_[i].a_grid_desc_ak0_m_ak1_.GetLength(I2);
 
                 if(GridwiseGemm::CalculateHasMainKBlockLoop(K) != has_main_k_block_loop)
                 {
@@ -733,18 +584,14 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
         }
     };
 
-    static constexpr bool IsValidCompilationParameter()
-    {
-        // TODO: properly implement this check
-        return true;
-    }
-
     static bool IsSupportedArgument(const Argument& arg)
     {
         if(ck::type_convert<ck::index_t>(arg.gemm_desc_kernel_arg_.size()) != arg.group_count_)
+        {
             return false;
-        else
-            return true;
+        }
+
+        return true;
     }
 
     // polymorphic
@@ -795,7 +642,7 @@ struct DeviceGroupedGemmXdl : public DeviceGroupedGemm<ALayout,
         auto str = std::stringstream();
 
         // clang-format off
-        str << "DeviceGroupedGemmXdl"
+        str << "DeviceGroupedGemm_Xdl"
             << "<"
             << BlockSize << ", "
             << MPerBlock << ", "
diff --git a/include/ck/tensor_operation/gpu/device/device_layernorm.hpp b/include/ck/tensor_operation/gpu/device/device_layernorm.hpp
index e7bb0116b3..d4c771c007 100644
--- a/include/ck/tensor_operation/gpu/device/device_layernorm.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_layernorm.hpp
@@ -13,8 +13,8 @@
 #include "ck/tensor_operation/gpu/device/device_reduce_common.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_layernorm.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_set_buffer_value.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_pool2d_fwd_nhwc_nhwc.hpp b/include/ck/tensor_operation/gpu/device/device_pool2d_fwd_nhwc_nhwc.hpp
index 3edf9bd3af..bfde40cda2 100644
--- a/include/ck/tensor_operation/gpu/device/device_pool2d_fwd_nhwc_nhwc.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_pool2d_fwd_nhwc_nhwc.hpp
@@ -11,8 +11,8 @@
 #include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"
 #include "ck/tensor_operation/gpu/device/device_pool2d_fwd.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_2d_reduction_threadwise.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_reduce_multiblock.hpp b/include/ck/tensor_operation/gpu/device/device_reduce_multiblock.hpp
index a903fc415e..328b002b77 100644
--- a/include/ck/tensor_operation/gpu/device/device_reduce_multiblock.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_reduce_multiblock.hpp
@@ -14,8 +14,8 @@
 #include "ck/tensor_operation/gpu/device/device_reduce_common.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_2d_reduction_multiblock.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_set_buffer_value.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_reduce_threadwise.hpp b/include/ck/tensor_operation/gpu/device/device_reduce_threadwise.hpp
index d916954951..2f0f239997 100644
--- a/include/ck/tensor_operation/gpu/device/device_reduce_threadwise.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_reduce_threadwise.hpp
@@ -6,8 +6,8 @@
 #include <iostream>
 #include <sstream>
 
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 #include "ck/tensor_operation/gpu/device/device_reduce.hpp"
 #include "ck/tensor_operation/gpu/device/device_reduce_common.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_2d_reduction_multiblock.hpp"
diff --git a/include/ck/tensor_operation/gpu/device/device_softmax.hpp b/include/ck/tensor_operation/gpu/device/device_softmax.hpp
index 6a5dfc4da4..7fd4c4d1b3 100644
--- a/include/ck/tensor_operation/gpu/device/device_softmax.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_softmax.hpp
@@ -14,8 +14,8 @@
 #include "ck/tensor_operation/gpu/device/device_reduce_common.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_softmax.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_set_buffer_value.hpp"
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/include/ck/tensor_operation/gpu/device/device_unary_elementwise.hpp b/include/ck/tensor_operation/gpu/device/device_unary_elementwise.hpp
index 054245429d..0e67ede13c 100644
--- a/include/ck/tensor_operation/gpu/device/device_unary_elementwise.hpp
+++ b/include/ck/tensor_operation/gpu/device/device_unary_elementwise.hpp
@@ -6,8 +6,8 @@
 #include <iostream>
 #include <vector>
 
-#include "ck/device_utility/device_prop.hpp"
-#include "ck/device_utility/kernel_launch.hpp"
+#include "ck/host_utility/device_prop.hpp"
+#include "ck/host_utility/kernel_launch.hpp"
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_unary_elementwise_1d.hpp"
 
diff --git a/include/ck/tensor_operation/gpu/device/matrix_padder.hpp b/include/ck/tensor_operation/gpu/device/matrix_padder.hpp
new file mode 100644
index 0000000000..3bb89eb130
--- /dev/null
+++ b/include/ck/tensor_operation/gpu/device/matrix_padder.hpp
@@ -0,0 +1,184 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include "ck/utility/common_header.hpp"
+#include "ck/tensor_description/tensor_descriptor.hpp"
+#include "ck/tensor_description/tensor_descriptor_helper.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+
+// M/N/KPerTileType could be index_t or Number<>
+template <GemmSpecialization GemmSpec,
+          typename MPerTileType,
+          typename NPerTileType,
+          typename KPerTileType>
+struct MatrixPadder
+{
+    static constexpr auto I0 = Number<0>{};
+    static constexpr auto I1 = Number<1>{};
+    static constexpr auto I2 = Number<2>{};
+    static constexpr auto I3 = Number<3>{};
+
+    template <typename ADesc_MRaw_KRaw>
+    __host__ __device__ constexpr auto
+    PadADescriptor_M_K(const ADesc_MRaw_KRaw& a_desc_mraw_kraw) const
+    {
+        const auto MRaw = a_desc_mraw_kraw.GetLength(I0);
+        const auto KRaw = a_desc_mraw_kraw.GetLength(I1);
+
+        const auto M = math::integer_divide_ceil(MRaw, MPerTile_) * MPerTile_;
+        const auto K = math::integer_divide_ceil(KRaw, KPerTile_) * KPerTile_;
+
+        const auto MPad = M - MRaw;
+        const auto KPad = K - KRaw;
+
+        if constexpr(GemmSpec == GemmSpecialization::MKPadding ||
+                     GemmSpec == GemmSpecialization::MNKPadding)
+        {
+            // pad both M and K
+            return transform_tensor_descriptor(a_desc_mraw_kraw,
+                                               make_tuple(make_right_pad_transform(MRaw, MPad),
+                                                          make_right_pad_transform(KRaw, KPad)),
+                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
+                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
+                          GemmSpec == GemmSpecialization::MNPadding)
+        {
+            // pad M, but not K
+            return transform_tensor_descriptor(
+                a_desc_mraw_kraw,
+                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(KRaw)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
+                          GemmSpec == GemmSpecialization::NKPadding)
+        {
+            // pad K, but not M
+            return transform_tensor_descriptor(
+                a_desc_mraw_kraw,
+                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(KRaw, KPad)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else
+        {
+            // not pad M or K
+            return a_desc_mraw_kraw;
+        }
+    }
+
+    template <typename BDesc_NRaw_KRaw>
+    __host__ __device__ constexpr auto
+    PadBDescriptor_N_K(const BDesc_NRaw_KRaw& b_desc_nraw_kraw) const
+    {
+        const auto NRaw = b_desc_nraw_kraw.GetLength(I0);
+        const auto KRaw = b_desc_nraw_kraw.GetLength(I1);
+
+        const auto N = math::integer_divide_ceil(NRaw, NPerTile_) * NPerTile_;
+        const auto K = math::integer_divide_ceil(KRaw, KPerTile_) * KPerTile_;
+
+        const auto NPad = N - NRaw;
+        const auto KPad = K - KRaw;
+
+        if constexpr(GemmSpec == GemmSpecialization::NKPadding ||
+                     GemmSpec == GemmSpecialization::MNKPadding)
+        {
+            // pad both N and K
+            return transform_tensor_descriptor(b_desc_nraw_kraw,
+                                               make_tuple(make_right_pad_transform(NRaw, NPad),
+                                                          make_right_pad_transform(KRaw, KPad)),
+                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
+                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
+                          GemmSpec == GemmSpecialization::MNPadding)
+        {
+            // pad N, but not K
+            return transform_tensor_descriptor(
+                b_desc_nraw_kraw,
+                make_tuple(make_right_pad_transform(NRaw, NPad), make_pass_through_transform(KRaw)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::KPadding ||
+                          GemmSpec == GemmSpecialization::MKPadding)
+        {
+            // pad K, but not N
+            return transform_tensor_descriptor(
+                b_desc_nraw_kraw,
+                make_tuple(make_pass_through_transform(NRaw), make_right_pad_transform(KRaw, KPad)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else
+        {
+            // not pad N or K
+            return b_desc_nraw_kraw;
+        }
+    }
+
+    template <typename CDesc_MRaw_NRaw>
+    __host__ __device__ constexpr auto
+    PadCDescriptor_M_N(const CDesc_MRaw_NRaw& c_desc_mraw_nraw) const
+    {
+        const auto MRaw = c_desc_mraw_nraw.GetLength(I0);
+        const auto NRaw = c_desc_mraw_nraw.GetLength(I1);
+
+        const auto M = math::integer_divide_ceil(MRaw, MPerTile_) * MPerTile_;
+        const auto N = math::integer_divide_ceil(NRaw, NPerTile_) * NPerTile_;
+
+        const auto MPad = M - MRaw;
+        const auto NPad = N - NRaw;
+
+        if constexpr(GemmSpec == GemmSpecialization::MNPadding ||
+                     GemmSpec == GemmSpecialization::MNKPadding)
+        {
+            // pad M and N
+            return transform_tensor_descriptor(c_desc_mraw_nraw,
+                                               make_tuple(make_right_pad_transform(MRaw, MPad),
+                                                          make_right_pad_transform(NRaw, NPad)),
+                                               make_tuple(Sequence<0>{}, Sequence<1>{}),
+                                               make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::MPadding ||
+                          GemmSpec == GemmSpecialization::MKPadding)
+        {
+            // pad M, but not N
+            return transform_tensor_descriptor(
+                c_desc_mraw_nraw,
+                make_tuple(make_right_pad_transform(MRaw, MPad), make_pass_through_transform(NRaw)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else if constexpr(GemmSpec == GemmSpecialization::NPadding ||
+                          GemmSpec == GemmSpecialization::NKPadding)
+        {
+            // pad N, but not M
+            return transform_tensor_descriptor(
+                c_desc_mraw_nraw,
+                make_tuple(make_pass_through_transform(MRaw), make_right_pad_transform(NRaw, NPad)),
+                make_tuple(Sequence<0>{}, Sequence<1>{}),
+                make_tuple(Sequence<0>{}, Sequence<1>{}));
+        }
+        else
+        {
+            // not pad M or N
+            return c_desc_mraw_nraw;
+        }
+    }
+
+    MPerTileType MPerTile_;
+    NPerTileType NPerTile_;
+    KPerTileType KPerTile_;
+};
+
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/include/ck/tensor_operation/gpu/device/tensor_layout.hpp b/include/ck/tensor_operation/gpu/device/tensor_layout.hpp
index 40c7eb7d5e..7b5eef51a9 100644
--- a/include/ck/tensor_operation/gpu/device/tensor_layout.hpp
+++ b/include/ck/tensor_operation/gpu/device/tensor_layout.hpp
@@ -25,86 +25,117 @@ struct ColumnMajor : public BaseTensorLayout
 
 namespace convolution {
 
-// 1D Conv
-struct NWC : public BaseTensorLayout
-{
-    static constexpr const char* name = "NWC";
-};
-
-struct KXC : public BaseTensorLayout
-{
-    static constexpr const char* name = "KXC";
-};
-
-struct NWK : public BaseTensorLayout
-{
-    static constexpr const char* name = "NWK";
-};
-
+// input tensor
+// packed NCW/NCHW/NCDHW
 struct NCW : public BaseTensorLayout
 {
     static constexpr const char* name = "NCW";
 };
 
-struct KCX : public BaseTensorLayout
-{
-    static constexpr const char* name = "KCX";
-};
-
-struct NKW : public BaseTensorLayout
-{
-    static constexpr const char* name = "NKW";
-};
-
-// 2D Conv
-struct NHWC : public BaseTensorLayout
-{
-    static constexpr const char* name = "NHWC";
-};
-
-struct KYXC : public BaseTensorLayout
-{
-    static constexpr const char* name = "KYXC";
-};
-
-struct NHWK : public BaseTensorLayout
-{
-    static constexpr const char* name = "NHWK";
-};
-
 struct NCHW : public BaseTensorLayout
 {
     static constexpr const char* name = "NCHW";
 };
 
-struct KCYX : public BaseTensorLayout
+struct NCDHW : public BaseTensorLayout
 {
-    static constexpr const char* name = "KCYX";
+    static constexpr const char* name = "NCDHW";
 };
 
-struct NKHW : public BaseTensorLayout
+// packed GNCW/GNCHW/GNCDHW
+struct GNCW : public BaseTensorLayout
 {
-    static constexpr const char* name = "NKHW";
+    static constexpr const char* name = "GNCW";
+};
+
+struct GNCHW : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNCHW";
+};
+
+struct GNCDHW : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNCDHW";
+};
+
+// input tensor
+// packed NWC/NHWC/NDHWC
+struct NWC : public BaseTensorLayout
+{
+    static constexpr const char* name = "NWC";
+};
+
+struct NHWC : public BaseTensorLayout
+{
+    static constexpr const char* name = "NHWC";
 };
 
-// 3D Conv
 struct NDHWC : public BaseTensorLayout
 {
     static constexpr const char* name = "NDHWC";
 };
 
-struct KZYXC : public BaseTensorLayout
+// input tensor
+// packed GNWC/GNHWC/GNDHWC
+struct GNWC : public BaseTensorLayout
 {
-    static constexpr const char* name = "KZYXC";
+    static constexpr const char* name = "GNWC";
 };
 
-struct NDHWK : public BaseTensorLayout
+struct GNHWC : public BaseTensorLayout
 {
-    static constexpr const char* name = "NDHWK";
+    static constexpr const char* name = "GNHWC";
 };
-struct NCDHW : public BaseTensorLayout
+
+struct GNDHWC : public BaseTensorLayout
 {
-    static constexpr const char* name = "NCDHW";
+    static constexpr const char* name = "GNDHWC";
+};
+
+// input tensor
+// packed GNWC/GNHWC/GNDHWC
+struct NWGC : public BaseTensorLayout
+{
+    static constexpr const char* name = "NWGC";
+};
+
+struct NHWGC : public BaseTensorLayout
+{
+    static constexpr const char* name = "NHWGC";
+};
+
+struct NDHWGC : public BaseTensorLayout
+{
+    static constexpr const char* name = "NDHWGC";
+};
+
+// input tensor
+// strided layout
+struct G_NW_C : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_NW_C";
+};
+
+struct G_NHW_C : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_NHW_C";
+};
+
+struct G_NDHW_C : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_NDHW_C";
+};
+
+// weight tensor
+// packed KCX/KCYX/KCZYX
+struct KCX : public BaseTensorLayout
+{
+    static constexpr const char* name = "KCX";
+};
+
+struct KCYX : public BaseTensorLayout
+{
+    static constexpr const char* name = "KCYX";
 };
 
 struct KCZYX : public BaseTensorLayout
@@ -112,11 +143,193 @@ struct KCZYX : public BaseTensorLayout
     static constexpr const char* name = "KCZYX";
 };
 
+// weight tensor
+// packed KCX/KCYX/KCZYX
+struct GKCX : public BaseTensorLayout
+{
+    static constexpr const char* name = "GKCX";
+};
+
+struct GKCYX : public BaseTensorLayout
+{
+    static constexpr const char* name = "GKCYX";
+};
+
+struct GKCZYX : public BaseTensorLayout
+{
+    static constexpr const char* name = "GKCZYX";
+};
+
+// weight tensor
+// packed KXC/KYXC/KZYXC
+struct KXC : public BaseTensorLayout
+{
+    static constexpr const char* name = "KXC";
+};
+
+struct KYXC : public BaseTensorLayout
+{
+    static constexpr const char* name = "KYXC";
+};
+
+struct KZYXC : public BaseTensorLayout
+{
+    static constexpr const char* name = "KZYXC";
+};
+
+// weight tensor
+// packed GKXC/GKYXC/GKZYXC
+struct GKXC : public BaseTensorLayout
+{
+    static constexpr const char* name = "GKXC";
+};
+
+struct GKYXC : public BaseTensorLayout
+{
+    static constexpr const char* name = "GKYXC";
+};
+
+struct GKZYXC : public BaseTensorLayout
+{
+    static constexpr const char* name = "GKZYXC";
+};
+
+// weight tensor
+// packed KXGC/KYXGC/KZYXGC
+struct KXGC : public BaseTensorLayout
+{
+    static constexpr const char* name = "KXGC";
+};
+
+struct KYXGC : public BaseTensorLayout
+{
+    static constexpr const char* name = "KYXGC";
+};
+
+struct KZYXGC : public BaseTensorLayout
+{
+    static constexpr const char* name = "KZYXGC";
+};
+
+// weight tensor
+// strided
+struct G_K_X_C : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_K_X_C";
+};
+
+struct G_K_YX_C : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_K_YX_C";
+};
+
+struct G_K_ZYX_C : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_K_ZYX_C";
+};
+
+// output tensor
+// packed NKW/NKHW/NKDHW
+struct NKW : public BaseTensorLayout
+{
+    static constexpr const char* name = "NKW";
+};
+
+struct NKHW : public BaseTensorLayout
+{
+    static constexpr const char* name = "NKHW";
+};
+
 struct NKDHW : public BaseTensorLayout
 {
     static constexpr const char* name = "NKDHW";
 };
 
+// output tensor
+// packed GNKW/GNKHW/GNKDHW
+struct GNKW : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNKW";
+};
+
+struct GNKHW : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNKHW";
+};
+
+struct GNKDHW : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNKDHW";
+};
+
+// output tensor
+// packed NWK/NHWK/NDHWK
+struct NWK : public BaseTensorLayout
+{
+    static constexpr const char* name = "NWK";
+};
+
+struct NHWK : public BaseTensorLayout
+{
+    static constexpr const char* name = "NHWK";
+};
+
+struct NDHWK : public BaseTensorLayout
+{
+    static constexpr const char* name = "NDHWK";
+};
+
+// output tensor
+// packed GNWK/GNHWK/GNDHWK
+struct GNWK : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNWK";
+};
+
+struct GNHWK : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNHWK";
+};
+
+struct GNDHWK : public BaseTensorLayout
+{
+    static constexpr const char* name = "GNDHWK";
+};
+
+// output tensor
+// packed NWGK/NHWGK/NDHWGK
+struct NWGK : public BaseTensorLayout
+{
+    static constexpr const char* name = "NWGK";
+};
+
+struct NHWGK : public BaseTensorLayout
+{
+    static constexpr const char* name = "NHWGK";
+};
+
+struct NDHWGK : public BaseTensorLayout
+{
+    static constexpr const char* name = "NDHWGK";
+};
+
+// output tensor
+// strided layout
+struct G_NW_K : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_NW_K";
+};
+
+struct G_NHW_K : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_NHW_K";
+};
+
+struct G_NDHW_K : public BaseTensorLayout
+{
+    static constexpr const char* name = "G_NDHW_K";
+};
+
 } // namespace convolution
 
 template <
diff --git a/include/ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp b/include/ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp
index ece1ecb865..0466702aba 100644
--- a/include/ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp
+++ b/include/ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp
@@ -104,6 +104,13 @@ struct Bilinear
         y = alpha_ * x0 + beta_ * x1;
     };
 
+    template <>
+    __host__ __device__ constexpr void
+    operator()<half_t, half_t, half_t>(half_t& y, const half_t& x0, const half_t& x1) const
+    {
+        y = type_convert<half_t>(alpha_) * x0 + type_convert<half_t>(beta_) * x1;
+    };
+
     template <>
     __host__ __device__ constexpr void
     operator()<half_t, float, half_t>(half_t& y, const float& x0, const half_t& x1) const
@@ -117,12 +124,12 @@ struct Bilinear
 
 struct AddRelu
 {
-    template <typename T>
-    __host__ __device__ constexpr void operator()(T& y, const T& x0, const T& x1) const;
+    template <typename Y, typename X0, typename X1>
+    __host__ __device__ constexpr void operator()(Y& y, const X0& x0, const X1& x1) const;
 
     template <>
     __host__ __device__ constexpr void
-    operator()<float>(float& y, const float& x0, const float& x1) const
+    operator()<float, float, float>(float& y, const float& x0, const float& x1) const
     {
         const float a = x0 + x1;
         y             = a > 0.0f ? a : 0.0f;
@@ -130,7 +137,7 @@ struct AddRelu
 
     template <>
     __host__ __device__ constexpr void
-    operator()<double>(double& y, const double& x0, const double& x1) const
+    operator()<double, double, double>(double& y, const double& x0, const double& x1) const
     {
         const double a = x0 + x1;
         y              = a > 0.0 ? a : 0.0;
@@ -138,11 +145,19 @@ struct AddRelu
 
     template <>
     __host__ __device__ constexpr void
-    operator()<half_t>(half_t& y, const half_t& x0, const half_t& x1) const
+    operator()<half_t, half_t, half_t>(half_t& y, const half_t& x0, const half_t& x1) const
     {
         const half_t a = x0 + x1;
         y              = a > type_convert<half_t>(0.0f) ? a : type_convert<half_t>(0.0f);
     };
+
+    template <>
+    __host__ __device__ constexpr void
+    operator()<half_t, float, half_t>(half_t& y, const float& x0, const half_t& x1) const
+    {
+        const float a = x0 + x1;
+        y             = a > type_convert<half_t>(0.0f) ? a : type_convert<half_t>(0.0f);
+    };
 };
 
 struct AddHardswish
diff --git a/include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp b/include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp
index 440f0de4d4..97e5d38feb 100644
--- a/include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp
+++ b/include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp
@@ -12,16 +12,65 @@ namespace element_wise {
 
 struct PassThrough
 {
-    template <typename T>
-    __host__ __device__ void operator()(T& y, const T& x) const
-    {
-        static_assert(is_same<T, float>::value || is_same<T, double>::value ||
-                          is_same<T, half_t>::value || is_same<T, bhalf_t>::value ||
-                          is_same<T, int32_t>::value || is_same<T, int8_t>::value,
-                      "Data type is not supported by this operation!");
+    template <typename Y, typename X>
+    __host__ __device__ void operator()(Y& y, const X& x) const;
 
+    template <>
+    __host__ __device__ void operator()<double, double>(double& y, const double& x) const
+    {
         y = x;
-    };
+    }
+
+    template <>
+    __host__ __device__ void operator()<float, float>(float& y, const float& x) const
+    {
+        y = x;
+    }
+
+    template <>
+    __host__ __device__ void operator()<half_t, half_t>(half_t& y, const half_t& x) const
+    {
+        y = x;
+    }
+
+    template <>
+    __host__ __device__ void operator()<bhalf_t, bhalf_t>(bhalf_t& y, const bhalf_t& x) const
+    {
+        y = x;
+    }
+
+    template <>
+    __host__ __device__ void operator()<int32_t, int32_t>(int32_t& y, const int32_t& x) const
+    {
+        y = x;
+    }
+
+    template <>
+    __host__ __device__ void operator()<bhalf_t, float>(bhalf_t& y, const float& x) const
+    {
+        y = type_convert<bhalf_t>(x);
+    }
+
+    template <>
+    __host__ __device__ void operator()<int8_t, int8_t>(int8_t& y, const int8_t& x) const
+    {
+        y = x;
+    }
+
+    template <>
+    __host__ __device__ void operator()<int8_t, int32_t>(int8_t& y, const int32_t& x) const
+    {
+        y = type_convert<int8_t>(x);
+    }
+};
+
+struct UnaryConvert
+{
+    template <typename Y, typename X>
+    __host__ __device__ void operator()(Y& y, const X& x) const
+    {
+        y = type_convert<Y>(x);
+    }
 };
 
 struct Scale
diff --git a/include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp
index 5ce7db0a97..4656ed439d 100644
--- a/include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp
+++ b/include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_d_xdl_cshuffle.hpp
@@ -18,25 +18,26 @@
 namespace ck {
 
 // GEMM:
-//   input : A[AK0, M, AK1]
-//   input : B[AK0, N, AK1]
+//   input : A[M, K]
+//   input : B[N, K]
 //   input : D0[M, N], D1[M, N], ...
 //   output : E[M, N]
 //   C = a_op(A) * b_op(B)
 //   E = cde_op(C, D0, D1, ...)
 // Assume:
 //   D0, D1, ... and E have the same layout
-template <typename FloatAB,
-          typename FloatGemmAcc,
-          typename FloatCShuffle,
+template <typename ABDataType, // FIXME: don't assume A/B have same datatype
+          typename AccDataType,
+          typename CShuffleDataType,
           typename DsDataType,
-          typename FloatE,
+          typename EDataType,
           typename AElementwiseOperation,
           typename BElementwiseOperation,
           typename CDEElementwiseOperation,
           InMemoryDataOperationEnum EGlobalMemoryDataOperation,
-          typename AGridDesc_AK0_M_AK1,
-          typename BGridDesc_BK0_N_BK1,
+          typename AGridDesc_M_K,
+          typename BGridDesc_N_K,
+          typename DsGridDesc_M_N,
           typename EGridDesc_M_N,
           index_t NumGemmKPrefetchStage,
           index_t BlockSize,
@@ -70,7 +71,7 @@ template <typename FloatAB,
           typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CDEShuffleBlockTransferScalarPerVector_NPerBlock,
           LoopScheduler LoopSched>
-struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
+struct GridwiseGemmMultipleD_xdl_cshuffle
 {
     static constexpr index_t NumDTensor = DsDataType::Size();
 
@@ -84,10 +85,10 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
     static constexpr auto I7 = Number<7>{};
 
     // K1 should be Number<...>
-    static constexpr auto AK0 = Number<KPerBlock / AK1Value>{};
-    static constexpr auto BK0 = Number<KPerBlock / BK1Value>{};
-    static constexpr auto AK1 = Number<AK1Value>{};
-    static constexpr auto BK1 = Number<BK1Value>{};
+    static constexpr auto AK1         = Number<AK1Value>{};
+    static constexpr auto BK1         = Number<BK1Value>{};
+    static constexpr auto AK0PerBlock = Number<KPerBlock / AK1Value>{};
+    static constexpr auto BK0PerBlock = Number<KPerBlock / BK1Value>{};
 
     using ThisThreadBlock = ThisThreadBlock<BlockSize>;
 
@@ -97,7 +98,7 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
     {
         // A matrix in LDS memory, dst of blockwise copy
         return make_naive_tensor_descriptor(
-            make_tuple(AK0, Number<MPerBlock>{}, AK1),
+            make_tuple(AK0PerBlock, Number<MPerBlock>{}, AK1),
             make_tuple(Number<MPerBlock + ABlockLdsExtraM>{} * AK1, AK1, I1));
     }
 
@@ -105,7 +106,7 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
     {
         // B matrix in LDS memory, dst of blockwise copy
         return make_naive_tensor_descriptor(
-            make_tuple(BK0, Number<NPerBlock>{}, BK1),
+            make_tuple(BK0PerBlock, Number<NPerBlock>{}, BK1),
             make_tuple(Number<NPerBlock + BBlockLdsExtraN>{} * BK1, BK1, I1));
     }
 
@@ -160,58 +161,46 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
             c_shuffle_block_desc_mblock_mperblock_nblock_nperblock.GetElementSpaceSize();
 
         return math::max((a_block_space_size_aligned + b_block_space_size_aligned) *
-                             sizeof(FloatAB),
-                         c_block_size * sizeof(FloatCShuffle));
-    }
-
-    // block_id to matrix tile idx (m0, n0) mapping are controlled by {M01, N01}
-    template <typename Block2ETileMap>
-    __host__ __device__ static constexpr bool
-    CheckValidity(const AGridDesc_AK0_M_AK1& a_grid_desc_ak0_m_ak1,
-                  const BGridDesc_BK0_N_BK1& b_grid_desc_bk0_n_bk1,
-                  const EGridDesc_M_N& e_grid_desc_m_n,
-                  const Block2ETileMap& block_2_etile_map)
-    {
-        static_assert((MPerBlock % (MPerXdl * MXdlPerWave) == 0) &&
-                          (NPerBlock % (NXdlPerWave * NPerXdl)) == 0,
-                      "Invalid tuning param!");
-
-        const auto M = a_grid_desc_ak0_m_ak1.GetLength(I1);
-        const auto N = b_grid_desc_bk0_n_bk1.GetLength(I1);
-        const auto K = a_grid_desc_ak0_m_ak1.GetLength(I0) * a_grid_desc_ak0_m_ak1.GetLength(I2);
-
-        if(!(M == e_grid_desc_m_n.GetLength(I0) && N == e_grid_desc_m_n.GetLength(I1)))
-            return false;
-
-        if(!(M % MPerBlock == 0 && N % NPerBlock == 0 && K % KPerBlock == 0))
-            return false;
-
-        // check gridwise gemm pipeline
-        const auto num_k_loop = K / KPerBlock;
-
-        if(!GridwiseGemmPipe::IsSupported(num_k_loop))
-        {
-            return false;
-        }
-
-        if(!block_2_etile_map.CheckValidity(e_grid_desc_m_n))
-        {
-            return false;
-        }
-
-        // TODO: also check validity of all components (blockwise-copy, threadwise-copy, etc)
-        return true;
-    }
-
-    __host__ __device__ static constexpr bool CalculateHasMainKBlockLoop(index_t K)
-    {
-        const index_t num_loop = K / KPerBlock;
-
-        return GridwiseGemmPipe::CalculateHasMainLoop(num_loop);
+                             sizeof(ABDataType),
+                         c_block_size * sizeof(CShuffleDataType));
     }
 
+    // A desc for source in blockwise copy
     __host__ __device__ static constexpr auto
-    MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(const EGridDesc_M_N& e_grid_desc_m_n)
+    MakeDefaultAGridDescriptor_AK0_M_AK1(const AGridDesc_M_K& a_grid_desc_m_k)
+    {
+        const auto M = a_grid_desc_m_k.GetLength(I0);
+        const auto K = a_grid_desc_m_k.GetLength(I1);
+
+        const auto AK0 = K / AK1;
+
+        return transform_tensor_descriptor(a_grid_desc_m_k,
+                                           make_tuple(make_unmerge_transform(make_tuple(AK0, AK1)),
+                                                      make_pass_through_transform(M)),
+                                           make_tuple(Sequence<1>{}, Sequence<0>{}),
+                                           make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
+    }
+
+    // B desc for source in blockwise copy
+    __host__ __device__ static constexpr auto
+    MakeDefaultBGridDescriptor_BK0_N_BK1(const BGridDesc_N_K& b_grid_desc_n_k)
+    {
+        const auto N = b_grid_desc_n_k.GetLength(I0);
+        const auto K = b_grid_desc_n_k.GetLength(I1);
+
+        const auto BK0 = K / BK1;
+
+        return transform_tensor_descriptor(b_grid_desc_n_k,
+                                           make_tuple(make_unmerge_transform(make_tuple(BK0, BK1)),
+                                                      make_pass_through_transform(N)),
+                                           make_tuple(Sequence<1>{}, Sequence<0>{}),
+                                           make_tuple(Sequence<0, 2>{}, Sequence<1>{}));
+    }
+
+    // E desc for destination in blockwise copy
+    template <typename EGridDescriptor_M_N>
+    __host__ __device__ static constexpr auto MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+        const EGridDescriptor_M_N& e_grid_desc_m_n)
     {
         const auto M = e_grid_desc_m_n.GetLength(I0);
         const auto N = e_grid_desc_m_n.GetLength(I1);
@@ -229,6 +218,19 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
         return e_grid_desc_mblock_mperblock_nblock_nperblock;
     }
 
+    // Ds desc for source in blockwise copy
+    template <typename DsGridDescriptor_M_N>
+    __host__ __device__ static constexpr auto
+    MakeDsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(
+        const DsGridDescriptor_M_N& ds_grid_desc_m_n)
+    {
+        return generate_tuple(
+            [&](auto i) {
+                return MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(ds_grid_desc_m_n[i]);
+            },
+            Number<NumDTensor>{});
+    }
+
     // return block_id to E matrix tile idx (m0, n0) mapping
     __host__ __device__ static constexpr auto
     MakeDefaultBlock2ETileMap(const EGridDesc_M_N& e_grid_desc_m_n)
@@ -237,33 +239,114 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
             e_grid_desc_m_n);
     }
 
-    using EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock = remove_cvref_t<decltype(
+    // block_id to matrix tile idx (m0, n0) mapping are controlled by {M01, N01}
+    template <typename Block2ETileMap>
+    __host__ __device__ static constexpr bool CheckValidity(const AGridDesc_M_K& a_grid_desc_m_k,
+                                                            const BGridDesc_N_K& b_grid_desc_n_k,
+                                                            const DsGridDesc_M_N& ds_grid_desc_m_n,
+                                                            const EGridDesc_M_N& e_grid_desc_m_n,
+                                                            const Block2ETileMap& block_2_etile_map)
+    {
+        static_assert((MPerBlock % (MPerXdl * MXdlPerWave) == 0) &&
+                          (NPerBlock % (NXdlPerWave * NPerXdl)) == 0,
+                      "Invalid tuning param!");
+
+        const auto M = a_grid_desc_m_k.GetLength(I0);
+        const auto N = b_grid_desc_n_k.GetLength(I0);
+        const auto K = a_grid_desc_m_k.GetLength(I1);
+
+        // check consistency of desc
+        if(!(M == e_grid_desc_m_n.GetLength(I0) && N == e_grid_desc_m_n.GetLength(I1)))
+        {
+            return false;
+        }
+
+        bool valid = true;
+
+        static_for<0, NumDTensor, 1>{}([&](auto i) {
+            valid = valid && (M == ds_grid_desc_m_n[i].GetLength(I0) &&
+                              N == ds_grid_desc_m_n[i].GetLength(I1));
+        });
+
+        if(!valid)
+        {
+            return false;
+        }
+
+        // check tile size
+        if(!(M % MPerBlock == 0 && N % NPerBlock == 0 && K % KPerBlock == 0))
+        {
+            return false;
+        }
+
+        // check gridwise gemm pipeline
+        const auto num_k_loop = K / KPerBlock;
+
+        if(!GridwiseGemmPipe::IsSupported(num_k_loop))
+        {
+            return false;
+        }
+
+        // check block-to-E-tile
+        if(!block_2_etile_map.CheckValidity(e_grid_desc_m_n))
+        {
+            return false;
+        }
+
+        // TODO: also check validity of all components (blockwise-copy, threadwise-copy, etc)
+        // check tensor size: cannot be larger than 2GB each
+        constexpr long_index_t TwoGB = (long_index_t{1} << 31);
+
+        if(!(a_grid_desc_m_k.GetElementSpaceSize() * sizeof(ABDataType) <= TwoGB &&
+             b_grid_desc_n_k.GetElementSpaceSize() * sizeof(ABDataType) <= TwoGB &&
+             e_grid_desc_m_n.GetElementSpaceSize() * sizeof(EDataType) <= TwoGB))
+        {
+            return false;
+        }
+
+        return true;
+    }
+
+    __host__ __device__ static constexpr bool CalculateHasMainKBlockLoop(index_t K)
+    {
+        const index_t num_loop = K / KPerBlock;
+
+        return GridwiseGemmPipe::CalculateHasMainLoop(num_loop);
+    }
+
+    using DefaultAGridDesc_AK0_M_AK1 =
+        remove_cvref_t<decltype(MakeDefaultAGridDescriptor_AK0_M_AK1(AGridDesc_M_K{}))>;
+    using DefaultBGridDesc_BK0_N_BK1 =
+        remove_cvref_t<decltype(MakeDefaultBGridDescriptor_BK0_N_BK1(BGridDesc_N_K{}))>;
+    using EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock  = remove_cvref_t<decltype(
         MakeEGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(EGridDesc_M_N{}))>;
+    using DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock = remove_cvref_t<decltype(
+        MakeDsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock(DsGridDesc_M_N{}))>;
 
     using DefaultBlock2ETileMap =
         remove_cvref_t<decltype(MakeDefaultBlock2ETileMap(EGridDesc_M_N{}))>;
 
     using DsGridPointer = decltype(MakeDsGridPointer());
 
-    template <bool HasMainKBlockLoop, typename Block2ETileMap>
-    __device__ static void
-    Run(const FloatAB* __restrict__ p_a_grid,
-        const FloatAB* __restrict__ p_b_grid,
-        DsGridPointer p_ds_grid,
-        FloatE* __restrict__ p_e_grid,
-        void* __restrict__ p_shared,
-        const AElementwiseOperation& a_element_op,
-        const BElementwiseOperation& b_element_op,
-        const CDEElementwiseOperation& cde_element_op,
-        const AGridDesc_AK0_M_AK1& a_grid_desc_ak0_m_ak1,
-        const BGridDesc_BK0_N_BK1& b_grid_desc_bk0_n_bk1,
-        const StaticallyIndexedArray<EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock,
-                                     NumDTensor>&
-            ds_grid_desc_mblock_mperblock_nblock_nperblock, // FIXME: Ds desc may be of different
-                                                            // type from E
-        const EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock&
-            e_grid_desc_mblock_mperblock_nblock_nperblock,
-        const Block2ETileMap& block_2_etile_map)
+    template <bool HasMainKBlockLoop,
+              typename AGridDesc_AK0_M_AK1,
+              typename BGridDesc_BK0_N_BK1,
+              typename Block2ETileMap>
+    __device__ static void Run(const ABDataType* __restrict__ p_a_grid,
+                               const ABDataType* __restrict__ p_b_grid,
+                               DsGridPointer p_ds_grid,
+                               EDataType* __restrict__ p_e_grid,
+                               void* __restrict__ p_shared,
+                               const AElementwiseOperation& a_element_op,
+                               const BElementwiseOperation& b_element_op,
+                               const CDEElementwiseOperation& cde_element_op,
+                               const AGridDesc_AK0_M_AK1& a_grid_desc_ak0_m_ak1,
+                               const BGridDesc_BK0_N_BK1& b_grid_desc_bk0_n_bk1,
+                               const DsGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock&
+                                   ds_grid_desc_mblock_mperblock_nblock_nperblock,
+                               const EGridDescriptor_MBlock_MPerBlock_NBlock_NPerBlock&
+                                   e_grid_desc_mblock_mperblock_nblock_nperblock,
+                               const Block2ETileMap& block_2_etile_map)
     {
         const auto a_grid_buf = make_dynamic_buffer<AddressSpaceEnum::Global>(
             p_a_grid, a_grid_desc_ak0_m_ak1.GetElementSpaceSize());
@@ -316,11 +399,11 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
                                                 AElementwiseOperation,
                                                 ck::tensor_operation::element_wise::PassThrough,
                                                 InMemoryDataOperationEnum::Set,
-                                                Sequence<AK0, MPerBlock, AK1>,
+                                                Sequence<AK0PerBlock, MPerBlock, AK1>,
                                                 ABlockTransferThreadClusterLengths_AK0_M_AK1,
                                                 ABlockTransferThreadClusterArrangeOrder,
-                                                FloatAB,
-                                                FloatAB,
+                                                ABDataType,
+                                                ABDataType,
                                                 decltype(a_grid_desc_ak0_m_ak1),
                                                 decltype(a_block_desc_ak0_m_ak1),
                                                 ABlockTransferSrcAccessOrder,
@@ -347,11 +430,11 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
                                                 BElementwiseOperation,
                                                 ck::tensor_operation::element_wise::PassThrough,
                                                 InMemoryDataOperationEnum::Set,
-                                                Sequence<BK0, NPerBlock, BK1>,
+                                                Sequence<BK0PerBlock, NPerBlock, BK1>,
                                                 BBlockTransferThreadClusterLengths_BK0_N_BK1,
                                                 BBlockTransferThreadClusterArrangeOrder,
-                                                FloatAB,
-                                                FloatAB,
+                                                ABDataType,
+                                                ABDataType,
                                                 decltype(b_grid_desc_bk0_n_bk1),
                                                 decltype(b_block_desc_bk0_n_bk1),
                                                 BBlockTransferSrcAccessOrder,
@@ -379,13 +462,14 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
         //     c_mtx[MPerBlock, NPerBlock] is distributed among threads, and saved in
         //       register
         // sanity check
-        constexpr index_t KPack = math::max(
-            math::lcm(AK1, BK1), MfmaSelector<FloatAB, MPerXdl, NPerXdl>::selected_mfma.k_per_blk);
+        constexpr index_t KPack =
+            math::max(math::lcm(AK1, BK1),
+                      MfmaSelector<ABDataType, MPerXdl, NPerXdl>::selected_mfma.k_per_blk);
 
         auto blockwise_gemm = BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_Selector<
             BlockSize,
-            FloatAB,
-            FloatGemmAcc,
+            ABDataType,
+            AccDataType,
             decltype(a_block_desc_ak0_m_ak1),
             decltype(b_block_desc_bk0_n_bk1),
             MPerXdl,
@@ -402,10 +486,10 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
             a_block_desc_ak0_m_ak1.GetElementSpaceSize(), max_lds_align);
 
         auto a_block_buf = make_dynamic_buffer<AddressSpaceEnum::Lds>(
-            static_cast<FloatAB*>(p_shared), a_block_desc_ak0_m_ak1.GetElementSpaceSize());
+            static_cast<ABDataType*>(p_shared), a_block_desc_ak0_m_ak1.GetElementSpaceSize());
 
         auto b_block_buf = make_dynamic_buffer<AddressSpaceEnum::Lds>(
-            static_cast<FloatAB*>(p_shared) + a_block_space_size_aligned,
+            static_cast<ABDataType*>(p_shared) + a_block_space_size_aligned,
             b_block_desc_bk0_n_bk1.GetElementSpaceSize());
 
         constexpr auto a_block_slice_copy_step = make_multi_index(KPerBlock / AK1, 0, 0);
@@ -466,7 +550,7 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
                 GetCShuffleBlockDescriptor_MBlock_MPerBlock_NBlock_NPerBlock();
 
             auto c_shuffle_block_buf = make_dynamic_buffer<AddressSpaceEnum::Lds>(
-                static_cast<FloatCShuffle*>(p_shared),
+                static_cast<CShuffleDataType*>(p_shared),
                 c_shuffle_block_desc_mblock_mperblock_nblock_nperblock.GetElementSpaceSize());
 
             constexpr auto c_block_desc_m0_n0_m1_n1_m2_m3_m4_n2 = transform_tensor_descriptor(
@@ -518,8 +602,8 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
 
             // shuffle: threadwise copy C from VGPR to LDS
             auto c_thread_copy_vgpr_to_lds =
-                ThreadwiseTensorSliceTransfer_v1r3<FloatGemmAcc,
-                                                   FloatCShuffle,
+                ThreadwiseTensorSliceTransfer_v1r3<AccDataType,
+                                                   CShuffleDataType,
                                                    decltype(c_thread_desc_m0_n0_m1_n1_m2_m3_m4_n2),
                                                    decltype(c_block_desc_m0_n0_m1_n1_m2_m3_m4_n2),
                                                    ck::tensor_operation::element_wise::PassThrough,
@@ -576,8 +660,8 @@ struct GridwiseGemmMultipleD_k0mk1_k0nk1_mn_xdl_cshuffle
             // blockwise copy C/D/E between LDS and global
             auto cde_block_copy_lds_and_global = ThreadGroupTensorSliceTransfer_v7<
                 ThisThreadBlock,
-                decltype(container_concat(make_tuple(FloatCShuffle{}), DsDataType{})),
-                Tuple<FloatE>,
+                decltype(container_concat(make_tuple(CShuffleDataType{}), DsDataType{})),
+                Tuple<EDataType>,
                 decltype(c_ds_desc_refs),
                 decltype(tie(e_grid_desc_mblock_mperblock_nblock_nperblock)),
                 CDEElementwiseOperation,
diff --git a/include/ck/utility/tuple.hpp b/include/ck/utility/tuple.hpp
index 07bf721d54..d8664be550 100644
--- a/include/ck/utility/tuple.hpp
+++ b/include/ck/utility/tuple.hpp
@@ -21,6 +21,8 @@ struct TupleElementKey
 template <typename Key, typename Data>
 struct TupleElementKeyData
 {
+    using DataType = Data;
+
 #if 0 // workaround compiler complaint about implicitly-deleted default constructor
     __host__ __device__ constexpr TupleElementKeyData() = default;
 #else
@@ -34,29 +36,40 @@ struct TupleElementKeyData
     {
     }
 
-    Data mData;
+    DataType mData;
 };
 
+// for read access of tuple element
 template <typename Key, typename Data>
 __host__ __device__ constexpr const Data&
-get_tuple_element_data(const TupleElementKeyData<Key, Data>& x)
+get_tuple_element_data_reference(const TupleElementKeyData<Key, Data>& x)
 {
     return static_cast<const Data&>(x.mData);
 }
 
+// for write access of tuple element
 template <typename Key, typename Data>
-__host__ __device__ constexpr Data& get_tuple_element_data(TupleElementKeyData<Key, Data>& x)
+__host__ __device__ constexpr Data&
+get_tuple_element_data_reference(TupleElementKeyData<Key, Data>& x)
 {
     return x.mData;
 }
 
 // TODO: not sure the use of reference is correct
 template <typename Key, typename Data>
-__host__ __device__ constexpr Data&& get_tuple_element_data(TupleElementKeyData<Key, Data>&& x)
+__host__ __device__ constexpr Data&&
+get_tuple_element_data_reference(TupleElementKeyData<Key, Data>&& x)
 {
     return static_cast<Data&&>(x.mData);
 }
 
+// for infering type of tuple element
+template <typename Key, typename Data>
+__host__ __device__ constexpr Data get_tuple_element_data(const TupleElementKeyData<Key, Data>& x)
+{
+    return std::forward(x.mData);
+}
+
 template <typename Indices, typename... Xs>
 struct TupleImpl;
 
@@ -87,13 +100,13 @@ struct TupleImpl<Sequence<Is...>, Xs...> : TupleElementKeyData<TupleElementKey<I
     template <index_t I>
     __host__ __device__ constexpr const auto& GetElementDataByKey(TupleElementKey<I>) const
     {
-        return get_tuple_element_data<TupleElementKey<I>>(*this);
+        return get_tuple_element_data_reference<TupleElementKey<I>>(*this);
     }
 
     template <index_t I>
     __host__ __device__ constexpr auto& GetElementDataByKey(TupleElementKey<I>)
     {
-        return get_tuple_element_data<TupleElementKey<I>>(*this);
+        return get_tuple_element_data_reference<TupleElementKey<I>>(*this);
     }
 };
 
@@ -185,7 +198,8 @@ struct Tuple<>
 template <index_t I, typename TTuple>
 struct tuple_element
 {
-    using type = decltype(TTuple{}.At(Number<I>{}));
+    // type should keep the cv/ref qualifier of original tuple element
+    using type = decltype(detail::get_tuple_element_data<detail::TupleElementKey<I>>(TTuple{}));
 };
 
 template <index_t I, typename TTuple>
diff --git a/library/CMakeLists.txt b/library/CMakeLists.txt
index a92fae9e26..90873fdd14 100644
--- a/library/CMakeLists.txt
+++ b/library/CMakeLists.txt
@@ -1,3 +1,2 @@
 add_subdirectory(src/tensor_operation_instance/gpu)
-add_subdirectory(src/host_tensor)
 add_subdirectory(src/utility)
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp
index 06e74a9e9a..97ce3dcacd 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp
@@ -7,7 +7,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_cgemm.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_cgemm.hpp
index cde0725789..ce0e337498 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_cgemm.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_cgemm.hpp
@@ -7,7 +7,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp
index 1239ca163a..225f7b7e36 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp
@@ -8,22 +8,24 @@
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
 namespace host {
 
-// out[N, K, Ho, Wo] = in[N, C, Hi, Wi] * wei[K, C, Y, X]
-template <typename InDataType,
+// input descriptor in [G, N, C, Do, Ho, Wo] order
+// weight descriptor in [G, K, C, Z, Y, X] order
+// output descriptor in [G, N, K, Di, Hi, Wi] order
+// phyiscal layout is irrelavent
+template <ck::index_t NDimSpatial,
+          typename InDataType,
           typename WeiDataType,
           typename OutDataType,
-          typename AccDataType,
           typename InElementwiseOperation,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation,
-          ck::index_t NumDimSpatial                                                    = 2,
-          typename ck::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
+          typename std::enable_if<NDimSpatial >= 1 && NDimSpatial <= 3, bool>::type = false>
 struct ReferenceConvBwdData : public device::BaseOperator
 {
     // Argument
@@ -73,36 +75,45 @@ struct ReferenceConvBwdData : public device::BaseOperator
 
         float Run(const Argument& arg)
         {
-            if constexpr(NumDimSpatial == 1)
+            if(!(arg.input_.GetNumOfDimension() == NDimSpatial + 3 &&
+                 arg.weight_.GetNumOfDimension() == NDimSpatial + 3 &&
+                 arg.output_.GetNumOfDimension() == NDimSpatial + 3))
             {
-                auto f_ncw = [&](auto n, auto c, auto wi) {
-                    std::size_t K  = arg.weight_.mDesc.GetLengths()[0];
-                    std::size_t X  = arg.weight_.mDesc.GetLengths()[2];
-                    std::size_t Wo = arg.output_.mDesc.GetLengths()[2];
+                throw std::runtime_error("wrong! inconsistent dimension");
+            }
 
-                    AccDataType v_acc = 0;
+            if constexpr(NDimSpatial == 1)
+            {
+                auto f_ncw = [&](auto g, auto n, auto c, auto wi) {
+                    std::size_t K  = arg.weight_.GetLengths()[1];
+                    std::size_t X  = arg.weight_.GetLengths()[3];
+                    std::size_t Wo = arg.output_.GetLengths()[3];
+
+                    float v_acc = 0;
 
                     for(std::size_t x = 0; x < X; ++x)
                     {
-                        auto w_tmp = ck::type_convert<ck::long_index_t>(wi) +
-                                     ck::type_convert<ck::long_index_t>(arg.in_left_pads_[0]) -
-                                     ck::type_convert<ck::long_index_t>(x * arg.conv_dilations_[0]);
+                        auto w_tmp = static_cast<ck::long_index_t>(wi) +
+                                     static_cast<ck::long_index_t>(arg.in_left_pads_[0]) -
+                                     static_cast<ck::long_index_t>(x * arg.conv_dilations_[0]);
+
                         if(w_tmp % arg.conv_strides_[0] == 0)
                         {
-                            auto wo = ck::type_convert<ck::long_index_t>(w_tmp) /
-                                      ck::type_convert<ck::long_index_t>(arg.conv_strides_[0]);
+                            auto wo = static_cast<ck::long_index_t>(w_tmp) /
+                                      static_cast<ck::long_index_t>(arg.conv_strides_[0]);
+
                             if(wo >= 0 && ck::type_convert<std::size_t>(wo) < Wo)
                             {
                                 for(std::size_t k = 0; k < K; ++k)
                                 {
-                                    AccDataType v_out = 0;
-                                    AccDataType v_wei = 0;
+                                    float v_out = 0;
+                                    float v_wei = 0;
 
                                     arg.out_element_op_(
-                                        v_out,
-                                        ck::type_convert<AccDataType>(arg.output_(n, k, wo)));
+                                        v_out, ck::type_convert<float>(arg.output_(g, n, k, wo)));
+
                                     arg.wei_element_op_(
-                                        v_wei, ck::type_convert<AccDataType>(arg.weight_(k, c, x)));
+                                        v_wei, ck::type_convert<float>(arg.weight_(g, k, c, x)));
 
                                     v_acc += v_out * v_wei;
                                 }
@@ -110,66 +121,72 @@ struct ReferenceConvBwdData : public device::BaseOperator
                         }
                     }
 
-                    arg.in_element_op_(v_acc, v_acc);
-                    arg.input_(n, c, wi) = ck::type_convert<InDataType>(v_acc);
+                    float v_in;
+
+                    arg.in_element_op_(v_in, v_acc);
+
+                    arg.input_(g, n, c, wi) = ck::type_convert<InDataType>(v_acc);
                 };
 
                 make_ParallelTensorFunctor(f_ncw,
-                                           arg.input_.mDesc.GetLengths()[0],
-                                           arg.input_.mDesc.GetLengths()[1],
-                                           arg.input_.mDesc.GetLengths()[2])(
+                                           arg.input_.GetLengths()[0],
+                                           arg.input_.GetLengths()[1],
+                                           arg.input_.GetLengths()[2],
+                                           arg.input_.GetLengths()[3])(
                     std::thread::hardware_concurrency());
 
                 return 0;
             }
-            else if constexpr(NumDimSpatial == 2)
+            else if constexpr(NDimSpatial == 2)
             {
-                auto f_nchw = [&](auto n, auto c, auto hi, auto wi) {
-                    std::size_t K = arg.weight_.mDesc.GetLengths()[0];
-                    std::size_t Y = arg.weight_.mDesc.GetLengths()[2];
-                    std::size_t X = arg.weight_.mDesc.GetLengths()[3];
+                auto f_nchw = [&](auto g, auto n, auto c, auto hi, auto wi) {
+                    std::size_t K = arg.weight_.GetLengths()[1];
+                    std::size_t Y = arg.weight_.GetLengths()[3];
+                    std::size_t X = arg.weight_.GetLengths()[4];
 
-                    std::size_t Ho = arg.output_.mDesc.GetLengths()[2];
-                    std::size_t Wo = arg.output_.mDesc.GetLengths()[3];
+                    std::size_t Ho = arg.output_.GetLengths()[3];
+                    std::size_t Wo = arg.output_.GetLengths()[4];
 
-                    AccDataType v_acc = 0;
+                    float v_acc = 0;
 
                     for(std::size_t y = 0; y < Y; ++y)
                     {
-                        auto h_tmp = ck::type_convert<ck::long_index_t>(hi) +
-                                     ck::type_convert<ck::long_index_t>(arg.in_left_pads_[0]) -
-                                     ck::type_convert<ck::long_index_t>(y * arg.conv_dilations_[0]);
+                        auto h_tmp = static_cast<ck::long_index_t>(hi) +
+                                     static_cast<ck::long_index_t>(arg.in_left_pads_[0]) -
+                                     static_cast<ck::long_index_t>(y * arg.conv_dilations_[0]);
                         if(h_tmp % arg.conv_strides_[0] == 0)
                         {
-                            auto ho = ck::type_convert<ck::long_index_t>(h_tmp) /
-                                      ck::type_convert<ck::long_index_t>(arg.conv_strides_[0]);
+                            auto ho = static_cast<ck::long_index_t>(h_tmp) /
+                                      static_cast<ck::long_index_t>(arg.conv_strides_[0]);
                             if(ho >= 0 && ck::type_convert<std::size_t>(ho) < Ho)
                             {
                                 for(std::size_t x = 0; x < X; ++x)
                                 {
                                     auto w_tmp =
-                                        ck::type_convert<ck::long_index_t>(wi) +
-                                        ck::type_convert<ck::long_index_t>(arg.in_left_pads_[1]) -
-                                        ck::type_convert<ck::long_index_t>(x *
-                                                                           arg.conv_dilations_[1]);
+                                        static_cast<ck::long_index_t>(wi) +
+                                        static_cast<ck::long_index_t>(arg.in_left_pads_[1]) -
+                                        static_cast<ck::long_index_t>(x * arg.conv_dilations_[1]);
                                     if(w_tmp % arg.conv_strides_[1] == 0)
                                     {
-                                        auto wo = ck::type_convert<ck::long_index_t>(w_tmp) /
-                                                  ck::type_convert<ck::long_index_t>(
-                                                      arg.conv_strides_[1]);
+                                        auto wo =
+                                            static_cast<ck::long_index_t>(w_tmp) /
+                                            static_cast<ck::long_index_t>(arg.conv_strides_[1]);
                                         if(wo >= 0 && ck::type_convert<std::size_t>(wo) < Wo)
                                         {
                                             for(std::size_t k = 0; k < K; ++k)
                                             {
-                                                AccDataType v_out = 0;
-                                                AccDataType v_wei = 0;
+                                                float v_out = 0;
+                                                float v_wei = 0;
 
-                                                arg.out_element_op_(v_out,
-                                                                    ck::type_convert<AccDataType>(
-                                                                        arg.output_(n, k, ho, wo)));
-                                                arg.wei_element_op_(v_wei,
-                                                                    ck::type_convert<AccDataType>(
-                                                                        arg.weight_(k, c, y, x)));
+                                                arg.out_element_op_(
+                                                    v_out,
+                                                    ck::type_convert<float>(
+                                                        arg.output_(g, n, k, ho, wo)));
+
+                                                arg.wei_element_op_(
+                                                    v_wei,
+                                                    ck::type_convert<float>(
+                                                        arg.weight_(g, k, c, y, x)));
 
                                                 v_acc += v_out * v_wei;
                                             }
@@ -180,90 +197,91 @@ struct ReferenceConvBwdData : public device::BaseOperator
                         }
                     }
 
-                    AccDataType v_in;
+                    float v_in;
+
                     arg.in_element_op_(v_in, v_acc);
-                    arg.input_(n, c, hi, wi) = ck::type_convert<InDataType>(v_in);
+
+                    arg.input_(g, n, c, hi, wi) = ck::type_convert<InDataType>(v_acc);
                 };
 
                 make_ParallelTensorFunctor(f_nchw,
-                                           arg.input_.mDesc.GetLengths()[0],
-                                           arg.input_.mDesc.GetLengths()[1],
-                                           arg.input_.mDesc.GetLengths()[2],
-                                           arg.input_.mDesc.GetLengths()[3])(
+                                           arg.input_.GetLengths()[0],
+                                           arg.input_.GetLengths()[1],
+                                           arg.input_.GetLengths()[2],
+                                           arg.input_.GetLengths()[3],
+                                           arg.input_.GetLengths()[4])(
                     std::thread::hardware_concurrency());
 
                 return 0;
             }
-            else if constexpr(NumDimSpatial == 3)
+            else if constexpr(NDimSpatial == 3)
             {
-                auto f_ncdhw = [&](auto n, auto c, auto di, auto hi, auto wi) {
-                    std::size_t K = arg.weight_.mDesc.GetLengths()[0];
-                    std::size_t Z = arg.weight_.mDesc.GetLengths()[2];
-                    std::size_t Y = arg.weight_.mDesc.GetLengths()[3];
-                    std::size_t X = arg.weight_.mDesc.GetLengths()[4];
+                auto f_ncdhw = [&](auto g, auto n, auto c, auto di, auto hi, auto wi) {
+                    std::size_t K = arg.weight_.GetLengths()[1];
+                    std::size_t Z = arg.weight_.GetLengths()[3];
+                    std::size_t Y = arg.weight_.GetLengths()[4];
+                    std::size_t X = arg.weight_.GetLengths()[5];
 
-                    std::size_t Do = arg.output_.mDesc.GetLengths()[2];
-                    std::size_t Ho = arg.output_.mDesc.GetLengths()[3];
-                    std::size_t Wo = arg.output_.mDesc.GetLengths()[4];
+                    std::size_t Do = arg.output_.GetLengths()[3];
+                    std::size_t Ho = arg.output_.GetLengths()[4];
+                    std::size_t Wo = arg.output_.GetLengths()[5];
 
-                    AccDataType v_acc = 0;
+                    float v_acc = 0;
 
                     for(std::size_t z = 0; z < Z; ++z)
                     {
-                        auto d_tmp = ck::type_convert<ck::long_index_t>(di) +
-                                     ck::type_convert<ck::long_index_t>(arg.in_left_pads_[0]) -
-                                     ck::type_convert<ck::long_index_t>(z * arg.conv_dilations_[0]);
+                        auto d_tmp = static_cast<ck::long_index_t>(di) +
+                                     static_cast<ck::long_index_t>(arg.in_left_pads_[0]) -
+                                     static_cast<ck::long_index_t>(z * arg.conv_dilations_[0]);
                         if(d_tmp % arg.conv_strides_[0] == 0)
                         {
-                            auto do_ = ck::type_convert<ck::long_index_t>(d_tmp) /
-                                       ck::type_convert<ck::long_index_t>(arg.conv_strides_[0]);
+                            auto do_ = static_cast<ck::long_index_t>(d_tmp) /
+                                       static_cast<ck::long_index_t>(arg.conv_strides_[0]);
                             if(do_ >= 0 && ck::type_convert<std::size_t>(do_) < Do)
                             {
                                 for(std::size_t y = 0; y < Y; ++y)
                                 {
                                     auto h_tmp =
-                                        ck::type_convert<ck::long_index_t>(hi) +
-                                        ck::type_convert<ck::long_index_t>(arg.in_left_pads_[1]) -
-                                        ck::type_convert<ck::long_index_t>(y *
-                                                                           arg.conv_dilations_[1]);
+                                        static_cast<ck::long_index_t>(hi) +
+                                        static_cast<ck::long_index_t>(arg.in_left_pads_[1]) -
+                                        static_cast<ck::long_index_t>(y * arg.conv_dilations_[1]);
                                     if(h_tmp % arg.conv_strides_[1] == 0)
                                     {
-                                        auto ho = ck::type_convert<ck::long_index_t>(h_tmp) /
-                                                  ck::type_convert<ck::long_index_t>(
-                                                      arg.conv_strides_[1]);
+                                        auto ho =
+                                            static_cast<ck::long_index_t>(h_tmp) /
+                                            static_cast<ck::long_index_t>(arg.conv_strides_[1]);
                                         if(ho >= 0 && ck::type_convert<std::size_t>(ho) < Ho)
                                         {
                                             for(std::size_t x = 0; x < X; ++x)
                                             {
-                                                auto w_tmp =
-                                                    ck::type_convert<ck::long_index_t>(wi) +
-                                                    ck::type_convert<ck::long_index_t>(
-                                                        arg.in_left_pads_[2]) -
-                                                    ck::type_convert<ck::long_index_t>(
-                                                        x * arg.conv_dilations_[2]);
+                                                auto w_tmp = static_cast<ck::long_index_t>(wi) +
+                                                             static_cast<ck::long_index_t>(
+                                                                 arg.in_left_pads_[2]) -
+                                                             static_cast<ck::long_index_t>(
+                                                                 x * arg.conv_dilations_[2]);
+
                                                 if(w_tmp % arg.conv_strides_[2] == 0)
                                                 {
-                                                    auto wo =
-                                                        ck::type_convert<ck::long_index_t>(w_tmp) /
-                                                        ck::type_convert<ck::long_index_t>(
-                                                            arg.conv_strides_[2]);
+                                                    auto wo = static_cast<ck::long_index_t>(w_tmp) /
+                                                              static_cast<ck::long_index_t>(
+                                                                  arg.conv_strides_[2]);
                                                     if(wo >= 0 &&
                                                        ck::type_convert<std::size_t>(wo) < Wo)
                                                     {
                                                         for(std::size_t k = 0; k < K; ++k)
                                                         {
-                                                            AccDataType v_out = 0;
-                                                            AccDataType v_wei = 0;
+                                                            float v_out = 0;
+                                                            float v_wei = 0;
 
                                                             arg.out_element_op_(
                                                                 v_out,
-                                                                ck::type_convert<AccDataType>(
-                                                                    arg.output_(
-                                                                        n, k, do_, ho, wo)));
+                                                                ck::type_convert<float>(arg.output_(
+                                                                    g, n, k, do_, ho, wo)));
+
                                                             arg.wei_element_op_(
                                                                 v_wei,
-                                                                ck::type_convert<AccDataType>(
-                                                                    arg.weight_(k, c, z, y, x)));
+                                                                ck::type_convert<float>(
+                                                                    arg.weight_(g, k, c, z, y, x)));
 
                                                             v_acc += v_out * v_wei;
                                                         }
@@ -277,17 +295,20 @@ struct ReferenceConvBwdData : public device::BaseOperator
                         }
                     }
 
-                    AccDataType v_in;
+                    float v_in;
+
                     arg.in_element_op_(v_in, v_acc);
-                    arg.input_(n, c, di, hi, wi) = ck::type_convert<InDataType>(v_in);
+
+                    arg.input_(g, n, c, di, hi, wi) = ck::type_convert<InDataType>(v_acc);
                 };
 
                 make_ParallelTensorFunctor(f_ncdhw,
-                                           arg.input_.mDesc.GetLengths()[0],
-                                           arg.input_.mDesc.GetLengths()[1],
-                                           arg.input_.mDesc.GetLengths()[2],
-                                           arg.input_.mDesc.GetLengths()[3],
-                                           arg.input_.mDesc.GetLengths()[4])(
+                                           arg.input_.GetLengths()[0],
+                                           arg.input_.GetLengths()[1],
+                                           arg.input_.GetLengths()[2],
+                                           arg.input_.GetLengths()[3],
+                                           arg.input_.GetLengths()[4],
+                                           arg.input_.GetLengths()[5])(
                     std::thread::hardware_concurrency());
 
                 return 0;
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_weight.hpp
similarity index 57%
rename from library/include/ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp
rename to library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_weight.hpp
index 6cab5f28f4..2911d5040d 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_weight.hpp
@@ -7,21 +7,25 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
 namespace host {
 
-// out[N, K, Ho, Wo] = in[N, C, Hi, Wi] * wei[K, C, Y, X]
-template <typename InDataType,
+// input descriptor in [G, N, C, Do, Ho, Wo] order
+// weight descriptor in [G, K, C, Z, Y, X] order
+// output descriptor in [G, N, K, Di, Hi, Wi] order
+// phyiscal layout is irrelavent
+template <ck::index_t NDimSpatial,
+          typename InDataType,
           typename WeiDataType,
           typename OutDataType,
           typename InElementwiseOperation,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation,
-          ck::index_t NumDimSpatial                                                    = 2,
-          typename ck::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
+          typename std::enable_if<NDimSpatial >= 1 && NDimSpatial <= 3, bool>::type = false>
 struct ReferenceConvBwdWeight : public device::BaseOperator
 {
     // Argument
@@ -71,156 +75,162 @@ struct ReferenceConvBwdWeight : public device::BaseOperator
 
         float Run(const Argument& arg)
         {
-            if constexpr(NumDimSpatial == 1)
+            if(!(arg.input_.GetNumOfDimension() == NDimSpatial + 3 &&
+                 arg.weight_.GetNumOfDimension() == NDimSpatial + 3 &&
+                 arg.output_.GetNumOfDimension() == NDimSpatial + 3))
             {
-                constexpr auto I0 = Number<0>{};
-                auto f_kcx        = [&](auto k, auto c, auto x) {
+                throw std::runtime_error("wrong! inconsistent dimension");
+            }
+
+            if constexpr(NDimSpatial == 1)
+            {
+                auto f_kcx = [&](auto g, auto k, auto c, auto x) {
                     float v_acc = 0;
-                    for(std::size_t n = 0; n < arg.output_.mDesc.GetLengths()[0]; ++n)
+
+                    for(std::size_t n = 0; n < arg.output_.GetLengths()[1]; ++n)
                     {
-                        for(std::size_t wo = 0; wo < arg.output_.mDesc.GetLengths()[2]; ++wo)
+                        for(std::size_t wo = 0; wo < arg.output_.GetLengths()[3]; ++wo)
                         {
-                            auto wi =
-                                ck::type_convert<ck::long_index_t>(wo * arg.conv_strides_[I0]) +
-                                ck::type_convert<ck::long_index_t>(x * arg.conv_dilations_[I0]) -
-                                ck::type_convert<ck::long_index_t>(arg.in_left_pads_[I0]);
+                            auto wi = static_cast<ck::long_index_t>(wo * arg.conv_strides_[0]) +
+                                      static_cast<ck::long_index_t>(x * arg.conv_dilations_[0]) -
+                                      static_cast<ck::long_index_t>(arg.in_left_pads_[0]);
+
                             if(wi >= 0 &&
-                               ck::type_convert<std::size_t>(wi) < arg.input_.mDesc.GetLengths()[2])
+                               ck::type_convert<std::size_t>(wi) < arg.input_.GetLengths()[3])
                             {
                                 float v_out;
                                 float v_in;
 
-                                arg.out_element_op_(v_out,
-                                                    ck::type_convert<float>(arg.output_(n, k, wo)));
-                                arg.in_element_op_(v_in,
-                                                   ck::type_convert<float>(arg.input_(n, c, wi)));
+                                arg.out_element_op_(
+                                    v_out, ck::type_convert<float>(arg.output_(g, n, k, wo)));
+
+                                arg.in_element_op_(
+                                    v_in, ck::type_convert<float>(arg.input_(g, n, c, wi)));
 
                                 v_acc += v_out * v_in;
                             }
                         }
                     }
+
                     float v_wei;
 
                     arg.wei_element_op_(v_wei, v_acc);
 
-                    arg.weight_(k, c, x) = ck::type_convert<WeiDataType>(v_wei);
+                    arg.weight_(g, k, c, x) = ck::type_convert<WeiDataType>(v_wei);
                 };
 
                 make_ParallelTensorFunctor(f_kcx,
-                                           arg.weight_.mDesc.GetLengths()[0],
-                                           arg.weight_.mDesc.GetLengths()[1],
-                                           arg.weight_.mDesc.GetLengths()[2])(
+                                           arg.weight_.GetLengths()[0],
+                                           arg.weight_.GetLengths()[1],
+                                           arg.weight_.GetLengths()[2],
+                                           arg.weight_.GetLengths()[3])(
                     std::thread::hardware_concurrency());
 
                 return 0;
             }
-            else if constexpr(NumDimSpatial == 2)
+            else if constexpr(NDimSpatial == 2)
             {
-                constexpr auto I0 = Number<0>{};
-                constexpr auto I1 = Number<1>{};
-                auto f_kcyx       = [&](auto k, auto c, auto y, auto x) {
+                auto f_kcyx = [&](auto g, auto k, auto c, auto y, auto x) {
                     float v_acc = 0;
-                    for(std::size_t n = 0; n < arg.output_.mDesc.GetLengths()[0]; ++n)
+
+                    for(std::size_t n = 0; n < arg.output_.GetLengths()[1]; ++n)
                     {
-                        for(std::size_t ho = 0; ho < arg.output_.mDesc.GetLengths()[2]; ++ho)
+                        for(std::size_t ho = 0; ho < arg.output_.GetLengths()[3]; ++ho)
                         {
-                            auto hi =
-                                ck::type_convert<ck::long_index_t>(ho * arg.conv_strides_[I0]) +
-                                ck::type_convert<ck::long_index_t>(y * arg.conv_dilations_[I0]) -
-                                ck::type_convert<ck::long_index_t>(arg.in_left_pads_[I0]);
-                            for(std::size_t wo = 0; wo < arg.output_.mDesc.GetLengths()[3]; ++wo)
+                            auto hi = static_cast<ck::long_index_t>(ho * arg.conv_strides_[0]) +
+                                      static_cast<ck::long_index_t>(y * arg.conv_dilations_[0]) -
+                                      static_cast<ck::long_index_t>(arg.in_left_pads_[0]);
+
+                            for(std::size_t wo = 0; wo < arg.output_.GetLengths()[4]; ++wo)
                             {
                                 auto wi =
-                                    ck::type_convert<ck::long_index_t>(wo * arg.conv_strides_[I1]) +
-                                    ck::type_convert<ck::long_index_t>(x *
-                                                                       arg.conv_dilations_[I1]) -
-                                    ck::type_convert<ck::long_index_t>(arg.in_left_pads_[I1]);
+                                    static_cast<ck::long_index_t>(wo * arg.conv_strides_[1]) +
+                                    static_cast<ck::long_index_t>(x * arg.conv_dilations_[1]) -
+                                    static_cast<ck::long_index_t>(arg.in_left_pads_[1]);
+
                                 if(hi >= 0 &&
-                                   ck::type_convert<std::size_t>(hi) <
-                                       arg.input_.mDesc.GetLengths()[2] &&
+                                   ck::type_convert<std::size_t>(hi) < arg.input_.GetLengths()[3] &&
                                    wi >= 0 &&
-                                   ck::type_convert<std::size_t>(wi) <
-                                       arg.input_.mDesc.GetLengths()[3])
+                                   ck::type_convert<std::size_t>(wi) < arg.input_.GetLengths()[4])
                                 {
                                     float v_out;
                                     float v_in;
 
                                     arg.out_element_op_(
-                                        v_out, ck::type_convert<float>(arg.output_(n, k, ho, wo)));
+                                        v_out,
+                                        ck::type_convert<float>(arg.output_(g, n, k, ho, wo)));
+
                                     arg.in_element_op_(
-                                        v_in, ck::type_convert<float>(arg.input_(n, c, hi, wi)));
+                                        v_in, ck::type_convert<float>(arg.input_(g, n, c, hi, wi)));
 
                                     v_acc += v_out * v_in;
                                 }
                             }
                         }
                     }
+
                     float v_wei;
 
                     arg.wei_element_op_(v_wei, v_acc);
 
-                    arg.weight_(k, c, y, x) = ck::type_convert<WeiDataType>(v_wei);
+                    arg.weight_(g, k, c, y, x) = ck::type_convert<WeiDataType>(v_wei);
                 };
 
                 make_ParallelTensorFunctor(f_kcyx,
-                                           arg.weight_.mDesc.GetLengths()[0],
-                                           arg.weight_.mDesc.GetLengths()[1],
-                                           arg.weight_.mDesc.GetLengths()[2],
-                                           arg.weight_.mDesc.GetLengths()[3])(
+                                           arg.weight_.GetLengths()[0],
+                                           arg.weight_.GetLengths()[1],
+                                           arg.weight_.GetLengths()[2],
+                                           arg.weight_.GetLengths()[3],
+                                           arg.weight_.GetLengths()[4])(
                     std::thread::hardware_concurrency());
 
                 return 0;
             }
-            else if constexpr(NumDimSpatial == 3)
+            else if constexpr(NDimSpatial == 3)
             {
-                constexpr auto I0 = Number<0>{};
-                constexpr auto I1 = Number<1>{};
-                constexpr auto I2 = Number<2>{};
-                auto f_kczyx      = [&](auto k, auto c, auto z, auto y, auto x) {
+                auto f_kczyx = [&](auto g, auto k, auto c, auto z, auto y, auto x) {
                     float v_acc = 0;
-                    for(std::size_t n = 0; n < arg.output_.mDesc.GetLengths()[0]; ++n)
+
+                    for(std::size_t n = 0; n < arg.output_.GetLengths()[1]; ++n)
                     {
-                        for(std::size_t do_ = 0; do_ < arg.output_.mDesc.GetLengths()[2]; ++do_)
+                        for(std::size_t do_ = 0; do_ < arg.output_.GetLengths()[3]; ++do_)
                         {
-                            auto di =
-                                ck::type_convert<ck::long_index_t>(do_ * arg.conv_strides_[I0]) +
-                                ck::type_convert<ck::long_index_t>(z * arg.conv_dilations_[I0]) -
-                                ck::type_convert<ck::long_index_t>(arg.in_left_pads_[I0]);
-                            for(std::size_t ho = 0; ho < arg.output_.mDesc.GetLengths()[3]; ++ho)
+                            auto di = static_cast<ck::long_index_t>(do_ * arg.conv_strides_[0]) +
+                                      static_cast<ck::long_index_t>(z * arg.conv_dilations_[0]) -
+                                      static_cast<ck::long_index_t>(arg.in_left_pads_[0]);
+                            for(std::size_t ho = 0; ho < arg.output_.GetLengths()[4]; ++ho)
                             {
                                 auto hi =
-                                    ck::type_convert<ck::long_index_t>(ho * arg.conv_strides_[I1]) +
-                                    ck::type_convert<ck::long_index_t>(y *
-                                                                       arg.conv_dilations_[I1]) -
-                                    ck::type_convert<ck::long_index_t>(arg.in_left_pads_[I1]);
-                                for(std::size_t wo = 0; wo < arg.output_.mDesc.GetLengths()[4];
-                                    ++wo)
+                                    static_cast<ck::long_index_t>(ho * arg.conv_strides_[1]) +
+                                    static_cast<ck::long_index_t>(y * arg.conv_dilations_[1]) -
+                                    static_cast<ck::long_index_t>(arg.in_left_pads_[1]);
+                                for(std::size_t wo = 0; wo < arg.output_.GetLengths()[5]; ++wo)
                                 {
                                     auto wi =
-                                        ck::type_convert<ck::long_index_t>(wo *
-                                                                           arg.conv_strides_[I2]) +
-                                        ck::type_convert<ck::long_index_t>(
-                                            x * arg.conv_dilations_[I2]) -
-                                        ck::type_convert<ck::long_index_t>(arg.in_left_pads_[I2]);
+                                        static_cast<ck::long_index_t>(wo * arg.conv_strides_[2]) +
+                                        static_cast<ck::long_index_t>(x * arg.conv_dilations_[2]) -
+                                        static_cast<ck::long_index_t>(arg.in_left_pads_[2]);
+
                                     if(di >= 0 &&
                                        ck::type_convert<std::size_t>(di) <
-                                           arg.input_.mDesc.GetLengths()[2] &&
+                                           arg.input_.GetLengths()[3] &&
                                        hi >= 0 &&
                                        ck::type_convert<std::size_t>(hi) <
-                                           arg.input_.mDesc.GetLengths()[3] &&
+                                           arg.input_.GetLengths()[4] &&
                                        wi >= 0 &&
                                        ck::type_convert<std::size_t>(wi) <
-                                           arg.input_.mDesc.GetLengths()[4])
+                                           arg.input_.GetLengths()[5])
                                     {
                                         float v_out;
                                         float v_in;
 
                                         arg.out_element_op_(v_out,
                                                             ck::type_convert<float>(
-                                                                arg.output_(n, k, do_, ho, wo)));
-                                        arg.in_element_op_(
-                                            v_in,
-                                            ck::type_convert<float>(arg.input_(n, c, di, hi, wi)));
+                                                                arg.output_(g, n, k, do_, ho, wo)));
+
+                                        arg.in_element_op_(v_in,
+                                                           ck::type_convert<float>(
+                                                               arg.input_(g, n, c, di, hi, wi)));
 
                                         v_acc += v_out * v_in;
                                     }
@@ -228,19 +238,21 @@ struct ReferenceConvBwdWeight : public device::BaseOperator
                             }
                         }
                     }
+
                     float v_wei;
 
                     arg.wei_element_op_(v_wei, v_acc);
 
-                    arg.weight_(k, c, z, y, x) = ck::type_convert<WeiDataType>(v_wei);
+                    arg.weight_(g, k, c, z, y, x) = ck::type_convert<WeiDataType>(v_wei);
                 };
 
                 make_ParallelTensorFunctor(f_kczyx,
-                                           arg.weight_.mDesc.GetLengths()[0],
-                                           arg.weight_.mDesc.GetLengths()[1],
-                                           arg.weight_.mDesc.GetLengths()[2],
-                                           arg.weight_.mDesc.GetLengths()[3],
-                                           arg.weight_.mDesc.GetLengths()[4])(
+                                           arg.weight_.GetLengths()[0],
+                                           arg.weight_.GetLengths()[1],
+                                           arg.weight_.GetLengths()[2],
+                                           arg.weight_.GetLengths()[3],
+                                           arg.weight_.GetLengths()[4],
+                                           arg.weight_.GetLengths()[5])(
                     std::thread::hardware_concurrency());
 
                 return 0;
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp
index fc333fbd6a..b8d47d218b 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp
@@ -8,7 +8,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -17,9 +17,10 @@ namespace host {
 //
 // @brief      Reference implementation for forward convolution.
 //
-// @paragraph  Supports both NCHW as well as NHWC formats (and their respective
-//             counterparts for weight and output) as long as tensor descriptor
-//             lengths is in NCHW.
+// @paragraph
+//             Tensor descriptor in GNCHW/GKCXY/GNKHW dimensional order
+//             Supports both GNCHW/NGCHW as well as GNHWC/NHWGC physical layout
+//             as long as dimensions in tensor descriptor is in GNCHW order
 //
 // @tparam     InDataType               Input tensor data type.
 // @tparam     WeiDataType              Weights tensor data type.
@@ -28,16 +29,20 @@ namespace host {
 //                                      operation.
 // @tparam     WeiElementwiseOperation  Functor for weights tensor elementwise
 //                                      operation.
-// @tparam     NumDimSpatial  Number of spatial dimensions.
+// @tparam     NDimSpatial  Number of spatial dimensions.
 //
-template <typename InDataType,
+// input descriptor in [G, N, C, Do, Ho, Wo] order
+// weight descriptor in [G, K, C, Z, Y, X] order
+// output descriptor in [G, N, K, Di, Hi, Wi] order
+// phyiscal layout is irrelavent
+template <ck::index_t NDimSpatial,
+          typename InDataType,
           typename WeiDataType,
           typename OutDataType,
           typename InElementwiseOperation,
           typename WeiElementwiseOperation,
           typename OutElementwiseOperation,
-          ck::index_t NumDimSpatial                                                     = 2,
-          typename std::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
+          typename std::enable_if<NDimSpatial >= 1 && NDimSpatial <= 3, bool>::type = false>
 struct ReferenceConvFwd : public device::BaseOperator
 {
     // Argument
@@ -86,29 +91,37 @@ struct ReferenceConvFwd : public device::BaseOperator
 
         float Run(const Argument& arg)
         {
-            if constexpr(NumDimSpatial == 1)
+            if(!(arg.input_.GetNumOfDimension() == NDimSpatial + 3 &&
+                 arg.weight_.GetNumOfDimension() == NDimSpatial + 3 &&
+                 arg.output_.GetNumOfDimension() == NDimSpatial + 3))
             {
-                auto f_ncw = [&](auto n, auto k, auto wo) {
+                throw std::runtime_error("wrong! inconsistent dimension");
+            }
+
+            if constexpr(NDimSpatial == 1)
+            {
+                auto func = [&](auto g, auto n, auto k, auto wo) {
                     float v_acc = 0;
 
-                    for(std::size_t c = 0; c < arg.weight_.mDesc.GetLengths()[1]; ++c)
+                    for(std::size_t c = 0; c < arg.weight_.GetLengths()[2]; ++c)
                     {
-                        for(std::size_t x = 0; x < arg.weight_.mDesc.GetLengths()[2]; ++x)
+                        for(std::size_t x = 0; x < arg.weight_.GetLengths()[3]; ++x)
                         {
-                            auto wi =
-                                ck::type_convert<ck::long_index_t>(wo * arg.conv_strides_[0]) +
-                                ck::type_convert<ck::long_index_t>(x * arg.conv_dilations_[0]) -
-                                ck::type_convert<ck::long_index_t>(arg.in_left_pads_[0]);
+                            auto wi = static_cast<ck::long_index_t>(wo * arg.conv_strides_[0]) +
+                                      static_cast<ck::long_index_t>(x * arg.conv_dilations_[0]) -
+                                      static_cast<ck::long_index_t>(arg.in_left_pads_[0]);
+
                             if(wi >= 0 &&
-                               ck::type_convert<std::size_t>(wi) < arg.input_.mDesc.GetLengths()[2])
+                               ck::type_convert<std::size_t>(wi) < arg.input_.GetLengths()[3])
                             {
                                 float v_in;
                                 float v_wei;
 
-                                arg.in_element_op_(v_in,
-                                                   ck::type_convert<float>(arg.input_(n, c, wi)));
-                                arg.wei_element_op_(v_wei,
-                                                    ck::type_convert<float>(arg.weight_(k, c, x)));
+                                arg.in_element_op_(
+                                    v_in, ck::type_convert<float>(arg.input_(g, n, c, wi)));
+
+                                arg.wei_element_op_(
+                                    v_wei, ck::type_convert<float>(arg.weight_(g, k, c, x)));
 
                                 v_acc += v_in * v_wei;
                             }
@@ -118,50 +131,53 @@ struct ReferenceConvFwd : public device::BaseOperator
                     float v_out;
 
                     arg.out_element_op_(v_out, v_acc);
-                    arg.output_(n, k, wo) = ck::type_convert<OutDataType>(v_out);
+
+                    arg.output_(g, n, k, wo) = ck::type_convert<OutDataType>(v_out);
                 };
 
-                make_ParallelTensorFunctor(f_ncw,
-                                           arg.output_.mDesc.GetLengths()[0],
-                                           arg.output_.mDesc.GetLengths()[1],
-                                           arg.output_.mDesc.GetLengths()[2])(
+                make_ParallelTensorFunctor(func,
+                                           arg.output_.GetLengths()[0],
+                                           arg.output_.GetLengths()[1],
+                                           arg.output_.GetLengths()[2],
+                                           arg.output_.GetLengths()[3])(
                     std::thread::hardware_concurrency());
 
                 return 0;
             }
-            else if constexpr(NumDimSpatial == 2)
+            else if constexpr(NDimSpatial == 2)
             {
-                auto f_nchw = [&](auto n, auto k, auto ho, auto wo) {
+                auto func = [&](auto g, auto n, auto k, auto ho, auto wo) {
                     float v_acc = 0;
 
-                    for(std::size_t c = 0; c < arg.weight_.mDesc.GetLengths()[1]; ++c)
+                    for(std::size_t c = 0; c < arg.weight_.GetLengths()[2]; ++c)
                     {
-                        for(std::size_t y = 0; y < arg.weight_.mDesc.GetLengths()[2]; ++y)
+                        for(std::size_t y = 0; y < arg.weight_.GetLengths()[3]; ++y)
                         {
-                            auto hi =
-                                ck::type_convert<ck::long_index_t>(ho * arg.conv_strides_[0]) +
-                                ck::type_convert<ck::long_index_t>(y * arg.conv_dilations_[0]) -
-                                ck::type_convert<ck::long_index_t>(arg.in_left_pads_[0]);
-                            for(std::size_t x = 0; x < arg.weight_.mDesc.GetLengths()[3]; ++x)
+                            auto hi = static_cast<ck::long_index_t>(ho * arg.conv_strides_[0]) +
+                                      static_cast<ck::long_index_t>(y * arg.conv_dilations_[0]) -
+                                      static_cast<ck::long_index_t>(arg.in_left_pads_[0]);
+
+                            for(std::size_t x = 0; x < arg.weight_.GetLengths()[4]; ++x)
                             {
                                 auto wi =
-                                    ck::type_convert<ck::long_index_t>(wo * arg.conv_strides_[1]) +
-                                    ck::type_convert<ck::long_index_t>(x * arg.conv_dilations_[1]) -
-                                    ck::type_convert<ck::long_index_t>(arg.in_left_pads_[1]);
+                                    static_cast<ck::long_index_t>(wo * arg.conv_strides_[1]) +
+                                    static_cast<ck::long_index_t>(x * arg.conv_dilations_[1]) -
+                                    static_cast<ck::long_index_t>(arg.in_left_pads_[1]);
+
                                 if(hi >= 0 &&
-                                   ck::type_convert<std::size_t>(hi) <
-                                       arg.input_.mDesc.GetLengths()[2] &&
+                                   ck::type_convert<std::size_t>(hi) < arg.input_.GetLengths()[3] &&
                                    wi >= 0 &&
-                                   ck::type_convert<std::size_t>(wi) <
-                                       arg.input_.mDesc.GetLengths()[3])
+                                   ck::type_convert<std::size_t>(wi) < arg.input_.GetLengths()[4])
                                 {
                                     float v_in;
                                     float v_wei;
 
                                     arg.in_element_op_(
-                                        v_in, ck::type_convert<float>(arg.input_(n, c, hi, wi)));
+                                        v_in, ck::type_convert<float>(arg.input_(g, n, c, hi, wi)));
+
                                     arg.wei_element_op_(
-                                        v_wei, ck::type_convert<float>(arg.weight_(k, c, y, x)));
+                                        v_wei, ck::type_convert<float>(arg.weight_(g, k, c, y, x)));
+
                                     v_acc += v_in * v_wei;
                                 }
                             }
@@ -171,64 +187,65 @@ struct ReferenceConvFwd : public device::BaseOperator
                     float v_out;
 
                     arg.out_element_op_(v_out, v_acc);
-                    arg.output_(n, k, ho, wo) = ck::type_convert<OutDataType>(v_out);
+
+                    arg.output_(g, n, k, ho, wo) = ck::type_convert<OutDataType>(v_out);
                 };
 
-                make_ParallelTensorFunctor(f_nchw,
-                                           arg.output_.mDesc.GetLengths()[0],
-                                           arg.output_.mDesc.GetLengths()[1],
-                                           arg.output_.mDesc.GetLengths()[2],
-                                           arg.output_.mDesc.GetLengths()[3])(
+                make_ParallelTensorFunctor(func,
+                                           arg.output_.GetLengths()[0],
+                                           arg.output_.GetLengths()[1],
+                                           arg.output_.GetLengths()[2],
+                                           arg.output_.GetLengths()[3],
+                                           arg.output_.GetLengths()[4])(
                     std::thread::hardware_concurrency());
 
                 return 0;
             }
-            else if constexpr(NumDimSpatial == 3)
+            else if constexpr(NDimSpatial == 3)
             {
-                auto f_nchw = [&](auto n, auto k, auto d_o, auto ho, auto wo) {
+                auto func = [&](auto g, auto n, auto k, auto d_o, auto ho, auto wo) {
                     float v_acc = 0;
 
-                    for(std::size_t c = 0; c < arg.weight_.mDesc.GetLengths()[1]; ++c)
+                    for(std::size_t c = 0; c < arg.weight_.GetLengths()[2]; ++c)
                     {
-                        for(std::size_t z = 0; z < arg.weight_.mDesc.GetLengths()[2]; ++z)
+                        for(std::size_t z = 0; z < arg.weight_.GetLengths()[3]; ++z)
                         {
-                            auto di =
-                                ck::type_convert<ck::long_index_t>(d_o * arg.conv_strides_[0]) +
-                                ck::type_convert<ck::long_index_t>(z * arg.conv_dilations_[0]) -
-                                ck::type_convert<ck::long_index_t>(arg.in_left_pads_[0]);
-                            for(std::size_t y = 0; y < arg.weight_.mDesc.GetLengths()[3]; ++y)
+                            auto di = static_cast<ck::long_index_t>(d_o * arg.conv_strides_[0]) +
+                                      static_cast<ck::long_index_t>(z * arg.conv_dilations_[0]) -
+                                      static_cast<ck::long_index_t>(arg.in_left_pads_[0]);
+                            for(std::size_t y = 0; y < arg.weight_.GetLengths()[4]; ++y)
                             {
                                 auto hi =
-                                    ck::type_convert<ck::long_index_t>(ho * arg.conv_strides_[1]) +
-                                    ck::type_convert<ck::long_index_t>(y * arg.conv_dilations_[1]) -
-                                    ck::type_convert<ck::long_index_t>(arg.in_left_pads_[1]);
-                                for(std::size_t x = 0; x < arg.weight_.mDesc.GetLengths()[4]; ++x)
+                                    static_cast<ck::long_index_t>(ho * arg.conv_strides_[1]) +
+                                    static_cast<ck::long_index_t>(y * arg.conv_dilations_[1]) -
+                                    static_cast<ck::long_index_t>(arg.in_left_pads_[1]);
+                                for(std::size_t x = 0; x < arg.weight_.GetLengths()[5]; ++x)
                                 {
                                     auto wi =
-                                        ck::type_convert<ck::long_index_t>(wo *
-                                                                           arg.conv_strides_[2]) +
-                                        ck::type_convert<ck::long_index_t>(x *
-                                                                           arg.conv_dilations_[2]) -
-                                        ck::type_convert<ck::long_index_t>(arg.in_left_pads_[2]);
+                                        static_cast<ck::long_index_t>(wo * arg.conv_strides_[2]) +
+                                        static_cast<ck::long_index_t>(x * arg.conv_dilations_[2]) -
+                                        static_cast<ck::long_index_t>(arg.in_left_pads_[2]);
                                     if(di >= 0 &&
                                        ck::type_convert<std::size_t>(di) <
-                                           arg.input_.mDesc.GetLengths()[2] &&
+                                           arg.input_.GetLengths()[3] &&
                                        hi >= 0 &&
                                        ck::type_convert<std::size_t>(hi) <
-                                           arg.input_.mDesc.GetLengths()[3] &&
+                                           arg.input_.GetLengths()[4] &&
                                        wi >= 0 &&
                                        ck::type_convert<std::size_t>(wi) <
-                                           arg.input_.mDesc.GetLengths()[4])
+                                           arg.input_.GetLengths()[5])
                                     {
                                         float v_in;
                                         float v_wei;
 
-                                        arg.in_element_op_(
-                                            v_in,
-                                            ck::type_convert<float>(arg.input_(n, c, di, hi, wi)));
+                                        arg.in_element_op_(v_in,
+                                                           ck::type_convert<float>(
+                                                               arg.input_(g, n, c, di, hi, wi)));
+
                                         arg.wei_element_op_(
                                             v_wei,
-                                            ck::type_convert<float>(arg.weight_(k, c, z, y, x)));
+                                            ck::type_convert<float>(arg.weight_(g, k, c, z, y, x)));
+
                                         v_acc += v_in * v_wei;
                                     }
                                 }
@@ -239,15 +256,17 @@ struct ReferenceConvFwd : public device::BaseOperator
                     float v_out;
 
                     arg.out_element_op_(v_out, v_acc);
-                    arg.output_(n, k, d_o, ho, wo) = ck::type_convert<OutDataType>(v_out);
+
+                    arg.output_(g, n, k, d_o, ho, wo) = ck::type_convert<OutDataType>(v_out);
                 };
 
-                make_ParallelTensorFunctor(f_nchw,
-                                           arg.output_.mDesc.GetLengths()[0],
-                                           arg.output_.mDesc.GetLengths()[1],
-                                           arg.output_.mDesc.GetLengths()[2],
-                                           arg.output_.mDesc.GetLengths()[3],
-                                           arg.output_.mDesc.GetLengths()[4])(
+                make_ParallelTensorFunctor(func,
+                                           arg.output_.GetLengths()[0],
+                                           arg.output_.GetLengths()[1],
+                                           arg.output_.GetLengths()[2],
+                                           arg.output_.GetLengths()[3],
+                                           arg.output_.GetLengths()[4],
+                                           arg.output_.GetLengths()[5])(
                     std::thread::hardware_concurrency());
 
                 return 0;
@@ -267,7 +286,10 @@ struct ReferenceConvFwd : public device::BaseOperator
         return true;
     }
 
-    bool IsSupportedArgument(const device::BaseArgument*) override { return true; }
+    bool IsSupportedArgument(const device::BaseArgument*) override
+    {
+        return NDimSpatial >= 1 && NDimSpatial <= 3;
+    }
 
     static auto MakeArgument(const Tensor<InDataType>& input,
                              const Tensor<WeiDataType>& weight,
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp
index 9309ef6e8f..be22003fd9 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp
@@ -7,7 +7,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp
index 44fa352024..f949f27fde 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp
@@ -7,7 +7,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm.hpp
index e3dd4de5df..6728bb1f47 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm.hpp
@@ -7,7 +7,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_2d.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_2d.hpp
index cd3383b994..c77d22f4cd 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_2d.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_2d.hpp
@@ -7,7 +7,7 @@
 #include <sstream>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation.hpp
index 33d7cbb837..7dfc3c1ed4 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation.hpp
@@ -8,7 +8,7 @@
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation_add.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation_add.hpp
index 1ae63d2f86..99102a40d4 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation_add.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation_add.hpp
@@ -8,7 +8,7 @@
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp
index 6487fe49ca..78eefe5795 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp
@@ -9,8 +9,8 @@
 #include <algorithm>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp b/library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp
index 5d9e90f71a..bfc6986d0c 100644
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp
@@ -9,8 +9,8 @@
 #include <algorithm>
 
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 namespace ck {
 namespace tensor_operation {
diff --git a/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp b/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
index 66230ac45c..783733feb6 100644
--- a/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
+++ b/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
@@ -10,22 +10,67 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-// aliasing, for commonly used type
+// aliasing, for commonly used data type
 using F64  = double;
 using F32  = float;
 using F16  = ck::half_t;
 using BF16 = ck::bhalf_t;
 
-using EMPTY_TUPLE = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
-using F16_TUPLE     = ck::Tuple<F16>;
-using F16_F16_TUPLE = ck::Tuple<F16, F16>;
+using F16_Tuple     = ck::Tuple<F16>;
+using F16_F16_Tuple = ck::Tuple<F16, F16>;
 
-using F32_TUPLE = ck::Tuple<F32>;
+using F32_Tuple = ck::Tuple<F32>;
 
+// GEMM layout
 using Row = ck::tensor_layout::gemm::RowMajor;
 using Col = ck::tensor_layout::gemm::ColumnMajor;
 
+using Row_Tuple     = ck::Tuple<Row>;
+using Row_Row_Tuple = ck::Tuple<Row, Row>;
+
+// Conv layout
+//
+using NWC   = ck::tensor_layout::convolution::NWC;
+using NHWC  = ck::tensor_layout::convolution::NHWC;
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+
+using KXC   = ck::tensor_layout::convolution::KXC;
+using KYXC  = ck::tensor_layout::convolution::KYXC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+
+using NWK   = ck::tensor_layout::convolution::NWK;
+using NHWK  = ck::tensor_layout::convolution::NHWK;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+//
+using GNWC   = ck::tensor_layout::convolution::GNWC;
+using GNHWC  = ck::tensor_layout::convolution::GNHWC;
+using GNDHWC = ck::tensor_layout::convolution::GNDHWC;
+
+using GKXC   = ck::tensor_layout::convolution::GKXC;
+using GKYXC  = ck::tensor_layout::convolution::GKYXC;
+using GKZYXC = ck::tensor_layout::convolution::GKZYXC;
+
+using GNWK   = ck::tensor_layout::convolution::GNWK;
+using GNHWK  = ck::tensor_layout::convolution::GNHWK;
+using GNDHWK = ck::tensor_layout::convolution::GNDHWK;
+
+//
+using NWGC   = ck::tensor_layout::convolution::NWGC;
+using NHWGC  = ck::tensor_layout::convolution::NHWGC;
+using NDHWGC = ck::tensor_layout::convolution::NDHWGC;
+
+using KXGC   = ck::tensor_layout::convolution::KXGC;
+using KYXGC  = ck::tensor_layout::convolution::KYXGC;
+using KZYXGC = ck::tensor_layout::convolution::KZYXGC;
+
+using NWGK   = ck::tensor_layout::convolution::NWGK;
+using NHWGK  = ck::tensor_layout::convolution::NHWGK;
+using NDHWGK = ck::tensor_layout::convolution::NDHWGK;
+
+// pointwise functor
 using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
 using Scale          = ck::tensor_operation::element_wise::Scale;
 using Bilinear       = ck::tensor_operation::element_wise::Bilinear;
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/contraction_bilinear.hpp b/library/include/ck/library/tensor_operation_instance/gpu/contraction_bilinear.hpp
index 9bb8e5ce52..a0cea7e390 100644
--- a/library/include/ck/library/tensor_operation_instance/gpu/contraction_bilinear.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/contraction_bilinear.hpp
@@ -25,7 +25,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
@@ -37,7 +37,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
@@ -49,7 +49,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
@@ -61,7 +61,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/contraction_scale.hpp b/library/include/ck/library/tensor_operation_instance/gpu/contraction_scale.hpp
index 6eb5b1d0cc..e921ecd47a 100644
--- a/library/include/ck/library/tensor_operation_instance/gpu/contraction_scale.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/contraction_scale.hpp
@@ -25,7 +25,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
@@ -37,7 +37,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
@@ -49,7 +49,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
@@ -61,7 +61,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp b/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp
new file mode 100644
index 0000000000..dd1f77b88b
--- /dev/null
+++ b/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp
@@ -0,0 +1,270 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/device_operation_instance_factory.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+// conv1d backward data
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<1,
+                                                  NWC,
+                                                  KXC,
+                                                  NWK,
+                                                  BF16,
+                                                  BF16,
+                                                  BF16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvBwdData<1, NWC, KXC, NWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvBwdData<1, NWC, KXC, NWK, F32, F32, F32, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<1,
+                                                  NWC,
+                                                  KXC,
+                                                  NWK,
+                                                  int8_t,
+                                                  int8_t,
+                                                  int8_t,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+// conv2d backward data
+void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  BF16,
+                                                  BF16,
+                                                  BF16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  F16,
+                                                  F16,
+                                                  F16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  F32,
+                                                  F32,
+                                                  F32,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  int8_t,
+                                                  int8_t,
+                                                  int8_t,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+// conv3d backward data
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  BF16,
+                                                  BF16,
+                                                  BF16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  F16,
+                                                  F16,
+                                                  F16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  F32,
+                                                  F32,
+                                                  F32,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  int8_t,
+                                                  int8_t,
+                                                  int8_t,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances);
+
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceConvBwdData<
+    NumDimSpatial,
+    InLayout,
+    WeiLayout,
+    OutLayout,
+    InDataType,
+    WeiDataType,
+    OutDataType,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough>>
+{
+    using DeviceOp = DeviceConvBwdData<NumDimSpatial,
+                                       InLayout,
+                                       WeiLayout,
+                                       OutLayout,
+                                       InDataType,
+                                       WeiDataType,
+                                       OutDataType,
+                                       ck::tensor_operation::element_wise::PassThrough,
+                                       ck::tensor_operation::element_wise::PassThrough,
+                                       ck::tensor_operation::element_wise::PassThrough>;
+
+    static auto GetInstances()
+    {
+        std::vector<std::unique_ptr<DeviceOp>> op_ptrs;
+
+        if constexpr(NumDimSpatial == 1 && is_same_v<InLayout, NWC> && is_same_v<WeiLayout, KXC> &&
+                     is_same_v<OutLayout, NWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, NHWC> &&
+                          is_same_v<WeiLayout, KYXC> && is_same_v<OutLayout, NHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 3 && is_same_v<InLayout, NDHWC> &&
+                          is_same_v<WeiLayout, KZYXC> && is_same_v<OutLayout, NDHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances(op_ptrs);
+            }
+        }
+
+        return op_ptrs;
+    }
+};
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_weight.hpp b/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_weight.hpp
new file mode 100644
index 0000000000..00b96a6cf8
--- /dev/null
+++ b/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_weight.hpp
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/device_operation_instance_factory.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+// conv1d backward weight
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+// conv2d backward weight
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+// conv3d backward weight
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceConvBwdWeight<
+    NumDimSpatial,
+    InLayout,
+    WeiLayout,
+    OutLayout,
+    InDataType,
+    WeiDataType,
+    OutDataType,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough>>
+{
+    using DeviceOp = DeviceConvBwdWeight<NumDimSpatial,
+                                         InLayout,
+                                         WeiLayout,
+                                         OutLayout,
+                                         InDataType,
+                                         WeiDataType,
+                                         OutDataType,
+                                         ck::tensor_operation::element_wise::PassThrough,
+                                         ck::tensor_operation::element_wise::PassThrough,
+                                         ck::tensor_operation::element_wise::PassThrough>;
+
+    static auto GetInstances()
+    {
+        std::vector<std::unique_ptr<DeviceOp>> op_ptrs;
+
+        if constexpr(NumDimSpatial == 1 && is_same_v<InLayout, NWC> && is_same_v<WeiLayout, KXC> &&
+                     is_same_v<OutLayout, NWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> && is_same_v<WeiDataType, float> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_f32_bf16_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, NHWC> &&
+                          is_same_v<WeiLayout, KYXC> && is_same_v<OutLayout, NHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> && is_same_v<WeiDataType, float> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_f32_bf16_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 3 && is_same_v<InLayout, NDHWC> &&
+                          is_same_v<WeiLayout, KZYXC> && is_same_v<OutLayout, NDHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> && is_same_v<WeiDataType, float> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances(op_ptrs);
+            }
+        }
+
+        return op_ptrs;
+    }
+};
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp b/library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp
new file mode 100644
index 0000000000..62f28c9b11
--- /dev/null
+++ b/library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/device_operation_instance_factory.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+// conv2d forward
+void add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvFwd<2, NHWC, KYXC, NHWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvFwd<2,
+                                              NHWC,
+                                              KYXC,
+                                              NHWK,
+                                              BF16,
+                                              BF16,
+                                              BF16,
+                                              PassThrough,
+                                              PassThrough,
+                                              PassThrough>>>& instances);
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvFwd<2, NHWC, KYXC, NHWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvFwd<2, NHWC, KYXC, NHWK, F32, F32, F32, PassThrough, PassThrough, PassThrough>>>&
+        instances);
+
+void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceConvFwd<2,
+                                              NHWC,
+                                              KYXC,
+                                              NHWK,
+                                              int8_t,
+                                              int8_t,
+                                              int8_t,
+                                              PassThrough,
+                                              PassThrough,
+                                              PassThrough>>>& instances);
+
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+struct DeviceOperationInstanceFactory<
+    ck::tensor_operation::device::DeviceConvFwd<NumDimSpatial,
+                                                InLayout,
+                                                WeiLayout,
+                                                OutLayout,
+                                                InDataType,
+                                                WeiDataType,
+                                                OutDataType,
+                                                ck::tensor_operation::element_wise::PassThrough,
+                                                ck::tensor_operation::element_wise::PassThrough,
+                                                ck::tensor_operation::element_wise::PassThrough>>
+{
+    using DeviceOp = DeviceConvFwd<NumDimSpatial,
+                                   InLayout,
+                                   WeiLayout,
+                                   OutLayout,
+                                   InDataType,
+                                   WeiDataType,
+                                   OutDataType,
+                                   ck::tensor_operation::element_wise::PassThrough,
+                                   ck::tensor_operation::element_wise::PassThrough,
+                                   ck::tensor_operation::element_wise::PassThrough>;
+
+    static auto GetInstances()
+    {
+        std::vector<std::unique_ptr<DeviceOp>> op_ptrs;
+
+        if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, NHWC> &&
+                     is_same_v<WeiLayout, KYXC> && is_same_v<OutLayout, NHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(op_ptrs);
+                add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(op_ptrs);
+            }
+        }
+
+        return op_ptrs;
+    }
+};
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/gemm_add_add_fastgelu.hpp b/library/include/ck/library/tensor_operation_instance/gpu/gemm_add_add_fastgelu.hpp
index e2cd64b34e..09d8e8b95b 100644
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm_add_add_fastgelu.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm_add_add_fastgelu.hpp
@@ -19,49 +19,53 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances(
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
                                                     Row,
+                                                    Row_Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_F16_TUPLE,
+                                                    F16_F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
                                                     AddAddFastGelu>>>&);
 
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
                                                     Col,
+                                                    Row_Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_F16_TUPLE,
+                                                    F16_F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
                                                     AddAddFastGelu>>>&);
 
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances(
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
                                                     Row,
+                                                    Row_Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_F16_TUPLE,
+                                                    F16_F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
                                                     AddAddFastGelu>>>&);
 
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances(
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
                                                     Col,
+                                                    Row_Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_F16_TUPLE,
+                                                    F16_F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
@@ -70,7 +74,9 @@ void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instanc
 // GEMM + Add + Add + FastGelu
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename D0Layout,
+          typename D1Layout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
           typename D0DataType,
@@ -79,7 +85,8 @@ template <typename ALayout,
 struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemmMultipleD<
     ALayout,
     BLayout,
-    DELayout,
+    ck::Tuple<D0Layout, D1Layout>,
+    ELayout,
     ADataType,
     BDataType,
     ck::Tuple<D0DataType, D1DataType>,
@@ -90,7 +97,8 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemmMu
 {
     using DeviceOp = DeviceGemmMultipleD<ALayout,
                                          BLayout,
-                                         DELayout,
+                                         ck::Tuple<D0Layout, D1Layout>,
+                                         ELayout,
                                          ADataType,
                                          BDataType,
                                          ck::Tuple<D0DataType, D1DataType>,
@@ -108,27 +116,31 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemmMu
                      is_same_v<EDataType, half_t>)
         {
             if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Row> &&
-                         is_same_v<DELayout, Row>)
+                         is_same_v<D0Layout, Row> && is_same_v<D1Layout, Row> &&
+                         is_same_v<ELayout, Row>)
             {
-                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances(
+                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instances(
                     op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Col> &&
-                              is_same_v<DELayout, Row>)
+                              is_same_v<D0Layout, Row> && is_same_v<D1Layout, Row> &&
+                              is_same_v<ELayout, Row>)
             {
-                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(
+                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instances(
                     op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Row> &&
-                              is_same_v<DELayout, Row>)
+                              is_same_v<D0Layout, Row> && is_same_v<D1Layout, Row> &&
+                              is_same_v<ELayout, Row>)
             {
-                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances(
+                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instances(
                     op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Col> &&
-                              is_same_v<DELayout, Row>)
+                              is_same_v<D0Layout, Row> && is_same_v<D1Layout, Row> &&
+                              is_same_v<ELayout, Row>)
             {
-                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances(
+                add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instances(
                     op_ptrs);
             }
         }
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/gemm_bilinear.hpp b/library/include/ck/library/tensor_operation_instance/gpu/gemm_bilinear.hpp
index 37731fde06..ef70504f29 100644
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm_bilinear.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm_bilinear.hpp
@@ -19,49 +19,53 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances(
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
                                                     Row,
+                                                    Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_TUPLE,
+                                                    F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
                                                     Bilinear>>>& instances);
 
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances(
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
                                                     Col,
+                                                    Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_TUPLE,
+                                                    F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
                                                     Bilinear>>>& instances);
 
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances(
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
                                                     Row,
+                                                    Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_TUPLE,
+                                                    F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
                                                     Bilinear>>>& instances);
 
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
                                                     Col,
+                                                    Row_Tuple,
                                                     Row,
                                                     F16,
                                                     F16,
-                                                    F16_TUPLE,
+                                                    F16_Tuple,
                                                     F16,
                                                     PassThrough,
                                                     PassThrough,
@@ -70,7 +74,8 @@ void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(
 // GEMM + Bilinear
 template <typename ALayout,
           typename BLayout,
-          typename DELayout,
+          typename DLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
           typename DDataType,
@@ -78,7 +83,8 @@ template <typename ALayout,
 struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemmMultipleD<
     ALayout,
     BLayout,
-    DELayout,
+    ck::Tuple<DLayout>,
+    ELayout,
     ADataType,
     BDataType,
     ck::Tuple<DDataType>,
@@ -89,7 +95,8 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemmMu
 {
     using DeviceOp = DeviceGemmMultipleD<ALayout,
                                          BLayout,
-                                         DELayout,
+                                         ck::Tuple<DLayout>,
+                                         ELayout,
                                          ADataType,
                                          BDataType,
                                          ck::Tuple<DDataType>,
@@ -106,24 +113,28 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGemmMu
                      is_same_v<DDataType, half_t> && is_same_v<EDataType, half_t>)
         {
             if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Row> &&
-                         is_same_v<DELayout, Row>)
+                         is_same_v<DLayout, Row> && is_same_v<ELayout, Row>)
             {
-                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
+                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instances(
+                    op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Col> &&
-                              is_same_v<DELayout, Row>)
+                              is_same_v<DLayout, Row> && is_same_v<ELayout, Row>)
             {
-                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
+                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instances(
+                    op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Row> &&
-                              is_same_v<DELayout, Row>)
+                              is_same_v<DLayout, Row> && is_same_v<ELayout, Row>)
             {
-                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances(op_ptrs);
+                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instances(
+                    op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Col> &&
-                              is_same_v<DELayout, Row>)
+                              is_same_v<DLayout, Row> && is_same_v<ELayout, Row>)
             {
-                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances(op_ptrs);
+                add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instances(
+                    op_ptrs);
             }
         }
 
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp b/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp
new file mode 100644
index 0000000000..aba28d3c3d
--- /dev/null
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp
@@ -0,0 +1,352 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/device_operation_instance_factory.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+// grouped conv1d forward, GNWC/GKXC/GNWK
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              BF16,
+                                                              BF16,
+                                                              Empty_Tuple,
+                                                              BF16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              F32,
+                                                              F32,
+                                                              Empty_Tuple,
+                                                              F32,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              int8_t,
+                                                              int8_t,
+                                                              Empty_Tuple,
+                                                              int8_t,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+// grouped conv2d forward, GNHWC/GKYXC/GNHWK
+void add_device_grouped_conv1d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              BF16,
+                                                              BF16,
+                                                              Empty_Tuple,
+                                                              BF16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              F32,
+                                                              F32,
+                                                              Empty_Tuple,
+                                                              F32,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              int8_t,
+                                                              int8_t,
+                                                              Empty_Tuple,
+                                                              int8_t,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+// grouped conv2d forward, NHWGC/KYXGC/NHWGK
+void add_device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              NHWGC,
+                                                              KYXGC,
+                                                              Empty_Tuple,
+                                                              NHWGK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+// grouped conv3d forward, GNDHWC/GKZYXC/GNDHWK
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              BF16,
+                                                              BF16,
+                                                              Empty_Tuple,
+                                                              BF16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              F32,
+                                                              F32,
+                                                              Empty_Tuple,
+                                                              F32,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              int8_t,
+                                                              int8_t,
+                                                              Empty_Tuple,
+                                                              int8_t,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances);
+
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD<
+    NumDimSpatial,
+    InLayout,
+    WeiLayout,
+    Empty_Tuple,
+    OutLayout,
+    InDataType,
+    WeiDataType,
+    Empty_Tuple,
+    OutDataType,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough>>
+{
+    using DeviceOp = DeviceGroupedConvFwdMultipleD<NumDimSpatial,
+                                                   InLayout,
+                                                   WeiLayout,
+                                                   Empty_Tuple,
+                                                   OutLayout,
+                                                   InDataType,
+                                                   WeiDataType,
+                                                   Empty_Tuple,
+                                                   OutDataType,
+                                                   ck::tensor_operation::element_wise::PassThrough,
+                                                   ck::tensor_operation::element_wise::PassThrough,
+                                                   ck::tensor_operation::element_wise::PassThrough>;
+
+    static auto GetInstances()
+    {
+        std::vector<std::unique_ptr<DeviceOp>> op_ptrs;
+
+        if constexpr(NumDimSpatial == 1 && is_same_v<InLayout, GNWC> &&
+                     is_same_v<WeiLayout, GKXC> && is_same_v<OutLayout, GNWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, GNHWC> &&
+                          is_same_v<WeiLayout, GKYXC> && is_same_v<OutLayout, GNHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_grouped_conv1d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, NHWGC> &&
+                          is_same_v<WeiLayout, KYXGC> && is_same_v<OutLayout, NHWGK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                // no instance
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                // no instance
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                // no instance
+            }
+        }
+        else if constexpr(NumDimSpatial == 3 && is_same_v<InLayout, GNDHWC> &&
+                          is_same_v<WeiLayout, GKZYXC> && is_same_v<OutLayout, GNDHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> &&
+                              is_same_v<WeiDataType, ck::bhalf_t> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, int8_t> && is_same_v<WeiDataType, int8_t> &&
+                              is_same_v<OutDataType, int8_t>)
+            {
+                add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instances(op_ptrs);
+            }
+        }
+
+        return op_ptrs;
+    }
+};
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/include/ck/library/tensor_operation_instance/gpu/grouped_gemm.hpp b/library/include/ck/library/tensor_operation_instance/gpu/grouped_gemm.hpp
index 30f8f809b0..c64598dadd 100644
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_gemm.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_gemm.hpp
@@ -16,15 +16,14 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-using DsType = Tuple<>;
-
 void add_device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Row,
                                                   Row,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
@@ -33,10 +32,11 @@ void add_device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(
 void add_device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Row,
                                                   Col,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
@@ -45,10 +45,11 @@ void add_device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(
 void add_device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Col,
                                                   Row,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
@@ -57,10 +58,11 @@ void add_device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instances(
 void add_device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Col,
                                                   Col,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
@@ -68,18 +70,18 @@ void add_device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instances(
 
 template <typename ALayout,
           typename BLayout,
-          typename CLayout,
+          typename ELayout,
           typename ADataType,
           typename BDataType,
-          typename DsDataType,
           typename EDataType>
 struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGroupedGemm<
     ALayout,
     BLayout,
-    CLayout,
+    Empty_Tuple,
+    ELayout,
     ADataType,
     BDataType,
-    DsDataType,
+    Empty_Tuple,
     EDataType,
     ck::tensor_operation::element_wise::PassThrough,
     ck::tensor_operation::element_wise::PassThrough,
@@ -87,10 +89,11 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGroupe
 {
     using DeviceOp = DeviceGroupedGemm<ALayout,
                                        BLayout,
-                                       CLayout,
+                                       Empty_Tuple,
+                                       ELayout,
                                        ADataType,
                                        BDataType,
-                                       DsDataType,
+                                       Empty_Tuple,
                                        EDataType,
                                        ck::tensor_operation::element_wise::PassThrough,
                                        ck::tensor_operation::element_wise::PassThrough,
@@ -104,22 +107,22 @@ struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceGroupe
                      is_same_v<EDataType, half_t>)
         {
             if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Row> &&
-                         is_same_v<CLayout, Row>)
+                         is_same_v<ELayout, Row>)
             {
                 add_device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Col> &&
-                              is_same_v<CLayout, Row>)
+                              is_same_v<ELayout, Row>)
             {
                 add_device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Row> &&
-                              is_same_v<CLayout, Row>)
+                              is_same_v<ELayout, Row>)
             {
                 add_device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
             }
             else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Col> &&
-                              is_same_v<CLayout, Row>)
+                              is_same_v<ELayout, Row>)
             {
                 add_device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
             }
diff --git a/library/include/ck/library/utility/check_err.hpp b/library/include/ck/library/utility/check_err.hpp
index fef0d8e033..de09ed873d 100644
--- a/library/include/ck/library/utility/check_err.hpp
+++ b/library/include/ck/library/utility/check_err.hpp
@@ -13,7 +13,9 @@
 #include <type_traits>
 #include <vector>
 
+#include "ck/ck.hpp"
 #include "ck/utility/data_type.hpp"
+#include "ck/host_utility/io.hpp"
 
 namespace ck {
 namespace utils {
@@ -194,10 +196,3 @@ check_err(const std::vector<T>& out,
 
 } // namespace utils
 } // namespace ck
-
-template <typename T>
-std::ostream& operator<<(std::ostream& os, const std::vector<T>& v)
-{
-    std::copy(std::begin(v), std::end(v), std::ostream_iterator<T>(os, " "));
-    return os;
-}
diff --git a/library/include/ck/library/host_tensor/conv_common.hpp b/library/include/ck/library/utility/conv_common.hpp
similarity index 100%
rename from library/include/ck/library/host_tensor/conv_common.hpp
rename to library/include/ck/library/utility/conv_common.hpp
diff --git a/library/include/ck/library/utility/conv_util.hpp b/library/include/ck/library/utility/conv_util.hpp
deleted file mode 100644
index e57bde8add..0000000000
--- a/library/include/ck/library/utility/conv_util.hpp
+++ /dev/null
@@ -1,574 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#pragma once
-
-#include <cstdlib>
-#include <functional>
-#include <iterator>
-#include <numeric>
-#include <sstream>
-#include <tuple>
-#include <type_traits>
-#include <vector>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/fill.hpp"
-#include "ck/library/utility/op_instance_engine.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-
-using DeviceConvFwdNoOpPtr = DeviceConvFwdPtr<element_wise::PassThrough,
-                                              element_wise::PassThrough,
-                                              element_wise::PassThrough>;
-namespace instance {
-
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-
-} // namespace instance
-namespace instance {
-
-void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-
-} // namespace instance
-namespace instance {
-
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-
-} // namespace instance
-
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
-
-namespace ck {
-namespace utils {
-namespace conv {
-
-using DeviceConvFwdNoOpPtr =
-    ck::tensor_operation::device::DeviceConvFwdPtr<ck::tensor_operation::element_wise::PassThrough,
-                                                   ck::tensor_operation::element_wise::PassThrough,
-                                                   ck::tensor_operation::element_wise::PassThrough>;
-
-/**
- * @brief      Calculate number of FLOPs for Convolution
- *
- * @param[in]  N                       Batch size.
- * @param[in]  C                       Number of input channels.
- * @param[in]  K                       Number of output channels.
- * @param[in]  filter_spatial_lengths  Filter spatial dimensions lengths.
- * @param[in]  output_spatial_lengths  Convolution output spatial dimensions
- *                                     lengths.
- *
- * @return     The number of flops.
- */
-std::size_t get_flops(ck::index_t N,
-                      ck::index_t C,
-                      ck::index_t K,
-                      const std::vector<ck::index_t>& filter_spatial_lengths,
-                      const std::vector<ck::index_t>& output_spatial_lengths);
-
-/**
- * @brief      Calculate number of bytes read/write by convolution algorithm.
- *
- * @param[in]  N                       Batch size.
- * @param[in]  C                       Number of input channels.
- * @param[in]  K                       Number of output channels.
- * @param[in]  input_spatial_lengths   Input spatial dimensions lengths.
- * @param[in]  filter_spatial_lengths  Filter spatial dimensions lengths.
- * @param[in]  output_spatial_lengths  Output spatial dimensions lengths
- *
- * @tparam     InDataType              Input tensor data type.
- * @tparam     WeiDataType             Weights tensor data type.
- * @tparam     OutDataType             Output tensor data type.
- *
- * @return     The number of used bytes.
- */
-template <typename InDataType  = float,
-          typename WeiDataType = InDataType,
-          typename OutDataType = InDataType>
-std::size_t get_btype(ck::index_t N,
-                      ck::index_t C,
-                      ck::index_t K,
-                      const std::vector<ck::index_t>& input_spatial_lengths,
-                      const std::vector<ck::index_t>& filter_spatial_lengths,
-                      const std::vector<ck::index_t>& output_spatial_lengths)
-{
-    // sizeof(InDataType) * (N * C * <input spatial lengths product>) +
-    // sizeof(WeiDataType) * (K * C * <filter spatial lengths product>) +
-    // sizeof(OutDataType) * (N * K * <output spatial lengths product>);
-    return sizeof(InDataType) * (N * C *
-                                 std::accumulate(std::begin(input_spatial_lengths),
-                                                 std::end(input_spatial_lengths),
-                                                 static_cast<std::size_t>(1),
-                                                 std::multiplies<std::size_t>())) +
-           sizeof(WeiDataType) * (K * C *
-                                  std::accumulate(std::begin(filter_spatial_lengths),
-                                                  std::end(filter_spatial_lengths),
-                                                  static_cast<std::size_t>(1),
-                                                  std::multiplies<std::size_t>())) +
-           sizeof(OutDataType) * (N * K *
-                                  std::accumulate(std::begin(output_spatial_lengths),
-                                                  std::end(output_spatial_lengths),
-                                                  static_cast<std::size_t>(1),
-                                                  std::multiplies<std::size_t>()));
-}
-
-struct ConvParams
-{
-    ConvParams();
-    ConvParams(ck::index_t n_dim,
-               ck::index_t n_batch,
-               ck::index_t n_out_channels,
-               ck::index_t n_in_channels,
-               const std::vector<ck::index_t>& filters_len,
-               const std::vector<ck::index_t>& input_len,
-               const std::vector<ck::index_t>& strides,
-               const std::vector<ck::index_t>& dilations,
-               const std::vector<ck::index_t>& left_pads,
-               const std::vector<ck::index_t>& right_pads);
-
-    ck::index_t num_dim_spatial_;
-    ck::index_t N_;
-    ck::index_t K_;
-    ck::index_t C_;
-
-    std::vector<ck::index_t> filter_spatial_lengths_;
-    std::vector<ck::index_t> input_spatial_lengths_;
-
-    std::vector<ck::index_t> conv_filter_strides_;
-    std::vector<ck::index_t> conv_filter_dilations_;
-
-    std::vector<ck::index_t> input_left_pads_;
-    std::vector<ck::index_t> input_right_pads_;
-
-    std::vector<ck::index_t> GetOutputSpatialLengths() const;
-};
-
-ConvParams parse_conv_params(int num_dim_spatial, int arg_idx, char* const argv[]);
-
-/**
- * @brief      Gets the host tensor descriptor.
- *
- * @param[in]  dims          The tensor dimensions lengths. Always in NCHW format.
- * @param[in]  layout        The tensor data layout.
- *
- * @tparam     TensorLayout  Layout type.
- *
- * @return     The host tensor descriptor object.
- */
-template <typename TensorLayout>
-HostTensorDescriptor get_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                const TensorLayout& layout)
-{
-    std::size_t C = dims[1];
-    // 1D
-    if constexpr(std::is_same<TensorLayout, ck::tensor_layout::convolution::NCW>::value ||
-                 std::is_same<TensorLayout, ck::tensor_layout::convolution::KCX>::value ||
-                 std::is_same<TensorLayout, ck::tensor_layout::convolution::NKW>::value)
-    {
-
-        return HostTensorDescriptor(dims, std::vector<std::size_t>{C * dims[2], dims[2], 1});
-    }
-    else if constexpr(std::is_same<TensorLayout, ck::tensor_layout::convolution::NWC>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::KXC>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::NWK>::value)
-    {
-        return HostTensorDescriptor(dims, std::vector<std::size_t>{C * dims[2], 1, C});
-    }
-    // 2D
-    else if constexpr(std::is_same<TensorLayout, ck::tensor_layout::convolution::NCHW>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::KCYX>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::NKHW>::value)
-    {
-
-        return HostTensorDescriptor(
-            dims, std::vector<std::size_t>{C * dims[2] * dims[3], dims[2] * dims[3], dims[3], 1});
-    }
-    else if constexpr(std::is_same<TensorLayout, ck::tensor_layout::convolution::NHWC>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::KYXC>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::NHWK>::value)
-    {
-        return HostTensorDescriptor(
-            dims, std::vector<std::size_t>{C * dims[2] * dims[3], 1, dims[3] * C, C});
-    }
-    // 3D
-    else if constexpr(std::is_same<TensorLayout, ck::tensor_layout::convolution::NCDHW>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::KCZYX>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::NKDHW>::value)
-    {
-
-        return HostTensorDescriptor(dims,
-                                    std::vector<std::size_t>{C * dims[2] * dims[3] * dims[4],
-                                                             dims[2] * dims[3] * dims[4],
-                                                             dims[3] * dims[4],
-                                                             dims[4],
-                                                             1});
-    }
-    else if constexpr(std::is_same<TensorLayout, ck::tensor_layout::convolution::NDHWC>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::KZYXC>::value ||
-                      std::is_same<TensorLayout, ck::tensor_layout::convolution::NDHWK>::value)
-    {
-        return HostTensorDescriptor(
-            dims,
-            std::vector<std::size_t>{
-                C * dims[2] * dims[3] * dims[4], 1, C * dims[3] * dims[4], C * dims[4], C});
-    }
-
-    std::stringstream err_msg;
-    err_msg << "Unsupported data layout provided: " << layout << "!";
-    throw std::runtime_error(err_msg.str());
-}
-
-HostTensorDescriptor get_output_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                       int num_dim_spatial = 2);
-
-HostTensorDescriptor get_filters_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                        int num_dim_spatial = 2);
-
-HostTensorDescriptor get_input_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                      int num_dim_spatial = 2);
-
-template <ck::index_t NDim,
-          typename InDataType  = float,
-          typename WeiDataType = float,
-          typename OutDataType = float>
-void run_reference_convolution_forward(const ConvParams& params,
-                                       const Tensor<InDataType>& input,
-                                       const Tensor<WeiDataType>& weights,
-                                       Tensor<OutDataType>& output)
-{
-    using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-    auto ref_conv     = ck::tensor_operation::host::ReferenceConvFwd<InDataType,
-                                                                 WeiDataType,
-                                                                 OutDataType,
-                                                                 PassThrough,
-                                                                 PassThrough,
-                                                                 PassThrough,
-                                                                 NDim>();
-    auto ref_invoker  = ref_conv.MakeInvoker();
-    auto ref_argument = ref_conv.MakeArgument(input,
-                                              weights,
-                                              output,
-                                              params.conv_filter_strides_,
-                                              params.conv_filter_dilations_,
-                                              params.input_left_pads_,
-                                              params.input_right_pads_,
-                                              PassThrough{},
-                                              PassThrough{},
-                                              PassThrough{});
-
-    ref_invoker.Run(ref_argument);
-}
-
-template <typename InDataType, typename WeiDataType, typename OutDataType>
-struct ConvolutionFwdInstances;
-
-template <>
-struct ConvolutionFwdInstances<float, float, float>
-{
-    template <int NumDimSpatial,
-              typename std::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
-    static std::vector<DeviceConvFwdNoOpPtr> Get()
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if constexpr(NumDimSpatial == 1)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 3)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <>
-struct ConvolutionFwdInstances<half_t, half_t, half_t>
-{
-    template <int NumDimSpatial,
-              typename std::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
-    static std::vector<DeviceConvFwdNoOpPtr> Get()
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if constexpr(NumDimSpatial == 1)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instances(conv_ptrs);
-            return conv_ptrs;
-        }
-        else if constexpr(NumDimSpatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(conv_ptrs);
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 3)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <>
-struct ConvolutionFwdInstances<bhalf_t, bhalf_t, bhalf_t>
-{
-    template <int NumDimSpatial,
-              typename std::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
-    static std::vector<DeviceConvFwdNoOpPtr> Get()
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if constexpr(NumDimSpatial == 1)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 3)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <>
-struct ConvolutionFwdInstances<int8_t, int8_t, int8_t>
-{
-    template <int NumDimSpatial,
-              typename std::enable_if<NumDimSpatial >= 1 && NumDimSpatial <= 3, bool>::type = false>
-    static std::vector<DeviceConvFwdNoOpPtr> Get()
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if constexpr(NumDimSpatial == 1)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(conv_ptrs);
-        }
-        else if constexpr(NumDimSpatial == 3)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <typename InDataType,
-          typename WeiDataType,
-          typename OutDataType,
-          typename InLayout         = ck::tensor_layout::convolution::NHWC,
-          typename WeiLayout        = ck::tensor_layout::convolution::KYXC,
-          typename OutLayout        = ck::tensor_layout::convolution::NHWK,
-          typename InElementwiseOp  = ck::tensor_operation::element_wise::PassThrough,
-          typename WeiElementwiseOp = ck::tensor_operation::element_wise::PassThrough,
-          typename OutElementwiseOp = ck::tensor_operation::element_wise::PassThrough,
-          typename InputInitFun     = FillUniformDistribution<InDataType>,
-          typename WeightsInitFun   = FillUniformDistribution<WeiDataType>>
-class ConvFwdOpInstance : public ck::utils::OpInstance<OutDataType, InDataType, WeiDataType>
-{
-    using DeviceConvFwdOp = tensor_operation::device::
-        DeviceConvFwd<InElementwiseOp, WeiElementwiseOp, OutElementwiseOp>;
-    using DeviceMemPtr  = std::unique_ptr<DeviceMem>;
-    using DeviceBuffers = std::vector<DeviceMemPtr>;
-    using BaseType      = ck::utils::OpInstance<OutDataType, InDataType, WeiDataType>;
-    template <typename T>
-    using TensorPtr      = std::unique_ptr<Tensor<T>>;
-    using InTensorsTuple = std::tuple<TensorPtr<InDataType>, TensorPtr<WeiDataType>>;
-
-    public:
-    ConvFwdOpInstance()                         = delete;
-    ConvFwdOpInstance(const ConvFwdOpInstance&) = default;
-    ConvFwdOpInstance& operator=(const ConvFwdOpInstance&) = default;
-
-    ConvFwdOpInstance(const ConvParams& params,
-                      bool do_init                         = true,
-                      const InputInitFun& input_init_f     = InputInitFun(),
-                      const WeightsInitFun& weights_init_f = WeightsInitFun())
-        : BaseType(),
-          params_{params},
-          output_spatial_lengths_{params.GetOutputSpatialLengths()},
-          do_init_{do_init},
-          input_init_f_{input_init_f},
-          weights_init_f_{weights_init_f}
-    {
-    }
-
-    virtual ~ConvFwdOpInstance() override{};
-
-    virtual InTensorsTuple GetInputTensors() const override
-    {
-        std::vector<std::size_t> input_dims{static_cast<std::size_t>(params_.N_),
-                                            static_cast<std::size_t>(params_.C_)};
-        input_dims.insert(std::end(input_dims),
-                          std::begin(params_.input_spatial_lengths_),
-                          std::end(params_.input_spatial_lengths_));
-
-        std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params_.K_),
-                                             static_cast<std::size_t>(params_.C_)};
-        filter_dims.insert(std::end(filter_dims),
-                           std::begin(params_.filter_spatial_lengths_),
-                           std::end(params_.filter_spatial_lengths_));
-
-        auto input = std::make_unique<Tensor<InDataType>>(
-            get_host_tensor_descriptor(input_dims, InLayout{}));
-        auto weights = std::make_unique<Tensor<WeiDataType>>(
-            get_host_tensor_descriptor(filter_dims, WeiLayout{}));
-
-        if(do_init_)
-        {
-            input_init_f_(input->begin(), input->end());
-            weights_init_f_(weights->begin(), weights->end());
-        }
-
-        return std::make_tuple(std::move(input), std::move(weights));
-    }
-
-    virtual TensorPtr<OutDataType> GetOutputTensor() const override
-    {
-        std::vector<std::size_t> output_dims{static_cast<std::size_t>(params_.N_),
-                                             static_cast<std::size_t>(params_.K_)};
-        output_dims.insert(std::end(output_dims),
-                           std::begin(output_spatial_lengths_),
-                           std::end(output_spatial_lengths_));
-        auto output = std::make_unique<Tensor<OutDataType>>(
-            get_host_tensor_descriptor(output_dims, OutLayout{}));
-
-        if(do_init_)
-        {
-            std::fill(output->begin(), output->end(), OutDataType(0.f));
-        }
-        return output;
-    }
-
-    virtual std::unique_ptr<tensor_operation::device::BaseInvoker>
-    MakeInvokerPointer(tensor_operation::device::BaseOperator* op_ptr) const override
-    {
-        static_assert(
-            std::is_same_v<InElementwiseOp, ck::tensor_operation::element_wise::PassThrough>);
-        static_assert(
-            std::is_same_v<OutElementwiseOp, ck::tensor_operation::element_wise::PassThrough>);
-        static_assert(
-            std::is_same_v<WeiElementwiseOp, ck::tensor_operation::element_wise::PassThrough>);
-
-        auto conv_ptr = dynamic_cast<DeviceConvFwdOp*>(op_ptr);
-        if(!conv_ptr)
-        {
-            throw std::runtime_error(
-                "[ConvFwdOpInstance]: couldn't cast op_ptr to DeviceConvFwdNoOpPtr type!");
-        }
-        return conv_ptr->MakeInvokerPointer();
-    }
-
-    virtual std::unique_ptr<tensor_operation::device::BaseArgument>
-    MakeArgumentPointer(tensor_operation::device::BaseOperator* op_ptr,
-                        const DeviceBuffers& in_device_buffers,
-                        const DeviceMemPtr& out_device_buffer) const override
-    {
-        static_assert(
-            std::is_same_v<InElementwiseOp, ck::tensor_operation::element_wise::PassThrough>);
-        static_assert(
-            std::is_same_v<OutElementwiseOp, ck::tensor_operation::element_wise::PassThrough>);
-        static_assert(
-            std::is_same_v<WeiElementwiseOp, ck::tensor_operation::element_wise::PassThrough>);
-
-        auto conv_ptr = dynamic_cast<DeviceConvFwdOp*>(op_ptr);
-        if(!conv_ptr)
-        {
-            throw std::runtime_error(
-                "[ConvFwdOpInstance]: couldn't cast op_ptr to DeviceConvFwdNoOpPtr type!");
-        }
-
-        return conv_ptr->MakeArgumentPointer(
-            static_cast<InDataType*>(in_device_buffers[0]->GetDeviceBuffer()),
-            static_cast<WeiDataType*>(in_device_buffers[1]->GetDeviceBuffer()),
-            static_cast<OutDataType*>(out_device_buffer->GetDeviceBuffer()),
-            params_.N_,
-            params_.K_,
-            params_.C_,
-            params_.input_spatial_lengths_,
-            params_.filter_spatial_lengths_,
-            output_spatial_lengths_,
-            params_.conv_filter_strides_,
-            params_.conv_filter_dilations_,
-            params_.input_left_pads_,
-            params_.input_right_pads_,
-            InElementwiseOp{},
-            WeiElementwiseOp{},
-            OutElementwiseOp{});
-    }
-
-    virtual std::size_t GetFlops() const override
-    {
-        return get_flops(params_.N_,
-                         params_.C_,
-                         params_.K_,
-                         params_.filter_spatial_lengths_,
-                         output_spatial_lengths_);
-    }
-
-    virtual std::size_t GetBtype() const override
-    {
-        return get_btype<InDataType, WeiDataType, OutDataType>(params_.N_,
-                                                               params_.C_,
-                                                               params_.K_,
-                                                               params_.input_spatial_lengths_,
-                                                               params_.filter_spatial_lengths_,
-                                                               output_spatial_lengths_);
-    }
-
-    private:
-    const ConvParams& params_;
-    const std::vector<ck::index_t> output_spatial_lengths_;
-    const bool do_init_;
-    InputInitFun input_init_f_;
-    WeightsInitFun weights_init_f_;
-};
-
-} // namespace conv
-} // namespace utils
-} // namespace ck
-
-std::ostream& operator<<(std::ostream& os, const ck::utils::conv::ConvParams& p);
diff --git a/library/include/ck/library/utility/convolution_host_tensor_descriptor_helper.hpp b/library/include/ck/library/utility/convolution_host_tensor_descriptor_helper.hpp
new file mode 100644
index 0000000000..6b34aa7999
--- /dev/null
+++ b/library/include/ck/library/utility/convolution_host_tensor_descriptor_helper.hpp
@@ -0,0 +1,354 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+
+#include "ck/library/utility/convolution_parameter.hpp"
+
+namespace ck {
+namespace utils {
+namespace conv {
+
+namespace detail {
+
+template <typename OldLayout>
+std::vector<std::size_t> get_layout_transpose_gnchw_to_old()
+{
+    // HACK: NHWC/KYXC/NHWK, which is treated as GNHWC/GKYXC/GNHWK by this function,
+    // is used by some legacy kernel. New kernel should use GNHWK/GKYXC/GNHWK
+    // TODO: remove this branch after removing legacy kernel
+    if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NWC> ||
+                 ck::is_same_v<OldLayout, ck::tensor_layout::convolution::KXC> ||
+                 ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NWK>)
+    {
+        return {0, 1, 3, 2};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NHWC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::KYXC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NHWK>)
+    {
+        return {0, 1, 4, 2, 3};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NDHWC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::KZYXC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NDHWK>)
+    {
+        return {0, 1, 5, 2, 3, 4};
+    }
+    // separate from legacy code above
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNCW> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GKCX> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNKW>)
+    {
+        return {0, 1, 2, 3};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNCHW> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GKCYX> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNKHW>)
+    {
+        return {0, 1, 2, 3, 4};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNCDHW> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GKCZYX> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNKDHW>)
+    {
+        return {0, 1, 2, 3, 4, 5};
+    }
+    if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNWC> ||
+                 ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GKXC> ||
+                 ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNWK>)
+    {
+        return {0, 1, 3, 2};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNHWC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GKYXC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNHWK>)
+    {
+        return {0, 1, 4, 2, 3};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNDHWC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GKZYXC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::GNDHWK>)
+    {
+        return {0, 1, 5, 2, 3, 4};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NWGC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::KXGC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NWGK>)
+    {
+        return {2, 0, 3, 1};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NHWGC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::KYXGC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NHWGK>)
+    {
+        return {3, 0, 4, 1, 2};
+    }
+    else if constexpr(ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NDHWGC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::KZYXGC> ||
+                      ck::is_same_v<OldLayout, ck::tensor_layout::convolution::NDHWGK>)
+    {
+        return {4, 0, 5, 1, 2, 3};
+    }
+    else
+    {
+        printf("%s\n", __func__);
+        throw std::runtime_error("wrong! unsupported layout");
+    }
+}
+
+} // namespace detail
+
+// make tensor descriptor for packed input tensor, and order the dimension in the order of GNCHW
+// regardless of physical layout
+template <typename InLayout>
+HostTensorDescriptor
+make_input_host_tensor_descriptor_g_n_c_wis_packed(const ck::utils::conv::ConvParam& param)
+{
+    std::vector<std::size_t> physical_lengths;
+
+    // HACK: NHWC/KYXC/NHWK, which is treated as GNHWC/GKYXC/GNHWK by this function,
+    // is used by some legacy kernel. New kernel should use GNHWK/GKYXC/GNHWK
+    // TODO: remove this branch after removing legacy kernel
+    if constexpr(ck::is_same_v<InLayout, ck::tensor_layout::convolution::NWC> ||
+                 ck::is_same_v<InLayout, ck::tensor_layout::convolution::NHWC> ||
+                 ck::is_same_v<InLayout, ck::tensor_layout::convolution::NDHWC>)
+    {
+        if(param.G_ != 1)
+        {
+            throw std::runtime_error("wrong! G != 1");
+        }
+
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 2,
+                                param.input_spatial_lengths_.begin(),
+                                param.input_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    // separate from legacy code above
+    else if constexpr(ck::is_same_v<InLayout, ck::tensor_layout::convolution::GNCW> ||
+                      ck::is_same_v<InLayout, ck::tensor_layout::convolution::GNCHW> ||
+                      ck::is_same_v<InLayout, ck::tensor_layout::convolution::GNCDHW>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.end(),
+                                param.input_spatial_lengths_.begin(),
+                                param.input_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<InLayout, ck::tensor_layout::convolution::GNWC> ||
+                      ck::is_same_v<InLayout, ck::tensor_layout::convolution::GNHWC> ||
+                      ck::is_same_v<InLayout, ck::tensor_layout::convolution::GNDHWC>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 2,
+                                param.input_spatial_lengths_.begin(),
+                                param.input_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<InLayout, ck::tensor_layout::convolution::NWGC> ||
+                      ck::is_same_v<InLayout, ck::tensor_layout::convolution::NHWGC> ||
+                      ck::is_same_v<InLayout, ck::tensor_layout::convolution::NDHWGC>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 1,
+                                param.input_spatial_lengths_.begin(),
+                                param.input_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else
+    {
+        printf("%s\n", __func__);
+        printf("%s\n", InLayout::name);
+        throw std::runtime_error("wrong! unsupported layout");
+    }
+
+    return transpose_host_tensor_descriptor_given_new2old(
+        HostTensorDescriptor(physical_lengths),
+        detail::get_layout_transpose_gnchw_to_old<InLayout>());
+}
+
+// make tensor descriptor for packed weight tensor, and order the dimension in the order of GKCYX
+// regardless of physical layout
+template <typename WeiLayout>
+HostTensorDescriptor
+make_weight_host_tensor_descriptor_g_k_c_xs_packed(const ck::utils::conv::ConvParam& param)
+{
+    std::vector<std::size_t> physical_lengths;
+
+    // HACK: NHWC/KYXC/NHWK, which is treated as GNHWC/GKYXC/GNHWK by this function,
+    // is used by some legacy kernel. New kernel should use GNHWK/GKYXC/GNHWK
+    // TODO: remove this branch after removing legacy kernel
+    if constexpr(ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KXC> ||
+                 ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KYXC> ||
+                 ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KZYXC>)
+    {
+        if(param.G_ != 1)
+        {
+            throw std::runtime_error("wrong! G != 1");
+        }
+
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.K_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 2,
+                                param.filter_spatial_lengths_.begin(),
+                                param.filter_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    // separate from legacy code above
+    else if constexpr(ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KXC> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KYXC> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KZYXC>)
+    {
+        if(param.G_ != 1)
+        {
+            throw std::runtime_error("wrong! G != 1");
+        }
+
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.K_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.end(),
+                                param.filter_spatial_lengths_.begin(),
+                                param.filter_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::GKCX> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::GKCYX> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::GKCZYX>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.K_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.end(),
+                                param.filter_spatial_lengths_.begin(),
+                                param.filter_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::GKXC> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::GKYXC> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::GKZYXC>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.K_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 2,
+                                param.filter_spatial_lengths_.begin(),
+                                param.filter_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KXGC> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KYXGC> ||
+                      ck::is_same_v<WeiLayout, ck::tensor_layout::convolution::KZYXGC>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.K_),
+                                                    static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.C_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 1,
+                                param.filter_spatial_lengths_.begin(),
+                                param.filter_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else
+    {
+        printf("%s\n", __func__);
+        printf("%s\n", WeiLayout::name);
+        throw std::runtime_error("wrong! unsupported layout");
+    }
+
+    return transpose_host_tensor_descriptor_given_new2old(
+        HostTensorDescriptor(physical_lengths),
+        detail::get_layout_transpose_gnchw_to_old<WeiLayout>());
+}
+
+// make tensor descriptor for packed output tensor, and order the dimension in the order of GNKHW
+// regardless of physical layout
+template <typename OutLayout>
+HostTensorDescriptor
+make_output_host_tensor_descriptor_g_n_k_wos_packed(const ck::utils::conv::ConvParam& param)
+{
+    std::vector<std::size_t> physical_lengths;
+
+    // HACK: NHWC/KYXC/NHWK, which is treated as GNHWC/GKYXC/GNHWK by this function,
+    // is used by some legacy kernel. New kernel should use GNHWK/GKYXC/GNHWK
+    // TODO: remove this branch after removing legacy kernel
+    if constexpr(ck::is_same_v<OutLayout, ck::tensor_layout::convolution::NWK> ||
+                 ck::is_same_v<OutLayout, ck::tensor_layout::convolution::NHWK> ||
+                 ck::is_same_v<OutLayout, ck::tensor_layout::convolution::NDHWK>)
+    {
+        if(param.G_ != 1)
+        {
+            throw std::runtime_error("wrong! G != 1");
+        }
+
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.K_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 2,
+                                param.output_spatial_lengths_.begin(),
+                                param.output_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    // separate from legacy code above
+    else if constexpr(ck::is_same_v<OutLayout, ck::tensor_layout::convolution::GNKW> ||
+                      ck::is_same_v<OutLayout, ck::tensor_layout::convolution::GNKHW> ||
+                      ck::is_same_v<OutLayout, ck::tensor_layout::convolution::GNKDHW>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.K_)};
+
+        physical_lengths.insert(physical_lengths.end(),
+                                param.output_spatial_lengths_.begin(),
+                                param.output_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<OutLayout, ck::tensor_layout::convolution::GNWK> ||
+                      ck::is_same_v<OutLayout, ck::tensor_layout::convolution::GNHWK> ||
+                      ck::is_same_v<OutLayout, ck::tensor_layout::convolution::GNDHWK>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.K_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 2,
+                                param.output_spatial_lengths_.begin(),
+                                param.output_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else if constexpr(ck::is_same_v<OutLayout, ck::tensor_layout::convolution::NWGK> ||
+                      ck::is_same_v<OutLayout, ck::tensor_layout::convolution::NHWGK> ||
+                      ck::is_same_v<OutLayout, ck::tensor_layout::convolution::NDHWGK>)
+    {
+        physical_lengths = std::vector<std::size_t>{static_cast<std::size_t>(param.N_),
+                                                    static_cast<std::size_t>(param.G_),
+                                                    static_cast<std::size_t>(param.K_)};
+
+        physical_lengths.insert(physical_lengths.begin() + 1,
+                                param.output_spatial_lengths_.begin(),
+                                param.output_spatial_lengths_.begin() + param.num_dim_spatial_);
+    }
+    else
+    {
+        printf("%s\n", __func__);
+        printf("%s\n", OutLayout::name);
+        throw std::runtime_error("wrong! unsupported layout");
+    }
+
+    return transpose_host_tensor_descriptor_given_new2old(
+        HostTensorDescriptor(physical_lengths),
+        detail::get_layout_transpose_gnchw_to_old<OutLayout>());
+}
+
+} // namespace conv
+} // namespace utils
+} // namespace ck
diff --git a/library/include/ck/library/utility/convolution_parameter.hpp b/library/include/ck/library/utility/convolution_parameter.hpp
new file mode 100644
index 0000000000..5f37e03e15
--- /dev/null
+++ b/library/include/ck/library/utility/convolution_parameter.hpp
@@ -0,0 +1,86 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+#include <numeric>
+#include <iterator>
+#include <vector>
+
+#include "ck/ck.hpp"
+
+namespace ck {
+namespace utils {
+namespace conv {
+
+struct ConvParam
+{
+    ConvParam();
+    ConvParam(ck::index_t n_dim,
+              ck::index_t group_count,
+              ck::index_t n_batch,
+              ck::index_t n_out_channels,
+              ck::index_t n_in_channels,
+              const std::vector<ck::index_t>& filters_len,
+              const std::vector<ck::index_t>& input_len,
+              const std::vector<ck::index_t>& strides,
+              const std::vector<ck::index_t>& dilations,
+              const std::vector<ck::index_t>& left_pads,
+              const std::vector<ck::index_t>& right_pads);
+
+    ck::index_t num_dim_spatial_;
+    ck::index_t G_;
+    ck::index_t N_;
+    ck::index_t K_;
+    ck::index_t C_;
+
+    std::vector<ck::index_t> filter_spatial_lengths_;
+    std::vector<ck::index_t> input_spatial_lengths_;
+    std::vector<ck::index_t> output_spatial_lengths_;
+
+    std::vector<ck::index_t> conv_filter_strides_;
+    std::vector<ck::index_t> conv_filter_dilations_;
+
+    std::vector<ck::index_t> input_left_pads_;
+    std::vector<ck::index_t> input_right_pads_;
+
+    std::vector<ck::index_t> GetOutputSpatialLengths() const;
+
+    std::size_t GetFlops() const;
+
+    template <typename InDataType, typename WeiDataType, typename OutDataType>
+    std::size_t GetByte() const
+    {
+        // sizeof(InDataType) * (G * N * C * <input spatial lengths product>) +
+        // sizeof(WeiDataType) * (G * K * C * <filter spatial lengths product>) +
+        // sizeof(OutDataType) * (G * N * K * <output spatial lengths product>);
+        return sizeof(InDataType) *
+                   (G_ * N_ * C_ *
+                    std::accumulate(std::begin(input_spatial_lengths_),
+                                    std::begin(input_spatial_lengths_) + num_dim_spatial_,
+                                    static_cast<std::size_t>(1),
+                                    std::multiplies<std::size_t>())) +
+               sizeof(WeiDataType) *
+                   (G_ * K_ * C_ *
+                    std::accumulate(std::begin(filter_spatial_lengths_),
+                                    std::begin(filter_spatial_lengths_) + num_dim_spatial_,
+                                    static_cast<std::size_t>(1),
+                                    std::multiplies<std::size_t>())) +
+               sizeof(OutDataType) * (G_ * N_ * K_ *
+                                      std::accumulate(std::begin(output_spatial_lengths_),
+                                                      std::end(output_spatial_lengths_),
+                                                      static_cast<std::size_t>(1),
+                                                      std::multiplies<std::size_t>()));
+    }
+};
+
+std::string get_conv_param_parser_helper_msg();
+
+ConvParam parse_conv_param(int num_dim_spatial, int arg_idx, char* const argv[]);
+
+} // namespace conv
+} // namespace utils
+} // namespace ck
+
+std::ostream& operator<<(std::ostream& os, const ck::utils::conv::ConvParam& p);
diff --git a/library/include/ck/library/host_tensor/device_memory.hpp b/library/include/ck/library/utility/device_memory.hpp
similarity index 100%
rename from library/include/ck/library/host_tensor/device_memory.hpp
rename to library/include/ck/library/utility/device_memory.hpp
diff --git a/library/include/ck/library/host_tensor/host_common_util.hpp b/library/include/ck/library/utility/host_common_util.hpp
similarity index 100%
rename from library/include/ck/library/host_tensor/host_common_util.hpp
rename to library/include/ck/library/utility/host_common_util.hpp
diff --git a/library/include/ck/library/host_tensor/host_conv.hpp b/library/include/ck/library/utility/host_conv.hpp
similarity index 100%
rename from library/include/ck/library/host_tensor/host_conv.hpp
rename to library/include/ck/library/utility/host_conv.hpp
diff --git a/library/include/ck/library/host_tensor/host_gemm.hpp b/library/include/ck/library/utility/host_gemm.hpp
similarity index 100%
rename from library/include/ck/library/host_tensor/host_gemm.hpp
rename to library/include/ck/library/utility/host_gemm.hpp
diff --git a/library/include/ck/library/host_tensor/host_reduction.hpp b/library/include/ck/library/utility/host_reduction.hpp
similarity index 99%
rename from library/include/ck/library/host_tensor/host_reduction.hpp
rename to library/include/ck/library/utility/host_reduction.hpp
index 57cf55edad..f02ebcd79a 100644
--- a/library/include/ck/library/host_tensor/host_reduction.hpp
+++ b/library/include/ck/library/utility/host_reduction.hpp
@@ -11,8 +11,8 @@
 #include "ck/utility/reduction_enums.hpp"
 #include "ck/utility/reduction_common.hpp"
 #include "ck/utility/reduction_functions_accumulate.hpp"
-#include "ck/library/host_tensor/host_common_util.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_common_util.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 template <int NDim>
 static void get_all_indexes(const std::array<size_t, NDim>& dimLengths,
diff --git a/library/include/ck/library/host_tensor/host_tensor.hpp b/library/include/ck/library/utility/host_tensor.hpp
similarity index 79%
rename from library/include/ck/library/host_tensor/host_tensor.hpp
rename to library/include/ck/library/utility/host_tensor.hpp
index caa18e6dd1..23596d553c 100644
--- a/library/include/ck/library/host_tensor/host_tensor.hpp
+++ b/library/include/ck/library/utility/host_tensor.hpp
@@ -73,22 +73,41 @@ auto construct_f_unpack_args(F, T args)
 
 struct HostTensorDescriptor
 {
-    HostTensorDescriptor() = delete;
-
-    template <typename X>
-    HostTensorDescriptor(const std::vector<X>& lens);
-
-    template <typename X, typename Y>
-    HostTensorDescriptor(const std::vector<X>& lens, const std::vector<Y>& strides);
+    HostTensorDescriptor() = default;
 
     void CalculateStrides();
 
+    template <typename X>
+    HostTensorDescriptor(const std::initializer_list<X>& lens) : mLens(lens.begin(), lens.end())
+    {
+        this->CalculateStrides();
+    }
+
+    template <typename X>
+    HostTensorDescriptor(const std::vector<X>& lens) : mLens(lens.begin(), lens.end())
+    {
+        this->CalculateStrides();
+    }
+
     template <typename Range>
     HostTensorDescriptor(const Range& lens) : mLens(lens.begin(), lens.end())
     {
         this->CalculateStrides();
     }
 
+    template <typename X, typename Y>
+    HostTensorDescriptor(const std::initializer_list<X>& lens,
+                         const std::initializer_list<Y>& strides)
+        : mLens(lens.begin(), lens.end()), mStrides(strides.begin(), strides.end())
+    {
+    }
+
+    template <typename X, typename Y>
+    HostTensorDescriptor(const std::vector<X>& lens, const std::vector<Y>& strides)
+        : mLens(lens.begin(), lens.end()), mStrides(strides.begin(), strides.end())
+    {
+    }
+
     template <typename Range1, typename Range2>
     HostTensorDescriptor(const Range1& lens, const Range2& strides)
         : mLens(lens.begin(), lens.end()), mStrides(strides.begin(), strides.end())
@@ -97,7 +116,7 @@ struct HostTensorDescriptor
 
     std::size_t GetNumOfDimension() const;
     std::size_t GetElementSize() const;
-    std::size_t GetElementSpace() const;
+    std::size_t GetElementSpaceSize() const;
 
     const std::vector<std::size_t>& GetLengths() const;
     const std::vector<std::size_t>& GetStrides() const;
@@ -122,6 +141,22 @@ struct HostTensorDescriptor
     std::vector<std::size_t> mStrides;
 };
 
+template <typename New2Old>
+HostTensorDescriptor transpose_host_tensor_descriptor_given_new2old(const HostTensorDescriptor& a,
+                                                                    const New2Old& new2old)
+{
+    std::vector<std::size_t> new_lengths(a.GetNumOfDimension());
+    std::vector<std::size_t> new_strides(a.GetNumOfDimension());
+
+    for(std::size_t i = 0; i < a.GetNumOfDimension(); i++)
+    {
+        new_lengths[i] = a.GetLengths()[new2old[i]];
+        new_strides[i] = a.GetStrides()[new2old[i]];
+    }
+
+    return HostTensorDescriptor(new_lengths, new_strides);
+}
+
 struct joinable_thread : std::thread
 {
     template <typename... Xs>
@@ -203,22 +238,22 @@ template <typename T>
 struct Tensor
 {
     template <typename X>
-    Tensor(std::initializer_list<X> lens) : mDesc(lens), mData(mDesc.GetElementSpace())
+    Tensor(std::initializer_list<X> lens) : mDesc(lens), mData(mDesc.GetElementSpaceSize())
     {
     }
 
     template <typename X>
-    Tensor(std::vector<X> lens) : mDesc(lens), mData(mDesc.GetElementSpace())
+    Tensor(std::vector<X> lens) : mDesc(lens), mData(mDesc.GetElementSpaceSize())
     {
     }
 
     template <typename X, typename Y>
     Tensor(std::vector<X> lens, std::vector<Y> strides)
-        : mDesc(lens, strides), mData(mDesc.GetElementSpace())
+        : mDesc(lens, strides), mData(mDesc.GetElementSpaceSize())
     {
     }
 
-    Tensor(const HostTensorDescriptor& desc) : mDesc(desc), mData(mDesc.GetElementSpace()) {}
+    Tensor(const HostTensorDescriptor& desc) : mDesc(desc), mData(mDesc.GetElementSpaceSize()) {}
 
     template <typename OutT>
     Tensor<OutT> CopyAsType()
@@ -240,6 +275,24 @@ struct Tensor
         return *this;
     }
 
+    const std::vector<std::size_t>& GetLengths() const { return mDesc.GetLengths(); }
+
+    const std::vector<std::size_t>& GetStrides() const { return mDesc.GetStrides(); }
+
+    std::size_t GetNumOfDimension() const { return mDesc.GetNumOfDimension(); }
+
+    std::size_t GetElementSize() const { return mDesc.GetElementSize(); }
+
+    std::size_t GetElementSpaceSize() const { return mDesc.GetElementSpaceSize(); }
+
+    void SetZero()
+    {
+        for(auto& v : mData)
+        {
+            v = T{0};
+        }
+    }
+
     template <typename F>
     void ForEach_impl(F&& f, std::vector<size_t>& idx, size_t rank)
     {
@@ -330,6 +383,19 @@ struct Tensor
                                        mDesc.GetLengths()[4])(num_thread);
             break;
         }
+        case 6: {
+            auto f = [&](auto i0, auto i1, auto i2, auto i3, auto i4, auto i5) {
+                (*this)(i0, i1, i2, i3, i4) = g(i0, i1, i2, i3, i4, i5);
+            };
+            make_ParallelTensorFunctor(f,
+                                       mDesc.GetLengths()[0],
+                                       mDesc.GetLengths()[1],
+                                       mDesc.GetLengths()[2],
+                                       mDesc.GetLengths()[3],
+                                       mDesc.GetLengths()[4],
+                                       mDesc.GetLengths()[5])(num_thread);
+            break;
+        }
         default: throw std::runtime_error("unspported dimension");
         }
     }
@@ -367,17 +433,3 @@ struct Tensor
     HostTensorDescriptor mDesc;
     std::vector<T> mData;
 };
-
-template <typename X>
-HostTensorDescriptor::HostTensorDescriptor(const std::vector<X>& lens)
-    : mLens(lens.begin(), lens.end())
-{
-    this->CalculateStrides();
-}
-
-template <typename X, typename Y>
-HostTensorDescriptor::HostTensorDescriptor(const std::vector<X>& lens,
-                                           const std::vector<Y>& strides)
-    : mLens(lens.begin(), lens.end()), mStrides(strides.begin(), strides.end())
-{
-}
diff --git a/library/include/ck/library/host_tensor/host_tensor_generator.hpp b/library/include/ck/library/utility/host_tensor_generator.hpp
similarity index 100%
rename from library/include/ck/library/host_tensor/host_tensor_generator.hpp
rename to library/include/ck/library/utility/host_tensor_generator.hpp
diff --git a/library/include/ck/library/utility/op_instance_engine.hpp b/library/include/ck/library/utility/op_instance_engine.hpp
index 8ba63f36e2..78812e8c81 100644
--- a/library/include/ck/library/utility/op_instance_engine.hpp
+++ b/library/include/ck/library/utility/op_instance_engine.hpp
@@ -16,8 +16,8 @@
 #include "ck/tensor_operation/gpu/device/device_base.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 namespace ck {
 namespace utils {
@@ -103,8 +103,8 @@ class OpInstanceRunEngine
             }
         }
         AllocateDeviceInputTensors(std::make_index_sequence<kNInArgs_>{});
-        out_device_buffer_ =
-            std::make_unique<DeviceMem>(sizeof(OutDataType) * out_tensor_->mDesc.GetElementSpace());
+        out_device_buffer_ = std::make_unique<DeviceMem>(sizeof(OutDataType) *
+                                                         out_tensor_->mDesc.GetElementSpaceSize());
         out_device_buffer_->SetZero();
     }
 
@@ -222,7 +222,7 @@ class OpInstanceRunEngine
         in_device_buffers_
             .emplace_back(
                 std::make_unique<DeviceMem>(sizeof(std::tuple_element_t<Index, InArgsTypesTuple>) *
-                                            ts->mDesc.GetElementSpace()))
+                                            ts->mDesc.GetElementSpaceSize()))
             ->ToDevice(ts->mData.data());
     }
 
diff --git a/library/src/host_tensor/CMakeLists.txt b/library/src/host_tensor/CMakeLists.txt
deleted file mode 100644
index eca22c6091..0000000000
--- a/library/src/host_tensor/CMakeLists.txt
+++ /dev/null
@@ -1,32 +0,0 @@
-## host_tensor
-set(HOST_TENSOR_SOURCE
-    device_memory.cpp
-    host_tensor.cpp
-)
-
-add_library(host_tensor STATIC ${HOST_TENSOR_SOURCE})
-add_library(composable_kernel::host_tensor ALIAS host_tensor)
-
-target_compile_features(host_tensor PUBLIC)
-set_target_properties(host_tensor PROPERTIES POSITION_INDEPENDENT_CODE ON)
-target_include_directories(host_tensor SYSTEM PUBLIC $<BUILD_INTERFACE:${HALF_INCLUDE_DIR}>)
-
-target_include_directories(host_tensor PUBLIC
-    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck>"
-    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/utility>"
-    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/host_tensor>"
-)
-
-rocm_install(
-    TARGETS host_tensor
-    EXPORT host_tensorTargets
-)
-
-rocm_install(
-    EXPORT host_tensorTargets
-    FILE composable_kernelhost_tensorTargets.cmake
-    NAMESPACE composable_kernel::
-    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
-)
-
-clang_tidy_check(host_tensor)
diff --git a/library/src/tensor_operation_instance/gpu/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
index f1ce23aae2..0a50d37c8a 100644
--- a/library/src/tensor_operation_instance/gpu/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
@@ -16,15 +16,18 @@ add_subdirectory(batched_gemm_reduce)
 add_subdirectory(grouped_gemm)
 add_subdirectory(contraction_scale)
 add_subdirectory(contraction_bilinear)
-add_subdirectory(conv1d_fwd)
+add_subdirectory(grouped_conv1d_fwd)
+add_subdirectory(grouped_conv2d_fwd)
+add_subdirectory(grouped_conv3d_fwd)
 add_subdirectory(conv2d_fwd)
-add_subdirectory(conv3d_fwd)
+add_subdirectory(conv1d_bwd_data)
+add_subdirectory(conv2d_bwd_data)
+add_subdirectory(conv3d_bwd_data)
+add_subdirectory(conv1d_bwd_weight)
+add_subdirectory(conv2d_bwd_weight)
+add_subdirectory(conv3d_bwd_weight)
 add_subdirectory(conv2d_fwd_bias_relu)
 add_subdirectory(conv2d_fwd_bias_relu_add)
-add_subdirectory(conv2d_bwd_data)
-add_subdirectory(convnd_bwd_data)
-add_subdirectory(conv2d_bwd_weight)
-add_subdirectory(convnd_bwd_weight)
 add_subdirectory(reduce)
 add_subdirectory(normalization)
 add_subdirectory(elementwise)
@@ -40,15 +43,17 @@ add_library(device_operations STATIC
     $<TARGET_OBJECTS:device_grouped_gemm_instance>
     $<TARGET_OBJECTS:device_contraction_scale_instance>
     $<TARGET_OBJECTS:device_contraction_bilinear_instance>
-    $<TARGET_OBJECTS:device_conv1d_fwd_instance>
-    $<TARGET_OBJECTS:device_conv2d_fwd_instance>
-    $<TARGET_OBJECTS:device_conv3d_fwd_instance>
+    $<TARGET_OBJECTS:device_grouped_conv1d_fwd_instance>
+    $<TARGET_OBJECTS:device_grouped_conv2d_fwd_instance>
+    $<TARGET_OBJECTS:device_grouped_conv3d_fwd_instance>
+    $<TARGET_OBJECTS:device_conv1d_bwd_data_instance>
+    $<TARGET_OBJECTS:device_conv2d_bwd_data_instance>
+    $<TARGET_OBJECTS:device_conv3d_bwd_data_instance>
+    $<TARGET_OBJECTS:device_conv1d_bwd_weight_instance>
+    $<TARGET_OBJECTS:device_conv2d_bwd_weight_instance>
+    $<TARGET_OBJECTS:device_conv3d_bwd_weight_instance>
     $<TARGET_OBJECTS:device_conv2d_fwd_bias_relu_instance>
     $<TARGET_OBJECTS:device_conv2d_fwd_bias_relu_add_instance>
-    $<TARGET_OBJECTS:device_conv2d_bwd_data_instance>
-    $<TARGET_OBJECTS:device_convnd_bwd_data_instance>
-    $<TARGET_OBJECTS:device_conv2d_bwd_weight_instance>
-    $<TARGET_OBJECTS:device_convnd_bwd_weight_instance>
     $<TARGET_OBJECTS:device_reduce_instance>
     $<TARGET_OBJECTS:device_normalization_instance>
     $<TARGET_OBJECTS:device_elementwise_instance>
@@ -75,7 +80,7 @@ target_include_directories(device_operations PUBLIC
     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/warp>
     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/thread>
     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/tensor_operation/gpu/element>
-    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/host_tensor>
+    $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/utility>
     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/tensor_operation_instance>
     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/tensor_operation_instance/gpu>
     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/tensor_operation_instance/gpu/reduce>
diff --git a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_instance.cpp
index 036818ee2c..230965867f 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32       = float;
-using F32_TUPLE = ck::Tuple<F32>;
+using F32_Tuple = ck::Tuple<F32>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,19 +40,19 @@ using device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn_in
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type|  Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |          |      |    Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |          |      |             |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>
     // clang-format on
     >;
 
@@ -62,7 +62,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_kknn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_instance.cpp
index b277fb86e8..f759f1cd6d 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32       = float;
-using F32_TUPLE = ck::Tuple<F32>;
+using F32_Tuple = ck::Tuple<F32>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,22 +40,22 @@ using device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn_in
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type|  Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |          |      |    Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |          |      |             |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   1,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   1,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   1,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   1,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   1,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   1,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
     // clang-format on
     >;
 
@@ -65,7 +65,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_knnn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_instance.cpp
index c03ce0b169..b1715740e6 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32       = float;
-using F32_TUPLE = ck::Tuple<F32>;
+using F32_Tuple = ck::Tuple<F32>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,22 +40,22 @@ using device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn_in
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type|  Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|          _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |          |      |    Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|          _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |          |      |             |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                              |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   4,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   4,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   4,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   4,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   4,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   4,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>
     // clang-format on
     >;
 
@@ -65,7 +65,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mknn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_instance.cpp
index ab56c4c159..378d1147b5 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_bilinear/device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32       = float;
-using F32_TUPLE = ck::Tuple<F32>;
+using F32_Tuple = ck::Tuple<F32>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,22 +40,22 @@ using device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn_in
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type|  Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |          |      |    Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |          |      |             |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   1,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   1,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   1,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_TUPLE,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   1,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   1,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   1,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, F32_Tuple,   F32,  PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
     // clang-format on
     >;
 
@@ -65,7 +65,7 @@ void add_device_contraction_bilinear_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_f32_mnnn
                                                            2,
                                                            F32,
                                                            F32,
-                                                           F32_TUPLE,
+                                                           F32_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp
index 7f49a98642..2c4141db26 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32         = float;
-using EMPTY_TUPLE = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,19 +40,19 @@ using device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instance =
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|        Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |            |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,               4>
     // clang-format on
     >;
 
@@ -62,7 +62,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_kkn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance.cpp
index 45ffa63ce2..972b2172cd 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32         = float;
-using EMPTY_TUPLE = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,22 +40,22 @@ using device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instance =
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|        Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |            |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   1,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   1,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   1,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   1,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   1,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   1,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
     // clang-format on
     >;
 
@@ -65,7 +65,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_knn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp
index cc63b06a56..c2fa6be202 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32         = float;
-using EMPTY_TUPLE = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,22 +40,22 @@ using device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instance =
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|        Type|  Type|  Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|          _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |            |      |    Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|          _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |            |      |             |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                              |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   4,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   4,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   4,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   4,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   4,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   4,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   4,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,               S<1, 16, 1, 16>,               4>
     // clang-format on
     >;
 
@@ -65,7 +65,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mkn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp
index ce11f255a6..d701a01a2d 100644
--- a/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/contraction_scale/device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance.cpp
@@ -22,7 +22,7 @@ namespace device {
 namespace instance {
 
 using F32         = float;
-using EMPTY_TUPLE = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
@@ -40,22 +40,22 @@ using device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instance =
         //#####################################|        |        |        |  Type|  Type|    Type| DataType|        Type|  Type|  Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
         //#####################################|        |        |        |      |      |        |         |            |      |    Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
         //#####################################|        |        |        |      |      |        |         |            |      |             |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   1,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   1,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   1,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
-        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, EMPTY_TUPLE,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   1,   1,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   1,   1,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   1,   1,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1,  8>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   1,   1,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   1,   1,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   1,   1,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              1,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              1,         0,           1,           1,              S<1, 16, 1, 16>,               4>,
+        DeviceContractionMultipleD_Xdl_CShuffle<       2,       2,       2,   F32,   F32,     F32,      F32, Empty_Tuple,   F32,  PassThrough, PassThrough,       Scale, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              4,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              4,         1,           1,           1,              S<1, 16, 1, 16>,               4>
     // clang-format on
     >;
 
@@ -65,7 +65,7 @@ void add_device_contraction_scale_m2_n2_k2_xdl_c_shuffle_f32_f32_f32_mnn_instanc
                                                            2,
                                                            F32,
                                                            F32,
-                                                           EMPTY_TUPLE,
+                                                           Empty_Tuple,
                                                            F32,
                                                            PassThrough,
                                                            PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/CMakeLists.txt
new file mode 100644
index 0000000000..fc72bed39f
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/CMakeLists.txt
@@ -0,0 +1,14 @@
+# device_conv1d_bwd_data_instance
+set(DEVICE_CONV1D_BWD_DATA_INSTANCE_SOURCE
+   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp;
+   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp;
+   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp;
+   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp;
+)
+
+add_library(device_conv1d_bwd_data_instance OBJECT ${DEVICE_CONV1D_BWD_DATA_INSTANCE_SOURCE})
+target_compile_features(device_conv1d_bwd_data_instance PUBLIC)
+set_target_properties(device_conv1d_bwd_data_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+rocm_install(TARGETS device_conv1d_bwd_data_instance)
+
+clang_tidy_check(device_conv1d_bwd_data_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp
new file mode 100644
index 0000000000..0666fba647
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<1,
+                                                  NWC,
+                                                  KXC,
+                                                  NWK,
+                                                  BF16,
+                                                  BF16,
+                                                  BF16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp
new file mode 100644
index 0000000000..5f33746fcb
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp
@@ -0,0 +1,95 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,   8,  32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,   8,  32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,   8,  32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,   8,  32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,   8,  32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,   8,  32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,   8,  32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,       
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,   8,  32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,   8,  32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,   8,  32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,   8,  32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvBwdData<1, NWC, KXC, NWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp
new file mode 100644
index 0000000000..3812d396a4
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp
@@ -0,0 +1,94 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                    |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                 |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances(
+    std::vector<std::unique_ptr<
+        DeviceConvBwdData<1, NWC, KXC, NWK, F32, F32, F32, PassThrough, PassThrough, PassThrough>>>&
+        instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp
new file mode 100644
index 0000000000..4f2a1129ca
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp
@@ -0,0 +1,99 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_int8_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       1, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<1,
+                                                  NWC,
+                                                  KXC,
+                                                  NWK,
+                                                  int8_t,
+                                                  int8_t,
+                                                  int8_t,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_int8_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/CMakeLists.txt
new file mode 100644
index 0000000000..5b80510899
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/CMakeLists.txt
@@ -0,0 +1,13 @@
+#device_conv1d_bwd_weight_instance
+set(DEVICE_CONV1D_BWD_WEIGHT_INSTANCE_SOURCE 
+    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp;
+    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp;
+    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp;
+)
+
+add_library(device_conv1d_bwd_weight_instance OBJECT ${DEVICE_CONV1D_BWD_WEIGHT_INSTANCE_SOURCE})
+target_compile_features(device_conv1d_bwd_weight_instance PUBLIC) 
+set_target_properties(device_conv1d_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+rocm_install(TARGETS device_conv1d_bwd_weight_instance)
+
+clang_tidy_check(device_conv1d_bwd_weight_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp
new file mode 100644
index 0000000000..98b62fc171
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_f32_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_f32_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp
new file mode 100644
index 0000000000..d43e954dfc
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp
new file mode 100644
index 0000000000..98c2653e72
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp
@@ -0,0 +1,101 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_fwd/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv1d_fwd/CMakeLists.txt
deleted file mode 100644
index 77aa6198f5..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv1d_fwd/CMakeLists.txt
+++ /dev/null
@@ -1,14 +0,0 @@
-# device_conv1d_fwd_instance
-set(DEVICE_CONV1D_FWD_INSTANCE_SOURCE
-   device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instance.cpp;
-   device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instance.cpp;
-   device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp;
-   device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instance.cpp;
-)
-
-add_library(device_conv1d_fwd_instance OBJECT ${DEVICE_CONV1D_FWD_INSTANCE_SOURCE}) 
-# target_compile_features(device_conv1d_fwd_instance PUBLIC)
-set_target_properties(device_conv1d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-# install(TARGETS device_conv1d_fwd_instance LIBRARY DESTINATION lib) 
-
-clang_tidy_check(device_conv1d_fwd_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instance.cpp
deleted file mode 100644
index d4c65ff54b..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instance.cpp
+++ /dev/null
@@ -1,115 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32  = float;
-using BF16 = bhalf_t;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances, device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_bf16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instance.cpp
deleted file mode 100644
index 166d25ba48..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instance.cpp
+++ /dev/null
@@ -1,115 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances, device_conv1d_fwd_xdl_nwc_kxc_nwk_f16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_f16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp
deleted file mode 100644
index 2cb296e472..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instance.cpp
+++ /dev/null
@@ -1,118 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-//------------------------------------------------------------------------------
-//            Conv1D
-//------------------------------------------------------------------------------
-
-// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-    // clang-format on
-    >;
-
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances, device_conv1d_fwd_xdl_nwc_kxc_nwk_f32_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_f32_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instance.cpp
deleted file mode 100644
index 2364c5ea32..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv1d_fwd/device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instance.cpp
+++ /dev/null
@@ -1,117 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |       |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      1,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |       |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      1,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |       |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      1,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances, device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_p0_int8_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv1d_fwd_xdl_nwc_kxc_nwk_1x1_s1_p0_int8_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
index 3b716d641c..add622e0c9 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
@@ -5,8 +5,11 @@
 
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+// FIXME: retire dedicated 2D version
 #include "ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -14,13 +17,18 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-using BF16 = ck::bhalf_t;
+using BF16 = bhalf_t;
 using F32  = float;
 
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
 static constexpr auto ConvBwdDataDefault =
     ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
 
@@ -29,6 +37,52 @@ static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
 
 // Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
 using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+// FIXME: retire dedicated 2D version
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances = std::tuple<
     // clang-format off
         //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
         //####################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
@@ -50,7 +104,8 @@ using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances = std::tuple<
     // clang-format on
     >;
 
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances =
+// FIXME: retire dedicated 2D version
+using device_conv_dedidecate_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances =
     std::tuple<
         // clang-format off
         //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
@@ -74,12 +129,26 @@ using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances =
         >;
 
 void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  BF16,
+                                                  BF16,
+                                                  BF16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances{});
     add_device_operation_instances(
         instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances{});
+    add_device_operation_instances(
+        instances,
+        device_conv_dedidecate_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances{});
 }
 
 } // namespace instance
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
index 5978ffcd10..71436dd47c 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
@@ -5,8 +5,11 @@
 
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+// FIXME: retire dedicated 2D version
 #include "ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -20,7 +23,12 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
 static constexpr auto ConvBwdDataDefault =
     ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
 
@@ -29,6 +37,52 @@ static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
 
 // Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
 using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,   8,  32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,   8,  32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,   8,  32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,   8,  32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,   8,  32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,   8,  32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,   8,  32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,       
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,   8,  32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,   8,  32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,   8,  32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,   8,  32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+// FIXME: retire dedicated 2D version
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances = std::tuple<
     // clang-format off
         //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
         //####################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
@@ -52,7 +106,8 @@ using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances = std::tuple<
     // clang-format on
     >;
 
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances =
+// FIXME: retire dedicated 2D version
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances =
     std::tuple<
         // clang-format off
         //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
@@ -76,12 +131,25 @@ using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances =
         >;
 
 void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  F16,
+                                                  F16,
+                                                  F16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances{});
     add_device_operation_instances(
         instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances{});
 }
 
 } // namespace instance
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
index 42e80be1a0..782f06da17 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
@@ -5,8 +5,11 @@
 
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+// FIXME: retire dedicated 2D version
 #include "ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -19,7 +22,12 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
 static constexpr auto ConvBwdDataDefault =
     ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
 
@@ -28,6 +36,52 @@ static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
 
 // Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
 using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                    |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                 |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
+    // clang-format on
+    >;
+
+// FIXME: retire dedicated 2D version
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances = std::tuple<
     // clang-format off
         //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
         //####################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
@@ -49,7 +103,8 @@ using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances = std::tuple<
     // clang-format on
     >;
 
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances =
+// FIXME: retire dedicated 2D version
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances =
     std::tuple<
         // clang-format off
         //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
@@ -73,12 +128,25 @@ using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances =
         >;
 
 void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  F32,
+                                                  F32,
+                                                  F32,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances{});
     add_device_operation_instances(
         instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances{});
 }
 
 } // namespace instance
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
index ff15c0238b..79a366d03a 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
@@ -5,8 +5,11 @@
 
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+// FIXME: retire dedicated 2D version
 #include "ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -14,13 +17,15 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-using DataType = int8_t;
-using AccType  = int32_t;
-
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
 static constexpr auto ConvBwdDataDefault =
     ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
 
@@ -30,56 +35,116 @@ static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
 // Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
 using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances = std::tuple<
     // clang-format off
-        //####################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //####################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //####################################################################|          |          |          |           |   Operation|   Operation|   Operation|      Specialization|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //####################################################################|          |          |          |           |            |            |            |                    |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
     // clang-format on
     >;
 
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances =
+using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       2, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+    // clang-format on
+    >;
+
+// FIXME: retire dedicated 2D version
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances = std::tuple<
+    // clang-format off
+        //####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //####################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //####################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //####################################################################|       |        |        |        |            |            |            |                    |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K< int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+    // clang-format on
+    >;
+
+// FIXME: retire dedicated 2D version
+using device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances =
     std::tuple<
         // clang-format off
-        //#####################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#####################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                             Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#####################################################################|          |          |          |           |   Operation|   Operation|   Operation|                   Specialization|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#####################################################################|          |          |          |           |            |            |            |                                 |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+        //#####################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //#####################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //#####################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //#####################################################################|       |        |        |        |            |            |            |                                 |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
         // clang-format on
         >;
 
 void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvBwdData<2,
+                                                  NHWC,
+                                                  KYXC,
+                                                  NHWK,
+                                                  int8_t,
+                                                  int8_t,
+                                                  int8_t,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances{});
     add_device_operation_instances(
         instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances{});
 }
 
 } // namespace instance
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/CMakeLists.txt
index 7d3c57b235..be60dc2aab 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/CMakeLists.txt
@@ -1,10 +1,12 @@
-# device_conv2d_bwd_weight_instance
-set(DEVICE_CONV2D_BWD_WEIGHT_INSTANCE_SOURCE
-   device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
-   device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
+#device_conv2d_bwd_weight_instance
+set(DEVICE_CONV2D_BWD_WEIGHT_INSTANCE_SOURCE 
+    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
+    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
+    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp;
 )
+
 add_library(device_conv2d_bwd_weight_instance OBJECT ${DEVICE_CONV2D_BWD_WEIGHT_INSTANCE_SOURCE})
-target_compile_features(device_conv2d_bwd_weight_instance PUBLIC)
+target_compile_features(device_conv2d_bwd_weight_instance PUBLIC) 
 set_target_properties(device_conv2d_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
 rocm_install(TARGETS device_conv2d_bwd_weight_instance)
 
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
new file mode 100644
index 0000000000..792cc33ae3
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_f32_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_f32_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
index ea9fb8c6a8..58b1e4a462 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
@@ -5,8 +5,11 @@
 
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+// TODO: retire dedicated 2d version
 #include "ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -20,36 +23,105 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
 // Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances = std::tuple<
+using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_default_instances = std::tuple<
     // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,     64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,    128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,     64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32,  PassThrough, PassThrough, PassThrough,     64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+// TODO: retire dedicated 2d version
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv_dedicated_2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances = std::tuple<
+    // clang-format off
+        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#################################################################################|       |        |        |        |            |            |            |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,     64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,    128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,     64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,     64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
     // clang-format on
     >;
 
 void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
 {
-    add_device_operation_instances(instances,
-                                   device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances{});
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances{});
 }
 
 } // namespace instance
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
index 744f2f91e8..b90044e74f 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
@@ -5,8 +5,11 @@
 
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+// TODO: retire dedicated 2d version
 #include "ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -19,36 +22,104 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
 // Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances = std::tuple<
+using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f32_default_instances = std::tuple<
     // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,     64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,    128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,     64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32,  PassThrough, PassThrough, PassThrough,     64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv_dedicated_2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances = std::tuple<
+    // clang-format off
+        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#################################################################################|       |        |        |        |            |            |            |       |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,     64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,    128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,     64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<     F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,     64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
     // clang-format on
     >;
 
 void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
 {
-    add_device_operation_instances(instances,
-                                   device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances{});
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f32_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances{});
+    add_device_operation_instances(
+        instances, device_conv_dedicated_2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances{});
 }
 
 } // namespace instance
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt
index 1ef4a9b07e..8d21aa2bc3 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/conv2d_fwd/CMakeLists.txt
@@ -6,18 +6,7 @@ set(DEVICE_CONV2D_FWD_INSTANCE_SOURCE
    device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp;
    device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp;
 )
-set(DEVICE_CONVND_2D_FWD_INSTANCE_SOURCE
-   device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
-   device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
-   device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp;
-   device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp;
-)
 
 add_library(device_conv2d_fwd_instance OBJECT ${DEVICE_CONV2D_FWD_INSTANCE_SOURCE}) 
-add_library(device_convnd_2d_fwd_instance OBJECT ${DEVICE_CONVND_2D_FWD_INSTANCE_SOURCE}) 
-
 set_target_properties(device_conv2d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-set_target_properties(device_convnd_2d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-
 clang_tidy_check(device_conv2d_fwd_instance)
-clang_tidy_check(device_convnd_2d_fwd_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp
index 7766a12eb9..55496f2ce6 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instance.cpp
@@ -20,6 +20,10 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto ConvFwdDefault =
@@ -131,7 +135,9 @@ using device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_odd_c_f16_instances = std::
     >;
 
 void add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<
+        DeviceConvFwd<2, NHWC, KYXC, NHWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances{});
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
index efb4bd875f..80754f94c4 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
@@ -20,6 +20,10 @@ using F32  = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto ConvFwdDefault =
@@ -99,7 +103,16 @@ using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances = std::tuple
     >;
 
 void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvFwd<2,
+                                              NHWC,
+                                              KYXC,
+                                              NHWK,
+                                              BF16,
+                                              BF16,
+                                              BF16,
+                                              PassThrough,
+                                              PassThrough,
+                                              PassThrough>>>& instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances{});
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
index 5c0110aa51..7b769949b6 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
@@ -20,6 +20,10 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto ConvFwdDefault =
@@ -99,7 +103,9 @@ using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances = std::tuple<
     >;
 
 void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<
+        DeviceConvFwd<2, NHWC, KYXC, NHWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
+        instances)
 {
     add_device_operation_instances(instances, device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances{});
     add_device_operation_instances(instances,
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
index 3e4c8debc9..cf2d451c7e 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
@@ -19,6 +19,10 @@ using F32 = float;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto ConvFwdDefault =
@@ -98,7 +102,9 @@ using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances = std::tuple<
     >;
 
 void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<
+        DeviceConvFwd<2, NHWC, KYXC, NHWK, F32, F32, F32, PassThrough, PassThrough, PassThrough>>>&
+        instances)
 {
     add_device_operation_instances(instances, device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances{});
     add_device_operation_instances(instances,
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
index cd1bf085fb..f8ea7a2011 100644
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
@@ -14,11 +14,13 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-using F32 = float;
-
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto ConvFwdDefault =
@@ -98,7 +100,16 @@ using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances = std::tuple
     >;
 
 void add_device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
+    std::vector<std::unique_ptr<DeviceConvFwd<2,
+                                              NHWC,
+                                              KYXC,
+                                              NHWK,
+                                              int8_t,
+                                              int8_t,
+                                              int8_t,
+                                              PassThrough,
+                                              PassThrough,
+                                              PassThrough>>>& instances)
 {
     add_device_operation_instances(instances,
                                    device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances{});
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
deleted file mode 100644
index 75351654ba..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
+++ /dev/null
@@ -1,118 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = ck::bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_bf16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
deleted file mode 100644
index c274e7e49d..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
+++ /dev/null
@@ -1,117 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_f16_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances, device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_f16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
deleted file mode 100644
index 22cb766415..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
+++ /dev/null
@@ -1,116 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_f32_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |        |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |        |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances, device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_f32_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
deleted file mode 100644
index 076faf7f3b..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv2d_fwd/device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
+++ /dev/null
@@ -1,117 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |        |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |        |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,       2,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |        |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |        |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,       2,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward|  NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization| Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |        |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |        |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,       2,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_p0_int8_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv2d_fwd_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/CMakeLists.txt
new file mode 100644
index 0000000000..215d4f7e86
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/CMakeLists.txt
@@ -0,0 +1,14 @@
+# device_conv3d_bwd_data_instance
+set(DEVICE_CONV3D_BWD_DATA_INSTANCE_SOURCE
+   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
+   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
+   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
+   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp;
+)
+
+add_library(device_conv3d_bwd_data_instance OBJECT ${DEVICE_CONV3D_BWD_DATA_INSTANCE_SOURCE})
+target_compile_features(device_conv3d_bwd_data_instance PUBLIC)
+set_target_properties(device_conv3d_bwd_data_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+rocm_install(TARGETS device_conv3d_bwd_data_instance)
+
+clang_tidy_check(device_conv3d_bwd_data_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
new file mode 100644
index 0000000000..63244018bd
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  BF16,
+                                                  BF16,
+                                                  BF16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
new file mode 100644
index 0000000000..975b2906c1
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,   8,  32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,   8,  32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,   8,  32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,   8,  32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,   8,  32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,   8,  32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,   8,  32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,       
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,   8,  32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,   8,  32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,   8,  32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,   8,  32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  F16,
+                                                  F16,
+                                                  F16,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
new file mode 100644
index 0000000000..20213e096e
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
@@ -0,0 +1,101 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                    |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                 |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  F32,
+                                                  F32,
+                                                  F32,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
new file mode 100644
index 0000000000..8d34f54825
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
@@ -0,0 +1,99 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdDataDefault =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
+
+static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                   |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_int8_instances = std::tuple<
+    // clang-format off
+        //##############################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                    ConvBackward| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
+        //##############################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Data|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
+        //##############################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                  Specialization|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
+        //##############################|        |       |        |        |        |            |            |            |                                |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,   128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
+        DeviceConvNdBwdDataNwcKxcNwk_Xdl<       3, int8_t,  int8_t,  int8_t, int32_t, PassThrough, PassThrough, PassThrough, ConvBwdDataFilter1x1Stride1Pad0,    64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdData<3,
+                                                  NDHWC,
+                                                  KZYXC,
+                                                  NDHWK,
+                                                  int8_t,
+                                                  int8_t,
+                                                  int8_t,
+                                                  PassThrough,
+                                                  PassThrough,
+                                                  PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_int8_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/CMakeLists.txt
new file mode 100644
index 0000000000..dfa03ea74a
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/CMakeLists.txt
@@ -0,0 +1,13 @@
+#device_conv3d_bwd_weight_instance
+set(DEVICE_CONV3D_BWD_WEIGHT_INSTANCE_SOURCE 
+    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
+    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
+    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
+)
+
+add_library(device_conv3d_bwd_weight_instance OBJECT ${DEVICE_CONV3D_BWD_WEIGHT_INSTANCE_SOURCE})
+target_compile_features(device_conv3d_bwd_weight_instance PUBLIC) 
+set_target_properties(device_conv3d_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+rocm_install(TARGETS device_conv3d_bwd_weight_instance)
+
+clang_tidy_check(device_conv3d_bwd_weight_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
new file mode 100644
index 0000000000..ff1a080dca
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
@@ -0,0 +1,104 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances{});
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_f32_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
new file mode 100644
index 0000000000..9d10187763
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
@@ -0,0 +1,103 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f16_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f16_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
new file mode 100644
index 0000000000..633b30b8c2
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f32_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f32_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_fwd/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/conv3d_fwd/CMakeLists.txt
deleted file mode 100644
index 91a299c742..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv3d_fwd/CMakeLists.txt
+++ /dev/null
@@ -1,12 +0,0 @@
-# device_conv3d_fwd_instance
-set(DEVICE_CONV3D_FWD_INSTANCE_SOURCE
-   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
-   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
-   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
-   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp;
-)
-add_library(device_conv3d_fwd_instance OBJECT ${DEVICE_CONV3D_FWD_INSTANCE_SOURCE}) 
-target_compile_features(device_conv3d_fwd_instance PUBLIC)
-set_target_properties(device_conv3d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-
-clang_tidy_check(device_conv3d_fwd_instance)
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
deleted file mode 100644
index e55a3d2b5b..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
+++ /dev/null
@@ -1,115 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32  = float;
-using BF16 = bhalf_t;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
deleted file mode 100644
index 01c6cc6b37..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
+++ /dev/null
@@ -1,115 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>
-    // clang-format on
-    >;
-
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f16_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_f16_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
deleted file mode 100644
index f881958c91..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
+++ /dev/null
@@ -1,114 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-    // clang-format on
-    >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |        |        |        |   Operation|   Operation|   Operation|               |       |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |        |        |        |            |            |            |               |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,      true,               7,               1>
-    // clang-format on
-    >;
-
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_f32_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_f32_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
deleted file mode 100644
index d7c0a30874..0000000000
--- a/library/src/tensor_operation_instance/gpu/conv3d_fwd/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
+++ /dev/null
@@ -1,117 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-static constexpr auto ConvFwd1x1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
-
-static constexpr auto ConvFwd1x1S1P0 =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |       |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault,      3,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |       |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0,      3,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //################################################################| InData|  WeiData| OutData|  AccData|          In|         Wei|         Out|    ConvForward| NumDim| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //################################################################|   Type|     Type|    Type|     Type| Elementwise| Elementwise| Elementwise| Specialization|Spatial|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //################################################################|       |         |        |         |   Operation|   Operation|   Operation|               |       |      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //################################################################|       |         |        |         |            |            |            |               |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,   128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>,
-        DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<  int8_t,  int8_t,  int8_t,  int32_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0,      3,    64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,             16,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instances(
-    std::vector<DeviceConvFwdPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_int8_instances{});
-    add_device_operation_instances(instances,
-                                   device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_p0_int8_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_int8_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/CMakeLists.txt
deleted file mode 100644
index dae633b7da..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/CMakeLists.txt
+++ /dev/null
@@ -1,22 +0,0 @@
-# device_convnd_bwd_data_instance
-set(DEVICE_CONVND_BWD_DATA_INSTANCE_SOURCE
-   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp;
-   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp;
-   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp;
-   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp;
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp;
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp;
-   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
-   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
-   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
-   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp;
-)
-
-add_library(device_convnd_bwd_data_instance OBJECT ${DEVICE_CONVND_BWD_DATA_INSTANCE_SOURCE})
-target_compile_features(device_convnd_bwd_data_instance PUBLIC)
-set_target_properties(device_convnd_bwd_data_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-rocm_install(TARGETS device_convnd_bwd_data_instance)
-
-clang_tidy_check(device_convnd_bwd_data_instance)
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp
deleted file mode 100644
index a449a9053f..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                    |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp
deleted file mode 100644
index fb97674032..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instance.cpp
+++ /dev/null
@@ -1,91 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                   |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     256,   256,   128,     4,   8,  32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     256,   128,   256,     4,   8,  32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     128,   128,   128,     4,   8,  32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-#if 1
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     256,   128,   128,     4,   8,  32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     256,    64,   128,     4,   8,  32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     128,    32,   128,     4,   8,  32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-#endif
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     128,   128,    64,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     128,    64,   128,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,      64,    64,    64,     4,   8,  32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     256,   128,    64,     4,   8,  32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,     128,   128,    32,     4,   8,  32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,      64,    64,    32,     4,   8,  32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   1,      64,    32,    64,     4,   8,  32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f16_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp
deleted file mode 100644
index e8f2a45b71..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instance.cpp
+++ /dev/null
@@ -1,88 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                    |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_f32_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp
deleted file mode 100644
index 6aad1f029f..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instance.cpp
+++ /dev/null
@@ -1,91 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using DataType = int8_t;
-using AccType  = int32_t;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|          |          |          |           |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|          |          |          |           |            |            |            |                    |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-    #if 1
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-    #endif
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,     128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   1,      64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //##############################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //##############################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //##############################################################################|          |          |          |           |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //##############################################################################|          |          |          |           |            |            |            |                                 |       |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,     128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   1,      64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv1d_bwd_data_xdl_nwc_kxc_nwk_int8_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_data_xdl_nwc_kxc_nwk_1x1_s1_p0_int8_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
deleted file mode 100644
index 010291cb47..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                    |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
deleted file mode 100644
index e7e147177a..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                   |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     256,   256,   128,     4,   8,  32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     256,   128,   256,     4,   8,  32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     128,   128,   128,     4,   8,  32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     256,   128,   128,     4,   8,  32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     256,    64,   128,     4,   8,  32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     128,    32,   128,     4,   8,  32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     128,    64,   128,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     256,   128,    64,     4,   8,  32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     128,   128,    64,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,       
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,      64,    64,    64,     4,   8,  32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,     128,   128,    32,     4,   8,  32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,      64,    64,    32,     4,   8,  32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   2,      64,    32,    64,     4,   8,  32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
deleted file mode 100644
index 357ddabd10..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
+++ /dev/null
@@ -1,88 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                    |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
deleted file mode 100644
index 3eadb0bdc9..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using DataType = int8_t;
-using AccType  = int32_t;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|          |          |          |           |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|          |          |          |           |            |            |            |                    |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,     128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   2,      64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //##############################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //##############################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //##############################################################################|          |          |          |           |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //##############################################################################|          |          |          |           |            |            |            |                                 |       |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,     128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   2,      64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_int8_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
deleted file mode 100644
index 6b5f71ff78..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                    |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |        ./       |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
deleted file mode 100644
index 214aea289b..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|       ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|               Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|     Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                   |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     256,   256,   128,     4,   8,  32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     128,   128,   128,     4,   8,  32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     256,   128,   256,     4,   8,  32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,  
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     256,   128,   128,     4,   8,  32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     256,    64,   128,     4,   8,  32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     128,    32,   128,     4,   8,  32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     128,    64,   128,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     256,   128,    64,     4,   8,  32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,      64,    32,    64,     4,   8,  32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     128,   128,    64,     4,   8,  32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,      64,    64,    64,     4,   8,  32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,     128,   128,    32,     4,   8,  32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdDataDefault,   3,      64,    64,    32,     4,   8,  32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   256,     4,  8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,   128,     4,  8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   128,     4,  8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    64,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    64,   128,     4,  8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    64,     4,  8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,    64,     4,  8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,    64,   128,     4,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    32,     4,  8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    32,   128,     4,  8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    32,     4,  8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    32,    64,     4,  8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              8,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f16_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
deleted file mode 100644
index c3e8b5e8c7..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
+++ /dev/null
@@ -1,88 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                    |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|       |        |        |        |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|       |        |        |        |            |            |            |                                 |       |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   256,   128,     4,  4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   256,     4,  4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,   128,     4,  4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   128,     4,  4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    64,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    64,   128,     4,  4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    64,     4,  4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,    64,     4,  4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,    64,   128,     4,  4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    32,     4,  4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    32,   128,     4,  4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    32,     4,  4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,              4,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    32,    64,     4,  4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,              4,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_f32_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
deleted file mode 100644
index 9142b8049b..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_data/device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instance.cpp
+++ /dev/null
@@ -1,89 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using DataType = int8_t;
-using AccType  = int32_t;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdDataDefault =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Default;
-
-static constexpr auto ConvBwdDataFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardDataSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances =
-    std::tuple<
-        // clang-format off
-        //#############################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|        ConvBackward|    Num| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //#############################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                Data|    Dim|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //#############################################################################|          |          |          |           |   Operation|   Operation|   Operation|      Specialization|Spatial|      |      |      |      |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //#############################################################################|          |          |          |           |            |            |            |                    |       |      |      |      |      |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   256,   128,     4,  16,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,   256,     4,  16,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,   128,     4,  16,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,   128,     4,  16,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,    64,   128,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,    64,   128,     4,  16,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,    32,   128,     4,  16,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    32,    64,     4,  16,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,    64,     4,  16,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    64,    64,     4,  16,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     256,   128,    64,     4,  16,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,     128,   128,    32,     4,  16,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<  DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataDefault,   3,      64,    64,    32,     4,  16,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-using device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_int8_instances =
-    std::tuple<
-        // clang-format off
-        //##############################################################################|    InData|   WeiData|   OutData|    AccData|          In|         Wei|         Out|                     ConvBackward|    Num| Block|  MPer|  NPer| K0Per|  K1| MPer| NPer| MXdl| NXdl|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CThreadTransfer| CThreadTransfer|
-        //##############################################################################|      Type|      Type|      Type|       Type| Elementwise| Elementwise| Elementwise|                             Data|    Dim|  Size| Block| Block| Block|    |  XDL|  XDL|  Per|  Per|    ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
-        //##############################################################################|          |          |          |           |   Operation|   Operation|   Operation|                   Specialization|Spatial|      |      |      |      |    |     |     | Wave| Wave|  Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
-        //##############################################################################|          |          |          |           |            |            |            |                                 |       |      |      |      |      |    |     |     |     |     |                 |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   256,   128,     4,  16,   32,   32,    4,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   256,     4,  16,   32,   32,    2,    4,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,   128,     4,  16,   32,   32,    4,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,   128,     4,  16,   32,   32,    2,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    64,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    64,   128,     4,  16,   32,   32,    2,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    64,     4,  16,   32,   32,    2,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,   128,    64,     4,  16,   32,   32,    2,    1,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     256,    64,   128,     4,  16,   32,   32,    1,    2,      S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 64, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,   128,    32,     4,  16,   32,   32,    2,    1,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              1,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,     128,    32,   128,     4,  16,   32,   32,    1,    2,      S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 32, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    64,    32,     4,  16,   32,   32,    2,    1,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              2,             16,      true,               7,               1>,
-        DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_K<   DataType,  DataType,  DataType,    AccType, PassThrough, PassThrough, PassThrough,  ConvBwdDataFilter1x1Stride1Pad0,   3,      64,    32,    64,     4,  16,   32,   32,    1,    2,      S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,             16,             16,      true,     S<4, 16, 1>,     S<2, 0, 1>,     S<0, 2, 1>,             1,              4,             16,      true,               7,               1>
-        // clang-format on
-        >;
-
-void add_device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances(
-    std::vector<DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(instances,
-                                   device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_int8_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_data_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_int8_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/CMakeLists.txt
deleted file mode 100644
index 7272163f2b..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/CMakeLists.txt
+++ /dev/null
@@ -1,19 +0,0 @@
-#device_convnd_bwd_weight_instance
-set(DEVICE_CONVND_BWD_WEIGHT_INSTANCE_SOURCE 
-    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp;
-    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp;
-    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp;
-    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
-    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
-    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp;
-    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
-    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
-    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
-)
-
-add_library(device_convnd_bwd_weight_instance OBJECT ${DEVICE_CONVND_BWD_WEIGHT_INSTANCE_SOURCE})
-target_compile_features(device_convnd_bwd_weight_instance PUBLIC) 
-set_target_properties(device_convnd_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-rocm_install(TARGETS device_convnd_bwd_weight_instance)
-
-clang_tidy_check(device_convnd_bwd_weight_instance)
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp
deleted file mode 100644
index c8aae435fe..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp
deleted file mode 100644
index 6e4964ce01..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp
deleted file mode 100644
index ed25442dc4..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp
+++ /dev/null
@@ -1,86 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
deleted file mode 100644
index 3a0dfeb6f4..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
deleted file mode 100644
index 025c7c86d8..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_default_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-void add_device_convnd_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_default_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
deleted file mode 100644
index cde50d779b..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
+++ /dev/null
@@ -1,86 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f32_default_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_convnd_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_f32_default_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
deleted file mode 100644
index 1e2ad43a31..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
deleted file mode 100644
index 647a898242..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp
+++ /dev/null
@@ -1,88 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f16_default_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances,
-        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f16_default_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
deleted file mode 100644
index 40754a09f0..0000000000
--- a/library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp
+++ /dev/null
@@ -1,87 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f32_default_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     3,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     3,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances,
-        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f32_default_instances{});
-    add_device_operation_instances(
-        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt
index 789c5b628f..194748ba67 100644
--- a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/CMakeLists.txt
@@ -1,9 +1,9 @@
 # device_gemm_add_add_fastgelu_instance
 set(DEVICE_GEMM_ADD_ADD_FASTGELU_INSTANCE_SOURCE
-   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp;
-   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp;
-   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp;
-   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp;
+   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp;
+   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp;
+   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp;
+   device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp;
 )
 
 add_library(device_gemm_add_add_fastgelu_instance OBJECT ${DEVICE_GEMM_ADD_ADD_FASTGELU_INSTANCE_SOURCE})
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp
new file mode 100644
index 0000000000..2adf0de661
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instance.cpp
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16           = ck::half_t;
+using F32           = float;
+using F16_F16_Tuple = ck::Tuple<F16, F16>;
+
+using Row           = ck::tensor_layout::gemm::RowMajor;
+using Col           = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Row_Tuple = ck::Tuple<Row, Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
+using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// e = elementwise((a * b), d0, d1)
+// outout: e[m, n]
+// input: a[k, m], b[k, n], d0[m, n], d1[m, n]
+using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##############################|      A|      B|            Ds|      E| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|        Layout| Layout|  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |              |       |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |              |       |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
+        // clang-format on
+        >;
+
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
+                                                    Row,
+                                                    Row_Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    AddAddFastGelu>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_kn_mn_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp
new file mode 100644
index 0000000000..0ce6b696a4
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instance.cpp
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16           = ck::half_t;
+using F32           = float;
+using F16_F16_Tuple = ck::Tuple<F16, F16>;
+
+using Row           = ck::tensor_layout::gemm::RowMajor;
+using Col           = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Row_Tuple = ck::Tuple<Row, Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
+using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// e = elementwise((a * b), d0, d1)
+// outout: e[m, n]
+// input: a[k, m], b[n, k], d0[m, n], d1[m, n]
+using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##############################|      A|      B|            Ds|      E| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|        Layout| Layout|  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |              |       |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |              |       |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
+        // clang-format on
+        >;
+
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
+                                                    Col,
+                                                    Row_Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    AddAddFastGelu>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_km_nk_mn_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp
new file mode 100644
index 0000000000..26ba43db45
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instance.cpp
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16           = ck::half_t;
+using F32           = float;
+using F16_F16_Tuple = ck::Tuple<F16, F16>;
+
+using Row           = ck::tensor_layout::gemm::RowMajor;
+using Col           = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Row_Tuple = ck::Tuple<Row, Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
+using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// e = elementwise((a * b), d0, d1)
+// outout: e[m, n]
+// input: a[m, k], b[k, n], d0[m, n], d1[m, n]
+using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##############################|      A|      B|            Ds|      E| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|        Layout| Layout|  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |              |       |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |              |       |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
+        // clang-format on
+        >;
+
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
+                                                    Row,
+                                                    Row_Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    AddAddFastGelu>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_kn_mn_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp
new file mode 100644
index 0000000000..66bf17bc25
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instance.cpp
@@ -0,0 +1,82 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16           = ck::half_t;
+using F32           = float;
+using F16_F16_Tuple = ck::Tuple<F16, F16>;
+
+using Row           = ck::tensor_layout::gemm::RowMajor;
+using Col           = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Row_Tuple = ck::Tuple<Row, Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
+using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
+
+static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
+
+// e = elementwise((a * b), d0, d1)
+// outout: e[m, n]
+// input: a[m, k], b[n, k], d0[m, n], d1[m ,n]
+using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instances =
+    std::tuple<
+        // clang-format off
+        //##############################|      A|      B|            Ds|      E| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|        Layout| Layout|  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |              |       |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |              |       |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Row_Tuple,    Row,   F16,   F16,     F32,      F32, F16_F16_Tuple,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
+                                                    Col,
+                                                    Row_Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    AddAddFastGelu>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_f16_f16_mk_nk_mn_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
deleted file mode 100644
index e00a66c5df..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
+++ /dev/null
@@ -1,81 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16           = ck::half_t;
-using F32           = float;
-using F16_F16_TUPLE = ck::Tuple<F16, F16>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
-using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// e = elementwise((a * b), d0, d1)
-// outout: e[m, n]
-// input: a[k, m], b[k, n], d0[m, n], d1[m, n]
-using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances = std::tuple<
-    // clang-format off
-        //##############################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
-                                                    Row,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    AddAddFastGelu>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
deleted file mode 100644
index a5f398937a..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
+++ /dev/null
@@ -1,81 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16           = ck::half_t;
-using F32           = float;
-using F16_F16_TUPLE = ck::Tuple<F16, F16>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
-using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// e = elementwise((a * b), d0, d1)
-// outout: e[m, n]
-// input: a[k, m], b[n, k], d0[m, n], d1[m, n]
-using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances = std::tuple<
-    // clang-format off
-        //##############################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
-                                                    Col,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    AddAddFastGelu>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
deleted file mode 100644
index 8e2b5cf669..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
+++ /dev/null
@@ -1,81 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16           = ck::half_t;
-using F32           = float;
-using F16_F16_TUPLE = ck::Tuple<F16, F16>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
-using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// e = elementwise((a * b), d0, d1)
-// outout: e[m, n]
-// input: a[m, k], b[k, n], d0[m, n], d1[m, n]
-using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances = std::tuple<
-    // clang-format off
-        //##############################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
-                                                    Row,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    AddAddFastGelu>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
deleted file mode 100644
index e28889a29d..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_add_add_fastgelu/device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
+++ /dev/null
@@ -1,78 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16           = ck::half_t;
-using F32           = float;
-using F16_F16_TUPLE = ck::Tuple<F16, F16>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough    = ck::tensor_operation::element_wise::PassThrough;
-using AddAddFastGelu = ck::tensor_operation::element_wise::AddAddFastGelu;
-
-static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
-
-// e = elementwise((a * b), d0, d1)
-// outout: e[m, n]
-// input: a[m, k], b[n, k], d0[m, n], d1[m ,n]
-using device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances = std::tuple<
-    // clang-format off
-        //##############################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle|        DsData| EData|           A|           B|            CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|          Type|  Type| Elementwise| Elementwise|    Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |              |      |   Operation|   Operation|      Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |              |      |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_F16_TUPLE,   F16, PassThrough, PassThrough, AddAddFastGelu,    GemmDefault,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
-                                                    Col,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    AddAddFastGelu>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_add_add_fastgelu_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt
index e6c93da88c..6bbebb7576 100644
--- a/library/src/tensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/gemm_bilinear/CMakeLists.txt
@@ -1,9 +1,9 @@
 # device_gemm_bilinear_instance
 set(DEVICE_GEMM_BILINEAR_INSTANCE_SOURCE
-   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp;
-   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp;
-   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp;
-   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp;
+   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp;
+   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp;
+   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp;
+   device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp;
 )
 
 add_library(device_gemm_bilinear_instance OBJECT ${DEVICE_GEMM_BILINEAR_INSTANCE_SOURCE})
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp
new file mode 100644
index 0000000000..e4bc35e24d
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instance.cpp
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16       = ck::half_t;
+using F32       = float;
+using F16_Tuple = ck::Tuple<ck::half_t>;
+
+using Row       = ck::tensor_layout::gemm::RowMajor;
+using Col       = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Tuple = ck::Tuple<Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
+
+static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// e[m, n] = bilinear(a[k, m] * b[k, n], d[m, n])
+using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instances = std::tuple<
+    // clang-format off
+        // no padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+
+        // M/N/K Padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
+    // clang-format on
+    >;
+
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
+                                                    Row,
+                                                    Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    Bilinear>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_kn_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp
new file mode 100644
index 0000000000..ad95c30a5e
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instance.cpp
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16       = ck::half_t;
+using F32       = float;
+using F16_Tuple = ck::Tuple<ck::half_t>;
+
+using Row       = ck::tensor_layout::gemm::RowMajor;
+using Col       = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Tuple = ck::Tuple<Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
+
+static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// e[m, n] = bilinear(a[k, m] * b[n, k], d[m, n])
+using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instances = std::tuple<
+    // clang-format off
+        // no padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+
+        // M/N/K Padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Col,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
+    // clang-format on
+    >;
+
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
+                                                    Col,
+                                                    Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    Bilinear>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_km_nk_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp
new file mode 100644
index 0000000000..2c6f6aa3dc
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instance.cpp
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16       = ck::half_t;
+using F32       = float;
+using F16_Tuple = ck::Tuple<ck::half_t>;
+
+using Row       = ck::tensor_layout::gemm::RowMajor;
+using Col       = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Tuple = ck::Tuple<Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
+
+static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// e[m, n] = bilinear(a[m, k] * b[k, n], d[m, n])
+using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instances = std::tuple<
+    // clang-format off
+        // no padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+
+        // M/N/K padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Row, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
+    // clang-format on
+    >;
+
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
+                                                    Row,
+                                                    Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    Bilinear>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_kn_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp
new file mode 100644
index 0000000000..9cfda63b9b
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instance.cpp
@@ -0,0 +1,99 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16       = ck::half_t;
+using F32       = float;
+using F16_Tuple = ck::Tuple<ck::half_t>;
+
+using Row       = ck::tensor_layout::gemm::RowMajor;
+using Col       = ck::tensor_layout::gemm::ColumnMajor;
+using Row_Tuple = ck::Tuple<Row>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
+
+static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// e[m, n] = bilinear(a[m, k] * b[n, k], d[m, n])
+using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instances = std::tuple<
+    // clang-format off
+        // no padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // M/N/N padding
+        //##############################|      A|      B|        Ds|      E| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //##############################| Layout| Layout|    Layout| Layout|  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //##############################|       |       |          |       |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //##############################|       |       |          |       |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGemmMultipleD_Xdl_CShuffle<    Row,    Col, Row_Tuple,    Row,   F16,   F16,     F32,      F16, F16_Tuple,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
+                                                    Col,
+                                                    Row_Tuple,
+                                                    Row,
+                                                    F16,
+                                                    F16,
+                                                    F16_Tuple,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    Bilinear>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_f16_mk_nk_mn_mn_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
deleted file mode 100644
index 4b8777a424..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
+++ /dev/null
@@ -1,103 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16       = ck::half_t;
-using F32       = float;
-using F16_TUPLE = ck::Tuple<ck::half_t>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
-
-static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
-static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
-
-// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
-using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances = std::tuple<
-    // clang-format off
-        // no padding
-        //##############################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-
-        // M/N/K Padding
-        //##############################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
-                                                    Row,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    Bilinear>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
deleted file mode 100644
index 589e4bf6d1..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
+++ /dev/null
@@ -1,103 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16       = ck::half_t;
-using F32       = float;
-using F16_TUPLE = ck::Tuple<ck::half_t>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
-
-static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
-static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
-
-// Compilation parameters for a[k, m] * b[n, k] = c[m, n]
-using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances = std::tuple<
-    // clang-format off
-        // no padding
-        //##############################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-
-        // M/N/K Padding
-        //##############################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Col,      Col,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Col,
-                                                    Col,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    Bilinear>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
deleted file mode 100644
index d18b7c2668..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
+++ /dev/null
@@ -1,104 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16       = ck::half_t;
-using F32       = float;
-using F16_TUPLE = ck::Tuple<F16>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
-;
-
-static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
-static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
-
-// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
-using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances = std::tuple<
-    // clang-format off
-        // no padding
-        //##############################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-
-        // M/N/K padding
-        //##############################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,      Row,    Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
-                                                    Row,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    Bilinear>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
deleted file mode 100644
index 29763ea4a2..0000000000
--- a/library/src/tensor_operation_instance/gpu/gemm_bilinear/device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
+++ /dev/null
@@ -1,97 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
-#include "ck/tensor_operation/gpu/device/device_gemm_multiple_d_xdl_cshuffle.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16       = ck::half_t;
-using F32       = float;
-using F16_TUPLE = ck::Tuple<F16>;
-
-using Row = ck::tensor_layout::gemm::RowMajor;
-using Col = ck::tensor_layout::gemm::ColumnMajor;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-using Bilinear    = ck::tensor_operation::element_wise::Bilinear;
-
-static constexpr auto GemmDefault    = ck::tensor_operation::device::GemmSpecialization::Default;
-static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
-
-// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
-using device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances = std::tuple<
-    // clang-format off
-        // no padding
-        //##############################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear,    GemmDefault,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-
-        // M/N/N padding
-        //##############################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle|    DsData| EData|           A|           B|         CDE|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##############################|        |        |        |  Type|  Type|    Type| DataType|      Type|  Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##############################|        |        |        |      |      |        |         |          |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##############################|        |        |        |      |      |        |         |          |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
-        DeviceGemmMultipleD_Xdl_CShuffle<     Row,     Col,     Row,   F16,   F16,     F32,      F32, F16_TUPLE,   F16, PassThrough, PassThrough,    Bilinear, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
-    // clang-format on
-    >;
-
-void add_device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances(
-    std::vector<std::unique_ptr<DeviceGemmMultipleD<Row,
-                                                    Col,
-                                                    Row,
-                                                    F16,
-                                                    F16,
-                                                    F16_TUPLE,
-                                                    F16,
-                                                    PassThrough,
-                                                    PassThrough,
-                                                    Bilinear>>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_gemm_bilinear_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/CMakeLists.txt
new file mode 100644
index 0000000000..43763f4675
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/CMakeLists.txt
@@ -0,0 +1,12 @@
+# device_grouped_conv1d_fwd_instance
+set(DEVICE_GROUPED_CONV1D_FWD_INSTANCE_SOURCE
+   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp;
+   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp;
+   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp;
+   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instance.cpp;
+)
+
+add_library(device_grouped_conv1d_fwd_instance OBJECT ${DEVICE_GROUPED_CONV1D_FWD_INSTANCE_SOURCE}) 
+set_target_properties(device_grouped_conv1d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
+clang_tidy_check(device_grouped_conv1d_fwd_instance)
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp
new file mode 100644
index 0000000000..1238c5796d
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = ck::bhalf_t;
+using F32  = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNWC = ck::tensor_layout::convolution::GNWC;
+using GKXC = ck::tensor_layout::convolution::GKXC;
+using GNWK = ck::tensor_layout::convolution::GNWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, wi, c] * wei[g, k, x, c] = out[g, n, wo, k]
+using device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              BF16,
+                                                              BF16,
+                                                              Empty_Tuple,
+                                                              BF16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp
new file mode 100644
index 0000000000..ead16d11ac
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNWC = ck::tensor_layout::convolution::GNWC;
+using GKXC = ck::tensor_layout::convolution::GKXC;
+using GNWK = ck::tensor_layout::convolution::GNWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, wi, c] * wei[g, k, x, c] = out[g, n, wo, k]
+using device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp
new file mode 100644
index 0000000000..dbb9f955f3
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNWC = ck::tensor_layout::convolution::GNWC;
+using GKXC = ck::tensor_layout::convolution::GKXC;
+using GNWK = ck::tensor_layout::convolution::GNWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, wi, c] * wei[g, k, x, c] = out[g, n, wo, k]
+using device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              F32,
+                                                              F32,
+                                                              Empty_Tuple,
+                                                              F32,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instance.cpp
new file mode 100644
index 0000000000..5d4c8a3271
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instance.cpp
@@ -0,0 +1,125 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNWC = ck::tensor_layout::convolution::GNWC;
+using GKXC = ck::tensor_layout::convolution::GKXC;
+using GNWK = ck::tensor_layout::convolution::GNWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, wi, c] * wei[g, k, x, c] = out[g, n, wo, k]
+using device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instances = std::tuple<
+    // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+                                                                                                                                                                                                                
+        // Filter1x1Stride1Pad0                                                                                                                                                                                 
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       1,   GNWC,   GKXC, Empty_Tuple,   GNWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
+                                                              GNWC,
+                                                              GKXC,
+                                                              Empty_Tuple,
+                                                              GNWK,
+                                                              int8_t,
+                                                              int8_t,
+                                                              Empty_Tuple,
+                                                              int8_t,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_int8_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/CMakeLists.txt
new file mode 100644
index 0000000000..330f6df787
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/CMakeLists.txt
@@ -0,0 +1,15 @@
+# device_grouped_conv2d_fwd_instance
+set(DEVICE_GROUPED_CONV2D_FWD_INSTANCE_SOURCE
+   # GNHWC, GKYXC, GNHWK
+   device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp;
+   device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instance.cpp;
+   device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instance.cpp;
+   device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instance.cpp;
+   # NHWGC, KYXGC, NHWGK
+   device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instance.cpp;
+)
+
+add_library(device_grouped_conv2d_fwd_instance OBJECT ${DEVICE_GROUPED_CONV2D_FWD_INSTANCE_SOURCE}) 
+set_target_properties(device_grouped_conv2d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
+clang_tidy_check(device_grouped_conv2d_fwd_instance)
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp
new file mode 100644
index 0000000000..c6742a0405
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = ck::bhalf_t;
+using F32  = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNHWC = ck::tensor_layout::convolution::GNHWC;
+using GKYXC = ck::tensor_layout::convolution::GKYXC;
+using GNHWK = ck::tensor_layout::convolution::GNHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, hi, wi, c] * wei[g, k, y, x, c] = out[g, n, ho, wo, k]
+using device_grouped_conv1d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv1d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              BF16,
+                                                              BF16,
+                                                              Empty_Tuple,
+                                                              BF16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_grouped_conv1d_fwd_xdl_gnhwc_gkyxc_gnhwk_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instance.cpp
new file mode 100644
index 0000000000..e9a5977f02
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNHWC = ck::tensor_layout::convolution::GNHWC;
+using GKYXC = ck::tensor_layout::convolution::GKYXC;
+using GNHWK = ck::tensor_layout::convolution::GNHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, hi ,wi, c] * wei[g, k, y, x, c] = out[g, n, ho, wo, k]
+using device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instance.cpp
new file mode 100644
index 0000000000..f1b4f52d86
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instance.cpp
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNHWC = ck::tensor_layout::convolution::GNHWC;
+using GKYXC = ck::tensor_layout::convolution::GKYXC;
+using GNHWK = ck::tensor_layout::convolution::GNHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, hi, wi, c] * wei[g, k, y, x, c] = out[g, n, ho, wo, k]
+using device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              F32,
+                                                              F32,
+                                                              Empty_Tuple,
+                                                              F32,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instance.cpp
new file mode 100644
index 0000000000..2494effda8
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instance.cpp
@@ -0,0 +1,125 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNHWC = ck::tensor_layout::convolution::GNHWC;
+using GKYXC = ck::tensor_layout::convolution::GKYXC;
+using GNHWK = ck::tensor_layout::convolution::GNHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[g, n, hi, wi, c] * wei[g, k, y, x, c] = out[g, n, ho, wo, k]
+using device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instances = std::tuple<
+    // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+                                                                                                                                                                                                                
+        // Filter1x1Stride1Pad0                                                                                                                                                                                 
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  GNHWC,  GKYXC, Empty_Tuple,  GNHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              GNHWC,
+                                                              GKYXC,
+                                                              Empty_Tuple,
+                                                              GNHWK,
+                                                              int8_t,
+                                                              int8_t,
+                                                              Empty_Tuple,
+                                                              int8_t,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_grouped_conv2d_fwd_xdl_gnhwc_gkyxc_gnhwk_int8_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instance.cpp
new file mode 100644
index 0000000000..475ff46aa1
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv2d_fwd/device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NHWGC = ck::tensor_layout::convolution::NHWGC;
+using KYXGC = ck::tensor_layout::convolution::KYXGC;
+using NHWGK = ck::tensor_layout::convolution::NHWGK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// Compilation parameters for in[n, hi, wi, g, c] * wei[k, y, x, g, c] = out[n, ho, wo, g, k]
+using device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       2,  NHWGC,  KYXGC, Empty_Tuple,  NHWGK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<2,
+                                                              NHWGC,
+                                                              KYXGC,
+                                                              Empty_Tuple,
+                                                              NHWGK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(instances,
+                                   device_grouped_conv2d_fwd_xdl_nhwgc_kyxgc_nhwgk_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
new file mode 100644
index 0000000000..ab7f60bf7f
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
@@ -0,0 +1,12 @@
+# device_grouped_conv3d_fwd_instance
+set(DEVICE_GROUPED_CONV3D_FWD_INSTANCE_SOURCE
+   device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp;
+   device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp;
+   device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp;
+   device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp;
+)
+
+add_library(device_grouped_conv3d_fwd_instance OBJECT ${DEVICE_GROUPED_CONV3D_FWD_INSTANCE_SOURCE}) 
+set_target_properties(device_grouped_conv3d_fwd_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
+clang_tidy_check(device_grouped_conv3d_fwd_instance)
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp
new file mode 100644
index 0000000000..e6578beeff
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = ck::bhalf_t;
+using F32  = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNDHWC = ck::tensor_layout::convolution::GNDHWC;
+using GKZYXC = ck::tensor_layout::convolution::GKZYXC;
+using GNDHWK = ck::tensor_layout::convolution::GNDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// in[g, n, di, hi, wi, c] * wei[g, k, z, y, x, c] = out[g, n, do, ho, wo, k]
+using device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,  BF16,  BF16,     F32,     BF16, Empty_Tuple,  BF16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              BF16,
+                                                              BF16,
+                                                              Empty_Tuple,
+                                                              BF16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp
new file mode 100644
index 0000000000..77a2e1e571
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instance.cpp
@@ -0,0 +1,129 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNDHWC = ck::tensor_layout::convolution::GNDHWC;
+using GKZYXC = ck::tensor_layout::convolution::GKZYXC;
+using GNDHWK = ck::tensor_layout::convolution::GNDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// in[g, n, di, hi, wi, c] * wei[g, k, z, y, x, c] = out[g, n, do, ho, wo, k]
+using device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              F16,
+                                                              F16,
+                                                              Empty_Tuple,
+                                                              F16,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp
new file mode 100644
index 0000000000..337d1183dd
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instance.cpp
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNDHWC = ck::tensor_layout::convolution::GNDHWC;
+using GKZYXC = ck::tensor_layout::convolution::GKZYXC;
+using GNDHWK = ck::tensor_layout::convolution::GNDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// in[g, n, di, hi, wi, c] * wei[g, k, z, y, x, c] = out[g, n, do, ho, wo, k]
+using device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instances =
+    std::tuple<
+        // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+
+        // Filter1x1Stride1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|          Ds| EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|    DataType|  Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |      |      |        |         |            |      |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    16,   4,   4,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    16,   4,   4,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    16,   4,   4,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    16,   4,   4,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    16,   4,   4,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1, 16, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    16,   4,   4,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1, 16>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    16,   4,   4,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK,   F32,   F32,     F32,      F32, Empty_Tuple,   F32, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    16,   4,   4,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              4,              4,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              4,              4,         1,           1,           1,              S<1,  8, 1,  8>,              4>
+        // clang-format on
+        >;
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              F32,
+                                                              F32,
+                                                              Empty_Tuple,
+                                                              F32,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp
new file mode 100644
index 0000000000..7cc2b10d2f
--- /dev/null
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instance.cpp
@@ -0,0 +1,125 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/device/convolution_forward_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using Empty_Tuple = ck::Tuple<>;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using GNDHWC = ck::tensor_layout::convolution::GNDHWC;
+using GKZYXC = ck::tensor_layout::convolution::GKZYXC;
+using GNDHWK = ck::tensor_layout::convolution::GNDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvFwdDefault =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto ConvFwd1x1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Pad0;
+
+static constexpr auto ConvFwd1x1S1P0 =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Filter1x1Stride1Pad0;
+
+static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+// in[g, n, di, hi, wi, c] * wei[g, k, z, y, x, c] = out[g, n, do, ho, wo, k]
+using device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instances = std::tuple<
+    // clang-format off
+        // Default
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwdDefault, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+
+        // Filter1x1Pad0
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough,   ConvFwd1x1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+                                                                                                                                                                                                                
+        // Filter1x1Stride1Pad0                                                                                                                                                                                 
+        //########################################|  NumDim|      A|      B|          Ds|      E|  AData|  BData| AccData| CShuffle|          Ds|  EData|           A|           B|         CDE|    ConvForward|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //########################################| Spatial| Layout| Layout|      Layout| Layout|   Type|   Type|    Type| DataType|    DataType|   Type| Elementwise| Elementwise| Elementwise| Specialization| Specialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |   Operation|   Operation|   Operation|               |               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //########################################|        |       |       |            |       |       |       |        |         |            |       |            |            |            |               |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedConvFwdMultipleD_Xdl_CShuffle<       3, GNDHWC, GKZYXC, Empty_Tuple, GNDHWK, int8_t, int8_t, int32_t,   int8_t, Empty_Tuple, int8_t, PassThrough, PassThrough, PassThrough, ConvFwd1x1S1P0, GemmMNKPadding,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<3,
+                                                              GNDHWC,
+                                                              GKZYXC,
+                                                              Empty_Tuple,
+                                                              GNDHWK,
+                                                              int8_t,
+                                                              int8_t,
+                                                              Empty_Tuple,
+                                                              int8_t,
+                                                              PassThrough,
+                                                              PassThrough,
+                                                              PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_grouped_conv3d_fwd_xdl_gndhwc_gkzyxc_gndhwk_int8_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
diff --git a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp
index abbbbb3335..e9901a06f2 100644
--- a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp
@@ -7,6 +7,7 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -23,45 +24,46 @@ using Col = ck::tensor_layout::gemm::ColumnMajor;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
-using DsType = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
-// Compilation parameters for a[k, m] * b[k, n] = c[m, n]
+// a[k, m] * b[k, n] = e[m, n]
 using device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instances = std::tuple<
     // clang-format off
-        //##################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle| DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##################|        |        |        |  Type|  Type|    Type| DataType|   Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##################|        |        |        |      |      |        |         |       |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##################|        |        |        |      |      |        |         |       |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>
+        //###################|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|      DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //###################| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|        Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //###################|       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //###################|       |       |            |       |      |      |        |         |            |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   2,   2,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   2,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   2,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   2,   2,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   2,   2,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   2,   2,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Col,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>
     // clang-format on
     >;
 
 void add_device_grouped_gemm_xdl_f16_f16_f16_km_kn_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Col,
                                                   Row,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp
index 8c7dc5d448..8ac934a188 100644
--- a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp
@@ -7,6 +7,7 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -23,44 +24,46 @@ using Col = ck::tensor_layout::gemm::ColumnMajor;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
-using DsType = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
+// a[k, m] * b[n, k] = e[m, n]
 using device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instances = std::tuple<
     // clang-format off
-        //##################| ALayout| BLayout| ELayout| AData| BData| AccData| CShuffle| DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##################|        |        |        |  Type|  Type|    Type| DataType|   Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##################|        |        |        |      |      |        |         |       |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##################|        |        |        |      |      |        |         |       |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Col,      Col,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>
+        //###################|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|      DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //###################| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|        Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //###################|       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //###################|       |       |            |       |      |      |        |         |            |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   2,   8,   32,   32,    2,    4,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   2,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   2,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   2,   8,   32,   32,    2,    2,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   2,   8,   32,   32,    2,    1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              2,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   2,   8,   32,   32,    1,    2,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              4,              2,         0,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Col,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,              1,              1,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>
     // clang-format on
     >;
 
 void add_device_grouped_gemm_xdl_f16_f16_f16_km_nk_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Col,
                                                   Col,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
index 3e330fa577..c8215085a8 100644
--- a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
@@ -7,6 +7,7 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -23,45 +24,46 @@ using Col = ck::tensor_layout::gemm::ColumnMajor;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
-using DsType = ck::Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
-// Compilation parameters for a[m, k] * b[k, n] = c[m, n]
+// a[m, k] * b[k, n] = e[m, n]
 using device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instances = std::tuple<
     // clang-format off
-        //##################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle| DsType| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##################|        |        |        |  Type|  Type|    Type| DataType|   Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##################|        |        |        |      |      |        |         |       |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##################|        |        |        |      |      |        |         |       |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,      Row,    Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>
+        //###################|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|      DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //###################| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|        Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //###################|       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //###################|       |       |            |       |      |      |        |         |            |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   2,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   2,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 4>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   2,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   2,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<16,16, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              1,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   2,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<8, 32, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              4,              2,         0,           1,           1,               S<1, 32, 1, 8>,              8>,
+        DeviceGroupedGemm_Xdl<    Row,    Row, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<0, 2, 1>,     S<0, 2, 1>,             1,              2,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>
     // clang-format on
     >;
 
 void add_device_grouped_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Row,
                                                   Row,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
diff --git a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
index 15bb9c1306..4f301315b7 100644
--- a/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_gemm/device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
@@ -7,6 +7,7 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/device_grouped_gemm_xdl.hpp"
+
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
 
 namespace ck {
@@ -23,42 +24,43 @@ using Col = ck::tensor_layout::gemm::ColumnMajor;
 template <ck::index_t... Is>
 using S = ck::Sequence<Is...>;
 
-using DsType = Tuple<>;
+using Empty_Tuple = ck::Tuple<>;
 
 using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 
 static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
 
-// Compilation parameters for a[m, k] * b[n, k] = c[m, n]
+// a[m, k] * b[n, k] = e[m, n]
 using device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances = std::tuple<
     // clang-format off
-        //##################| ALayout| BLayout| CLayout| AData| BData| AccData| CShuffle| DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
-        //##################|        |        |        |  Type|  Type|    Type| DataType|   Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
-        //##################|        |        |        |      |      |        |         |       |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
-        //##################|        |        |        |      |      |        |         |       |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,              8>,
-        DeviceGroupedGemmXdl<     Row,     Col,     Row,   F16,   F16,     F32,      F16, DsType,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,              8>
+        //###################|      A|      B|          Ds|      E| AData| BData| AccData| CShuffle|      DsData| EData|           A|           B|           C|           GEMM| NumGemmK| Block|  MPer|  NPer|  KPer| AK1| BK1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle| CBlockTransferClusterLengths|  CBlockTransfer|
+        //###################| Layout| Layout|      Layout| Layout|  Type|  Type|    Type| DataType|        Type|  Type| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
+        //###################|       |       |            |       |      |      |        |         |            |      |   Operation|   Operation|   Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
+        //###################|       |       |            |       |      |      |        |         |            |      |            |            |            |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   256,    32,   8,   8,   32,   32,    2,    4,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    64,   128,    32,   8,   8,   32,   32,    2,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,    64,    64,    64,    32,   8,   8,   32,   32,    2,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,   128,    64,    32,   8,   8,   32,   32,    2,    1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,   128,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 4>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,   128,    32,   128,    32,   8,   8,   32,   32,    1,    2,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 32, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 8>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,    64,    64,    32,    32,   8,   8,   32,   32,    2,    1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>,
+        DeviceGroupedGemm_Xdl<    Row,    Col, Empty_Tuple,    Row,   F16,   F16,     F32,      F16, Empty_Tuple,   F16, PassThrough, PassThrough, PassThrough,    GemmDefault,        1,    64,    32,    64,    32,   8,   8,   32,   32,    1,    2,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 16, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 16, 1, 4>,               8>
     // clang-format on
     >;
 
 void add_device_grouped_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(
     std::vector<std::unique_ptr<DeviceGroupedGemm<Row,
                                                   Col,
+                                                  Empty_Tuple,
                                                   Row,
                                                   F16,
                                                   F16,
-                                                  DsType,
+                                                  Empty_Tuple,
                                                   F16,
                                                   PassThrough,
                                                   PassThrough,
diff --git a/library/src/utility/CMakeLists.txt b/library/src/utility/CMakeLists.txt
index afa6de5119..7f6a59eebe 100644
--- a/library/src/utility/CMakeLists.txt
+++ b/library/src/utility/CMakeLists.txt
@@ -1,11 +1,28 @@
-set(CONV_UTIL_SOURCE
-    conv_util.cpp
+## utility
+set(UTILITY_SOURCE
+    device_memory.cpp
+    host_tensor.cpp
+    convolution_parameter.cpp
 )
 
-add_library(conv_util SHARED ${CONV_UTIL_SOURCE})
-target_link_libraries(conv_util PRIVATE host_tensor)
-target_compile_features(conv_util PUBLIC)
-set_target_properties(conv_util PROPERTIES POSITION_INDEPENDENT_CODE ON)
-target_include_directories(conv_util SYSTEM PUBLIC $<BUILD_INTERFACE:${HALF_INCLUDE_DIR}>)
+add_library(utility STATIC ${UTILITY_SOURCE})
+add_library(composable_kernel::utility ALIAS utility)
 
-clang_tidy_check(conv_util)
+target_include_directories(utility PUBLIC
+    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck>"
+    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}/ck/library/utility>"
+)
+
+rocm_install(
+    TARGETS utility
+    EXPORT utilityTargets
+)
+
+rocm_install(
+    EXPORT utilityTargets
+    FILE composable_kernelutilityTargets.cmake
+    NAMESPACE composable_kernel::
+    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
+)
+
+clang_tidy_check(utility)
diff --git a/library/src/utility/conv_util.cpp b/library/src/utility/conv_util.cpp
deleted file mode 100644
index 3a223770cd..0000000000
--- a/library/src/utility/conv_util.cpp
+++ /dev/null
@@ -1,242 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include "ck/library/utility/conv_util.hpp"
-
-namespace ck {
-namespace utils {
-namespace conv {
-
-/**
- * @brief      Calculate number of FLOPs for Convolution
- *
- * @param[in]  N                       Batch size.
- * @param[in]  C                       Number of input channels.
- * @param[in]  K                       Number of output channels.
- * @param[in]  filter_spatial_lengths  Filter spatial dimensions lengths.
- * @param[in]  output_spatial_lengths  Convolution output spatial dimensions
- *                                     lengths.
- *
- * @return     The number of flops.
- */
-std::size_t get_flops(ck::index_t N,
-                      ck::index_t C,
-                      ck::index_t K,
-                      const std::vector<ck::index_t>& filter_spatial_lengths,
-                      const std::vector<ck::index_t>& output_spatial_lengths)
-{
-    // 2 * N * K * <output spatial lengths product> * C * <filter spatial lengths product>
-    return static_cast<std::size_t>(2) * N * K *
-           std::accumulate(std::begin(output_spatial_lengths),
-                           std::end(output_spatial_lengths),
-                           static_cast<std::size_t>(1),
-                           std::multiplies<std::size_t>()) *
-           C *
-           std::accumulate(std::begin(filter_spatial_lengths),
-                           std::end(filter_spatial_lengths),
-                           static_cast<std::size_t>(1),
-                           std::multiplies<std::size_t>());
-}
-
-ConvParams::ConvParams()
-    : num_dim_spatial_(2),
-      N_(128),
-      K_(256),
-      C_(192),
-      filter_spatial_lengths_(2, 3),
-      input_spatial_lengths_(2, 71),
-      conv_filter_strides_(2, 2),
-      conv_filter_dilations_(2, 1),
-      input_left_pads_(2, 1),
-      input_right_pads_(2, 1)
-{
-}
-
-ConvParams::ConvParams(ck::index_t n_dim,
-                       ck::index_t n_batch,
-                       ck::index_t n_out_channels,
-                       ck::index_t n_in_channels,
-                       const std::vector<ck::index_t>& filters_len,
-                       const std::vector<ck::index_t>& input_len,
-                       const std::vector<ck::index_t>& strides,
-                       const std::vector<ck::index_t>& dilations,
-                       const std::vector<ck::index_t>& left_pads,
-                       const std::vector<ck::index_t>& right_pads)
-    : num_dim_spatial_(n_dim),
-      N_(n_batch),
-      K_(n_out_channels),
-      C_(n_in_channels),
-      filter_spatial_lengths_(filters_len),
-      input_spatial_lengths_(input_len),
-      conv_filter_strides_(strides),
-      conv_filter_dilations_(dilations),
-      input_left_pads_(left_pads),
-      input_right_pads_(right_pads)
-{
-    if(ck::type_convert<ck::index_t>(filter_spatial_lengths_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(input_spatial_lengths_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(conv_filter_strides_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(conv_filter_dilations_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(input_left_pads_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(input_right_pads_.size()) != num_dim_spatial_)
-    {
-        throw(
-            std::runtime_error("ConvParams::GetOutputSpatialLengths: "
-                               "parameter size is different from number of declared dimensions!"));
-    }
-}
-
-std::vector<ck::index_t> ConvParams::GetOutputSpatialLengths() const
-{
-    if(ck::type_convert<ck::index_t>(filter_spatial_lengths_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(input_spatial_lengths_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(conv_filter_strides_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(conv_filter_dilations_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(input_left_pads_.size()) != num_dim_spatial_ ||
-       ck::type_convert<ck::index_t>(input_right_pads_.size()) != num_dim_spatial_)
-    {
-        throw(
-            std::runtime_error("ConvParams::GetOutputSpatialLengths: "
-                               "parameter size is different from number of declared dimensions!"));
-    }
-
-    std::vector<ck::index_t> out_spatial_len(num_dim_spatial_, 0);
-    for(ck::index_t i = 0; i < num_dim_spatial_; ++i)
-    {
-        // XEff = (X - 1) * conv_dilation_w + 1;
-        // Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
-        const ck::index_t idx_eff =
-            (filter_spatial_lengths_[i] - 1) * conv_filter_dilations_[i] + 1;
-        out_spatial_len[i] =
-            (input_spatial_lengths_[i] + input_left_pads_[i] + input_right_pads_[i] - idx_eff) /
-                conv_filter_strides_[i] +
-            1;
-    }
-    return out_spatial_len;
-}
-
-ConvParams parse_conv_params(int num_dim_spatial, int arg_idx, char* const argv[])
-{
-    ck::utils::conv::ConvParams params;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-HostTensorDescriptor get_output_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                       int num_dim_spatial)
-{
-    namespace tl = ck::tensor_layout::convolution;
-
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::NDHWK{});
-    }
-    case 2: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::NHWK{});
-    }
-    case 1: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::NWK{});
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-HostTensorDescriptor get_filters_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                        int num_dim_spatial)
-{
-    namespace tl = ck::tensor_layout::convolution;
-
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::KZYXC{});
-    }
-    case 2: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::KYXC{});
-    }
-    case 1: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::KXC{});
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-HostTensorDescriptor get_input_host_tensor_descriptor(const std::vector<std::size_t>& dims,
-                                                      int num_dim_spatial)
-{
-    namespace tl = ck::tensor_layout::convolution;
-
-    switch(num_dim_spatial)
-    {
-    case 3: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::NDHWC{});
-    }
-    case 2: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::NHWC{});
-    }
-    case 1: {
-        return ck::utils::conv::get_host_tensor_descriptor(dims, tl::NWC{});
-    }
-    default: {
-        throw std::runtime_error("Unsupported number of spatial dimensions provided!");
-    }
-    }
-}
-
-} // namespace conv
-} // namespace utils
-} // namespace ck
-
-std::ostream& operator<<(std::ostream& os, const ck::utils::conv::ConvParams& p)
-{
-    os << "ConvParams {"
-       << "\nnum_dim_spatial: " << p.num_dim_spatial_ << "\nN: " << p.N_ << "\nK: " << p.K_
-       << "\nC: " << p.C_ << "\nfilter_spatial_lengths: " << p.filter_spatial_lengths_
-       << "\ninput_spatial_lengths: " << p.input_spatial_lengths_
-       << "\nconv_filter_strides: " << p.conv_filter_strides_
-       << "\nconv_filter_dilations: " << p.conv_filter_dilations_
-       << "\ninput_left_pads: " << p.input_left_pads_
-       << "\ninput_right_pads: " << p.input_right_pads_;
-    return os;
-}
diff --git a/library/src/utility/convolution_parameter.cpp b/library/src/utility/convolution_parameter.cpp
new file mode 100644
index 0000000000..82bb09e60c
--- /dev/null
+++ b/library/src/utility/convolution_parameter.cpp
@@ -0,0 +1,175 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "ck/host_utility/io.hpp"
+
+#include "ck/library/utility/convolution_parameter.hpp"
+
+namespace ck {
+namespace utils {
+namespace conv {
+
+ConvParam::ConvParam(ck::index_t n_dim,
+                     ck::index_t group_count,
+                     ck::index_t n_batch,
+                     ck::index_t n_out_channels,
+                     ck::index_t n_in_channels,
+                     const std::vector<ck::index_t>& filters_len,
+                     const std::vector<ck::index_t>& input_len,
+                     const std::vector<ck::index_t>& strides,
+                     const std::vector<ck::index_t>& dilations,
+                     const std::vector<ck::index_t>& left_pads,
+                     const std::vector<ck::index_t>& right_pads)
+    : num_dim_spatial_(n_dim),
+      G_(group_count),
+      N_(n_batch),
+      K_(n_out_channels),
+      C_(n_in_channels),
+      filter_spatial_lengths_(filters_len),
+      input_spatial_lengths_(input_len),
+      output_spatial_lengths_(num_dim_spatial_),
+      conv_filter_strides_(strides),
+      conv_filter_dilations_(dilations),
+      input_left_pads_(left_pads),
+      input_right_pads_(right_pads)
+{
+    if(static_cast<ck::index_t>(filter_spatial_lengths_.size()) != num_dim_spatial_ ||
+       static_cast<ck::index_t>(input_spatial_lengths_.size()) != num_dim_spatial_ ||
+       static_cast<ck::index_t>(conv_filter_strides_.size()) != num_dim_spatial_ ||
+       static_cast<ck::index_t>(conv_filter_dilations_.size()) != num_dim_spatial_ ||
+       static_cast<ck::index_t>(input_left_pads_.size()) != num_dim_spatial_ ||
+       static_cast<ck::index_t>(input_right_pads_.size()) != num_dim_spatial_)
+    {
+        throw(
+            std::runtime_error("ConvParam::ConvParam: "
+                               "parameter size is different from number of declared dimensions!"));
+    }
+
+    for(ck::index_t i = 0; i < num_dim_spatial_; ++i)
+    {
+        // XEff = (X - 1) * conv_dilation_w + 1;
+        // Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
+        const ck::index_t x_eff = (filter_spatial_lengths_[i] - 1) * conv_filter_dilations_[i] + 1;
+
+        output_spatial_lengths_[i] =
+            (input_spatial_lengths_[i] + input_left_pads_[i] + input_right_pads_[i] - x_eff) /
+                conv_filter_strides_[i] +
+            1;
+    }
+}
+
+ConvParam::ConvParam()
+    : ConvParam::ConvParam(2, 1, 128, 256, 192, {3, 3}, {71, 71}, {2, 2}, {1, 1}, {1, 1}, {1, 1})
+{
+}
+
+std::vector<ck::index_t> ConvParam::GetOutputSpatialLengths() const
+{
+    return output_spatial_lengths_;
+}
+
+std::size_t ConvParam::GetFlops() const
+{
+    // 2 * G * N * K * C * <output spatial lengths product> * <filter spatial lengths product>
+    return static_cast<std::size_t>(2) * G_ * N_ * K_ * C_ *
+           std::accumulate(std::begin(output_spatial_lengths_),
+                           std::begin(output_spatial_lengths_) + num_dim_spatial_,
+                           static_cast<std::size_t>(1),
+                           std::multiplies<std::size_t>()) *
+           std::accumulate(std::begin(filter_spatial_lengths_),
+                           std::begin(filter_spatial_lengths_) + num_dim_spatial_,
+                           static_cast<std::size_t>(1),
+                           std::multiplies<std::size_t>());
+}
+
+std::string get_conv_param_parser_helper_msg()
+{
+    std::string msg;
+
+    msg += "Following arguments (depending on number of spatial dims):\n"
+           " Number of spatial dimensions (1=Conv1d, 2=Conv2d, 3=Conv3d)\n"
+           " G, N, K, C, \n"
+           " <filter spatial dimensions>, (ie Y, X for 2D)\n"
+           " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
+           " <strides>, (ie Sy, Sx for 2D)\n"
+           " <dilations>, (ie Dy, Dx for 2D)\n"
+           " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
+           " <right padding>, (ie RightPy, RightPx for 2D)\n";
+
+    return msg;
+}
+
+ck::utils::conv::ConvParam parse_conv_param(int num_dim_spatial, int arg_idx, char* const argv[])
+{
+    const ck::index_t G = std::stoi(argv[arg_idx++]);
+    const ck::index_t N = std::stoi(argv[arg_idx++]);
+    const ck::index_t K = std::stoi(argv[arg_idx++]);
+    const ck::index_t C = std::stoi(argv[arg_idx++]);
+
+    std::vector<ck::index_t> filter_spatial_lengths(num_dim_spatial);
+    std::vector<ck::index_t> input_spatial_lengths(num_dim_spatial);
+    std::vector<ck::index_t> conv_filter_strides(num_dim_spatial);
+    std::vector<ck::index_t> conv_filter_dilations(num_dim_spatial);
+    std::vector<ck::index_t> input_left_pads(num_dim_spatial);
+    std::vector<ck::index_t> input_right_pads(num_dim_spatial);
+
+    for(int i = 0; i < num_dim_spatial; ++i)
+    {
+        filter_spatial_lengths[i] = std::stoi(argv[arg_idx++]);
+    }
+
+    for(int i = 0; i < num_dim_spatial; ++i)
+    {
+        input_spatial_lengths[i] = std::stoi(argv[arg_idx++]);
+    }
+
+    for(int i = 0; i < num_dim_spatial; ++i)
+    {
+        conv_filter_strides[i] = std::stoi(argv[arg_idx++]);
+    }
+
+    for(int i = 0; i < num_dim_spatial; ++i)
+    {
+        conv_filter_dilations[i] = std::stoi(argv[arg_idx++]);
+    }
+
+    for(int i = 0; i < num_dim_spatial; ++i)
+    {
+        input_left_pads[i] = std::stoi(argv[arg_idx++]);
+    }
+
+    for(int i = 0; i < num_dim_spatial; ++i)
+    {
+        input_right_pads[i] = std::stoi(argv[arg_idx++]);
+    }
+
+    return ck::utils::conv::ConvParam{num_dim_spatial,
+                                      G,
+                                      N,
+                                      K,
+                                      C,
+                                      filter_spatial_lengths,
+                                      input_spatial_lengths,
+                                      conv_filter_strides,
+                                      conv_filter_dilations,
+                                      input_left_pads,
+                                      input_right_pads};
+}
+} // namespace conv
+} // namespace utils
+} // namespace ck
+
+std::ostream& operator<<(std::ostream& os, const ck::utils::conv::ConvParam& p)
+{
+    os << "ConvParam {"
+       << "\nnum_dim_spatial: " << p.num_dim_spatial_ << "\nG: " << p.G_ << "\nN: " << p.N_
+       << "\nK: " << p.K_ << "\nC: " << p.C_
+       << "\nfilter_spatial_lengths: " << p.filter_spatial_lengths_
+       << "\ninput_spatial_lengths: " << p.input_spatial_lengths_
+       << "\nconv_filter_strides: " << p.conv_filter_strides_
+       << "\nconv_filter_dilations: " << p.conv_filter_dilations_
+       << "\ninput_left_pads: " << p.input_left_pads_
+       << "\ninput_right_pads: " << p.input_right_pads_ << "}\n";
+
+    return os;
+}
diff --git a/library/src/host_tensor/device_memory.cpp b/library/src/utility/device_memory.cpp
similarity index 88%
rename from library/src/host_tensor/device_memory.cpp
rename to library/src/utility/device_memory.cpp
index 5e7157e4e0..99d5248706 100644
--- a/library/src/host_tensor/device_memory.cpp
+++ b/library/src/utility/device_memory.cpp
@@ -1,8 +1,9 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include "ck/device_utility/hip_check_error.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
+#include "ck/host_utility/hip_check_error.hpp"
+
+#include "ck/library/utility/device_memory.hpp"
 
 DeviceMem::DeviceMem(std::size_t mem_size) : mMemSize(mem_size)
 {
diff --git a/library/src/host_tensor/host_tensor.cpp b/library/src/utility/host_tensor.cpp
similarity index 92%
rename from library/src/host_tensor/host_tensor.cpp
rename to library/src/utility/host_tensor.cpp
index dc9f5699dc..24f8224bef 100644
--- a/library/src/host_tensor/host_tensor.cpp
+++ b/library/src/utility/host_tensor.cpp
@@ -3,7 +3,7 @@
 
 #include <cassert>
 
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
 
 void HostTensorDescriptor::CalculateStrides()
 {
@@ -26,7 +26,7 @@ std::size_t HostTensorDescriptor::GetElementSize() const
         mLens.begin(), mLens.end(), std::size_t{1}, std::multiplies<std::size_t>());
 }
 
-std::size_t HostTensorDescriptor::GetElementSpace() const
+std::size_t HostTensorDescriptor::GetElementSpaceSize() const
 {
     std::size_t space = 1;
     for(std::size_t i = 0; i < mLens.size(); ++i)
diff --git a/profiler/CMakeLists.txt b/profiler/CMakeLists.txt
index eca6a0171f..274cfd5f21 100644
--- a/profiler/CMakeLists.txt
+++ b/profiler/CMakeLists.txt
@@ -14,20 +14,19 @@ set(PROFILER_SOURCE
     src/profile_batched_gemm.cpp
     src/profile_batched_gemm_reduce.cpp
     src/profile_grouped_gemm.cpp
+    src/profile_conv_fwd.cpp
     src/profile_conv_fwd_bias_relu.cpp
     src/profile_conv_fwd_bias_relu_add.cpp
-    src/profile_convnd_fwd.cpp
-    src/profile_convnd_bwd_data.cpp
+    src/profile_conv_bwd_data.cpp
     src/profile_conv_bwd_weight.cpp
-    src/profile_convnd_bwd_weight.cpp
+    src/profile_grouped_conv_fwd.cpp
     src/profile_reduce.cpp
     src/profile_normalization.cpp
 )
 
 add_executable(ckProfiler ${PROFILER_SOURCE})
 
-target_link_libraries(ckProfiler PRIVATE host_tensor)
-target_link_libraries(ckProfiler PRIVATE conv_util)
+target_link_libraries(ckProfiler PRIVATE utility)
 target_link_libraries(ckProfiler PRIVATE device_gemm_instance)
 target_link_libraries(ckProfiler PRIVATE device_gemm_splitk_instance)
 target_link_libraries(ckProfiler PRIVATE device_gemm_bilinear_instance)
@@ -37,13 +36,17 @@ target_link_libraries(ckProfiler PRIVATE device_gemm_bias_add_reduce_instance)
 target_link_libraries(ckProfiler PRIVATE device_batched_gemm_instance)
 target_link_libraries(ckProfiler PRIVATE device_batched_gemm_reduce_instance)
 target_link_libraries(ckProfiler PRIVATE device_grouped_gemm_instance)
-target_link_libraries(ckProfiler PRIVATE device_conv1d_fwd_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_fwd_instance)
-target_link_libraries(ckProfiler PRIVATE device_conv3d_fwd_instance)
+target_link_libraries(ckProfiler PRIVATE device_grouped_conv1d_fwd_instance)
+target_link_libraries(ckProfiler PRIVATE device_grouped_conv2d_fwd_instance)
+target_link_libraries(ckProfiler PRIVATE device_grouped_conv3d_fwd_instance)
+target_link_libraries(ckProfiler PRIVATE device_conv1d_bwd_data_instance)
+target_link_libraries(ckProfiler PRIVATE device_conv2d_bwd_data_instance)
+target_link_libraries(ckProfiler PRIVATE device_conv3d_bwd_data_instance)
+target_link_libraries(ckProfiler PRIVATE device_conv1d_bwd_weight_instance)
+target_link_libraries(ckProfiler PRIVATE device_conv2d_bwd_weight_instance)
+target_link_libraries(ckProfiler PRIVATE device_conv3d_bwd_weight_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_fwd_bias_relu_instance)
 target_link_libraries(ckProfiler PRIVATE device_conv2d_fwd_bias_relu_add_instance)
-target_link_libraries(ckProfiler PRIVATE device_convnd_bwd_data_instance)
-target_link_libraries(ckProfiler PRIVATE device_conv2d_bwd_weight_instance)
-target_link_libraries(ckProfiler PRIVATE device_convnd_bwd_weight_instance)
 target_link_libraries(ckProfiler PRIVATE device_normalization_instance)
 target_link_libraries(ckProfiler PRIVATE device_reduce_instance)
diff --git a/profiler/include/profile_batched_gemm_impl.hpp b/profiler/include/profile_batched_gemm_impl.hpp
index 0da9a26cf5..d50710a3c1 100644
--- a/profiler/include/profile_batched_gemm_impl.hpp
+++ b/profiler/include/profile_batched_gemm_impl.hpp
@@ -13,9 +13,9 @@
 #include "ck/library/tensor_operation_instance/gpu/batched_gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp"
 
 namespace ck {
@@ -114,9 +114,9 @@ bool profile_batched_gemm_impl(int do_verification,
         ref_invoker.Run(ref_argument);
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_g_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_g_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_g_m_k.mData.data());
     b_device_buf.ToDevice(b_g_k_n.mData.data());
diff --git a/profiler/include/profile_batched_gemm_reduce_impl.hpp b/profiler/include/profile_batched_gemm_reduce_impl.hpp
index d1a989348a..5f1aa0a980 100644
--- a/profiler/include/profile_batched_gemm_reduce_impl.hpp
+++ b/profiler/include/profile_batched_gemm_reduce_impl.hpp
@@ -10,10 +10,10 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp"
 
 namespace ck {
@@ -193,13 +193,13 @@ bool profile_batched_gemm_reduce_impl(int do_verification,
         }
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_g_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_g_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_g_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_g_m_n_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce0_device_buf(sizeof(ReduceDataType) *
-                                 d0_g_m_device_result.mDesc.GetElementSpace());
+                                 d0_g_m_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce1_device_buf(sizeof(ReduceDataType) *
-                                 d1_g_m_device_result.mDesc.GetElementSpace());
+                                 d1_g_m_device_result.mDesc.GetElementSpaceSize());
 
     std::array<void*, 2> p_reduces = {reduce0_device_buf.GetDeviceBuffer(),
                                       reduce1_device_buf.GetDeviceBuffer()};
@@ -319,11 +319,11 @@ bool profile_batched_gemm_reduce_impl(int do_verification,
                 reduce1_device_buf.FromDevice(d1_g_m_device_result.mData.data());
 
                 bool c_error =
-                    ck::utils::check_err(c_g_m_n_host_result.mData, c_g_m_n_device_result.mData);
+                    ck::utils::check_err(c_g_m_n_device_result.mData, c_g_m_n_host_result.mData);
                 bool d0_error =
-                    ck::utils::check_err(d0_g_m_host_result.mData, d0_g_m_device_result.mData);
+                    ck::utils::check_err(d0_g_m_device_result.mData, d0_g_m_host_result.mData);
                 bool d1_error =
-                    ck::utils::check_err(d1_g_m_host_result.mData, d1_g_m_device_result.mData);
+                    ck::utils::check_err(d1_g_m_device_result.mData, d1_g_m_host_result.mData);
 
                 pass = pass && (c_error == true);
                 pass = pass && (d0_error == true);
diff --git a/profiler/include/profile_conv_bwd_data_impl.hpp b/profiler/include/profile_conv_bwd_data_impl.hpp
new file mode 100644
index 0000000000..b0243e1b25
--- /dev/null
+++ b/profiler/include/profile_conv_bwd_data_impl.hpp
@@ -0,0 +1,249 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_bwd_data.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp"
+
+namespace ck {
+namespace profiler {
+
+template <typename DataType>
+void show_data_nhwc_layout(Tensor<DataType>& nhwc)
+{
+    std::cout << "[";
+    for(int n = 0; n < ck::type_convert<int>(nhwc.mDesc.GetLengths()[0]); n++)
+    {
+        std::cout << "[";
+        for(int hi = 0; hi < ck::type_convert<int>(nhwc.mDesc.GetLengths()[2]); hi++)
+        {
+            std::cout << "[";
+            for(int wi = 0; wi < ck::type_convert<int>(nhwc.mDesc.GetLengths()[3]); wi++)
+            {
+                std::cout << "[";
+                for(int c = 0; c < ck::type_convert<int>(nhwc.mDesc.GetLengths()[1]); c++)
+                {
+                    std::cout << static_cast<float>(nhwc(n, c, hi, wi)) << "  ";
+                }
+                std::cout << "]";
+            }
+            std::cout << "]";
+        }
+        std::cout << "]";
+    }
+    std::cout << "]";
+}
+
+template <ck::index_t NDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+bool profile_conv_bwd_data_impl(int do_verification,
+                                int init_method,
+                                bool do_log,
+                                bool time_kernel,
+                                const ck::utils::conv::ConvParam& conv_param)
+{
+    using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+    using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+    using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    const auto in_g_n_c_wis_desc =
+        ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);
+
+    const auto wei_g_k_c_xs_desc =
+        ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(conv_param);
+
+    const auto out_g_n_k_wos_desc =
+        ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(conv_param);
+
+    Tensor<InDataType> input_host_result(in_g_n_c_wis_desc);
+    Tensor<InDataType> input_device_result(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> weight(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> output(out_g_n_k_wos_desc);
+
+    std::cout << "input: " << input_host_result.mDesc << std::endl;
+    std::cout << "weight: " << weight.mDesc << std::endl;
+    std::cout << "output: " << output.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        output.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
+        weight.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        break;
+    default:
+        output.GenerateTensorValue(GeneratorTensor_3<OutDataType>{0.0, 1.0});
+        weight.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * input_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * weight.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * output.mDesc.GetElementSpaceSize());
+
+    out_device_buf.ToDevice(output.mData.data());
+    wei_device_buf.ToDevice(weight.mData.data());
+
+    if(do_verification)
+    {
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvBwdData<NDimSpatial,
+                                                                         InDataType,
+                                                                         WeiDataType,
+                                                                         OutDataType,
+                                                                         InElementOp,
+                                                                         WeiElementOp,
+                                                                         OutElementOp>{};
+
+        auto ref_invoker = ref_conv.MakeInvoker();
+
+        auto ref_argument = ref_conv.MakeArgument(input_host_result,
+                                                  weight,
+                                                  output,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  InElementOp{},
+                                                  WeiElementOp{},
+                                                  OutElementOp{});
+        ref_invoker.Run(ref_argument);
+    }
+
+    using DeviceOp = ck::tensor_operation::device::DeviceConvBwdData<NDimSpatial,
+                                                                     InLayout,
+                                                                     WeiLayout,
+                                                                     OutLayout,
+                                                                     InDataType,
+                                                                     WeiDataType,
+                                                                     OutDataType,
+                                                                     InElementOp,
+                                                                     WeiElementOp,
+                                                                     OutElementOp>;
+
+    // get device op instances
+    const auto op_ptrs = ck::tensor_operation::device::instance::DeviceOperationInstanceFactory<
+        DeviceOp>::GetInstances();
+
+    std::cout << "found " << op_ptrs.size() << " instances" << std::endl;
+
+    std::string best_op_name;
+    float best_avg_time   = 0;
+    float best_tflops     = 0;
+    float best_gb_per_sec = 0;
+
+    // profile device Conv instances
+    bool pass = true;
+
+    for(auto& op_ptr : op_ptrs)
+    {
+        auto argument_ptr =
+            op_ptr->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
+                                        static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
+                                        static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
+                                        conv_param.N_,
+                                        conv_param.K_,
+                                        conv_param.C_,
+                                        conv_param.input_spatial_lengths_,
+                                        conv_param.filter_spatial_lengths_,
+                                        conv_param.output_spatial_lengths_,
+                                        conv_param.conv_filter_strides_,
+                                        conv_param.conv_filter_dilations_,
+                                        conv_param.input_left_pads_,
+                                        conv_param.input_right_pads_,
+                                        in_element_op,
+                                        wei_element_op,
+                                        out_element_op);
+
+        if(op_ptr->IsSupportedArgument(argument_ptr.get()))
+        {
+            // for conv bwd data, some input tensor element are zero, but not written by kernel,
+            // need to set zero
+            in_device_buf.SetZero();
+
+            std::string op_name = op_ptr->GetTypeString();
+
+            auto invoker_ptr = op_ptr->MakeInvokerPointer();
+
+            float avg_time =
+                invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, time_kernel});
+
+            std::size_t flop      = conv_param.GetFlops();
+            std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+            float tflops     = static_cast<float>(flop) / 1.E9 / avg_time;
+            float gb_per_sec = num_btype / 1.E6 / avg_time;
+
+            std::cout << "Perf: " << avg_time << " ms, " << tflops << " TFlops, " << gb_per_sec
+                      << " GB/s" << std::endl;
+
+            if(tflops > best_tflops)
+            {
+                best_op_name    = op_name;
+                best_tflops     = tflops;
+                best_avg_time   = avg_time;
+                best_gb_per_sec = gb_per_sec;
+            }
+
+            if(do_verification)
+            {
+                in_device_buf.FromDevice(input_device_result.mData.data());
+
+                pass =
+                    pass & ck::utils::check_err(input_device_result.mData, input_host_result.mData);
+
+                if(do_log)
+                {
+                    std::cout << "in : ";
+                    show_data_nhwc_layout(output);
+                    std::cout << std::endl;
+
+                    std::cout << "wei: ";
+                    show_data_nhwc_layout(weight);
+                    std::cout << std::endl;
+
+                    std::cout << "out_host  : ";
+                    show_data_nhwc_layout(input_host_result);
+                    std::cout << std::endl;
+
+                    std::cout << "out_device: ";
+                    show_data_nhwc_layout(input_device_result);
+                    std::cout << std::endl;
+                }
+            }
+        }
+        else
+        {
+            std::cout << op_ptr->GetTypeString() << " does not support this problem" << std::endl;
+        }
+    }
+
+    std::cout << "Best configuration parameters:"
+              << "\nname: " << best_op_name << "\navg_time: " << best_avg_time
+              << "\ntflops: " << best_tflops << "\nGB/s: " << best_gb_per_sec << std::endl;
+
+    return pass;
+}
+
+} // namespace profiler
+} // namespace ck
diff --git a/profiler/include/profile_conv_bwd_weight_impl.hpp b/profiler/include/profile_conv_bwd_weight_impl.hpp
index c677eb3538..7712ad3363 100644
--- a/profiler/include/profile_conv_bwd_weight_impl.hpp
+++ b/profiler/include/profile_conv_bwd_weight_impl.hpp
@@ -3,114 +3,69 @@
 
 #pragma once
 
+#include "ck/ck.hpp"
+#include <iomanip>
+#include <iostream>
+#include <typeinfo>
+
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_conv_backward_weight.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
+#include "ck/library/tensor_operation_instance/gpu/convolution_backward_weight.hpp"
+
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
-#include "ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using DeviceConvBwdWeightNoOpPtr =
-    DeviceConvBwdWeightPtr<ck::tensor_operation::element_wise::PassThrough,
-                           ck::tensor_operation::element_wise::PassThrough,
-                           ck::tensor_operation::element_wise::PassThrough>;
-
-void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvBwdWeightNoOpPtr>&);
-
-void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvBwdWeightNoOpPtr>&);
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_bwd_weight.hpp"
 
 namespace ck {
 namespace profiler {
 
-template <int NDimSpatial,
-          typename InDataType,
-          typename WeiDataType,
-          typename OutDataType,
+template <typename DataType>
+void show_data_nhwc_layout(Tensor<DataType>& nhwc)
+{
+    std::cout << "[";
+    for(int n = 0; n < ck::type_convert<int>(nhwc.mDesc.GetLengths()[0]); n++)
+    {
+        std::cout << "[";
+        for(int hi = 0; hi < ck::type_convert<int>(nhwc.mDesc.GetLengths()[2]); hi++)
+        {
+            std::cout << "[";
+            for(int wi = 0; wi < ck::type_convert<int>(nhwc.mDesc.GetLengths()[3]); wi++)
+            {
+                std::cout << "[";
+                for(int c = 0; c < ck::type_convert<int>(nhwc.mDesc.GetLengths()[1]); c++)
+                {
+                    std::cout << static_cast<float>(nhwc(n, c, hi, wi)) << "  ";
+                }
+                std::cout << "]";
+            }
+            std::cout << "]";
+        }
+        std::cout << "]";
+    }
+    std::cout << "]";
+}
+
+template <ck::index_t NDimSpatial,
           typename InLayout,
           typename WeiLayout,
-          typename OutLayout>
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
 bool profile_conv_bwd_weight_impl(int do_verification,
                                   int init_method,
                                   bool do_log,
                                   bool time_kernel,
-                                  ck::index_t N,
-                                  ck::index_t K,
-                                  ck::index_t C,
-                                  std::vector<ck::index_t> input_spatial_lengths,
-                                  std::vector<ck::index_t> filter_spatial_lengths,
-                                  std::vector<ck::index_t> output_spatial_lengths,
-                                  std::vector<ck::index_t> conv_filter_strides,
-                                  std::vector<ck::index_t> conv_filter_dilations,
-                                  std::vector<ck::index_t> input_left_pads,
-                                  std::vector<ck::index_t> input_right_pads,
+                                  const ck::utils::conv::ConvParam& conv_param,
                                   ck::index_t split_k)
 {
-    const ck::index_t Y = filter_spatial_lengths[0];
-    const ck::index_t X = filter_spatial_lengths[1];
-
-    const ck::index_t Hi = input_spatial_lengths[0];
-    const ck::index_t Wi = input_spatial_lengths[1];
-
-    const ck::index_t Ho = output_spatial_lengths[0];
-    const ck::index_t Wo = output_spatial_lengths[1];
-
-    auto f_host_tensor_descriptor =
-        [](std::size_t N_, std::size_t C_, std::size_t H, std::size_t W, auto layout) {
-            if constexpr(is_same<decltype(layout), ck::tensor_layout::convolution::NCHW>::value ||
-                         is_same<decltype(layout), ck::tensor_layout::convolution::KCYX>::value ||
-                         is_same<decltype(layout), ck::tensor_layout::convolution::NKHW>::value)
-            {
-                return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
-                                            std::vector<std::size_t>({C_ * H * W, H * W, W, 1}));
-            }
-            else if constexpr(is_same<decltype(layout), tensor_layout::convolution::NHWC>::value ||
-                              is_same<decltype(layout), tensor_layout::convolution::KYXC>::value ||
-                              is_same<decltype(layout), tensor_layout::convolution::NHWK>::value)
-            {
-                return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
-                                            std::vector<std::size_t>({C_ * H * W, 1, W * C_, C_}));
-            }
-        };
-
-    Tensor<InDataType> in_n_c_hi_wi(f_host_tensor_descriptor(N, C, Hi, Wi, InLayout{}));
-    Tensor<WeiDataType> wei_k_c_y_x_host_result(f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
-    Tensor<WeiDataType> wei_k_c_y_x_device_result(
-        f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
-    Tensor<OutDataType> out_n_k_ho_wo(f_host_tensor_descriptor(N, K, Ho, Wo, OutLayout{}));
-
-    std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
-    std::cout << "wei_k_c_y_x: " << wei_k_c_y_x_host_result.mDesc << std::endl;
-    std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-    switch(init_method)
-    {
-    case 0: break;
-    case 1:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        break;
-    default:
-        out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-        in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_1<InDataType>{1});
-    }
-
     using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
     using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
     using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
@@ -119,25 +74,63 @@ bool profile_conv_bwd_weight_impl(int do_verification,
     const auto wei_element_op = WeiElementOp{};
     const auto out_element_op = OutElementOp{};
 
+    const auto in_g_n_c_wis_desc =
+        ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);
+
+    const auto wei_g_k_c_xs_desc =
+        ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(conv_param);
+
+    const auto out_g_n_k_wos_desc =
+        ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(conv_param);
+
+    Tensor<InDataType> input(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> weight_host_result(wei_g_k_c_xs_desc);
+    Tensor<WeiDataType> weight_device_result(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> output(out_g_n_k_wos_desc);
+
+    std::cout << "input: " << input.mDesc << std::endl;
+    std::cout << "weight: " << weight_host_result.mDesc << std::endl;
+    std::cout << "output: " << output.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
+        output.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
+        break;
+    default:
+        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
+        output.GenerateTensorValue(GeneratorTensor_3<OutDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) *
+                             weight_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * output.mDesc.GetElementSpaceSize());
+
+    in_device_buf.ToDevice(input.mData.data());
+    out_device_buf.ToDevice(output.mData.data());
+
     if(do_verification)
     {
-        using ReferenceConvBwdWeightInstance =
-            ck::tensor_operation::host::ReferenceConvBwdWeight<InDataType,
-                                                               WeiDataType,
-                                                               OutDataType,
-                                                               InElementOp,
-                                                               WeiElementOp,
-                                                               OutElementOp>;
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvBwdWeight<NDimSpatial,
+                                                                           InDataType,
+                                                                           WeiDataType,
+                                                                           OutDataType,
+                                                                           InElementOp,
+                                                                           WeiElementOp,
+                                                                           OutElementOp>{};
 
-        auto ref_conv     = ReferenceConvBwdWeightInstance{};
-        auto ref_invoker  = ref_conv.MakeInvoker();
-        auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi,
-                                                  wei_k_c_y_x_host_result,
-                                                  out_n_k_ho_wo,
-                                                  conv_filter_strides,
-                                                  conv_filter_dilations,
-                                                  input_left_pads,
-                                                  input_right_pads,
+        auto ref_invoker = ref_conv.MakeInvoker();
+
+        auto ref_argument = ref_conv.MakeArgument(input,
+                                                  weight_host_result,
+                                                  output,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
                                                   in_element_op,
                                                   wei_element_op,
                                                   out_element_op);
@@ -145,140 +138,126 @@ bool profile_conv_bwd_weight_impl(int do_verification,
         ref_invoker.Run(ref_argument);
     }
 
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) *
-                             wei_k_c_y_x_device_result.mDesc.GetElementSpace());
-    DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
+    using DeviceOp = ck::tensor_operation::device::DeviceConvBwdWeight<NDimSpatial,
+                                                                       InLayout,
+                                                                       WeiLayout,
+                                                                       OutLayout,
+                                                                       InDataType,
+                                                                       WeiDataType,
+                                                                       OutDataType,
+                                                                       InElementOp,
+                                                                       WeiElementOp,
+                                                                       OutElementOp>;
 
-    out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-    in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
+    // get device op instances
+    const auto op_ptrs = ck::tensor_operation::device::instance::DeviceOperationInstanceFactory<
+        DeviceOp>::GetInstances();
 
-    using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+    std::cout << "found " << op_ptrs.size() << " instances" << std::endl;
 
-    using DeviceConvBwdWeightNoOpPtr =
-        ck::tensor_operation::device::DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>;
-
-    // add device Conv instances
-    std::vector<DeviceConvBwdWeightNoOpPtr> conv_ptrs;
-
-    if constexpr(ck::is_same_v<ck::remove_cv_t<InDataType>, float> &&
-                 ck::is_same_v<ck::remove_cv_t<WeiDataType>, float> &&
-                 ck::is_same_v<ck::remove_cv_t<OutDataType>, float>)
-    {
-        ck::tensor_operation::device::instance::
-            add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(conv_ptrs);
-    }
-    else if constexpr(ck::is_same_v<ck::remove_cv_t<InDataType>, ck::half_t> &&
-                      ck::is_same_v<ck::remove_cv_t<WeiDataType>, ck::half_t> &&
-                      ck::is_same_v<ck::remove_cv_t<OutDataType>, ck::half_t>)
-    {
-        ck::tensor_operation::device::instance::
-            add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(conv_ptrs);
-    }
-
-    if(conv_ptrs.size() <= 0)
-    {
-        throw std::runtime_error("wrong! no device Conv instance found");
-    }
-
-    std::string best_conv_name;
-    float best_ave_time   = 0;
+    std::string best_op_name;
+    float best_avg_time   = 0;
     float best_tflops     = 0;
     float best_gb_per_sec = 0;
 
     // profile device Conv instances
-    bool pass = true;
+    bool all_pass = true;
 
-    for(auto& conv_ptr : conv_ptrs)
+    for(auto& op_ptr : op_ptrs)
     {
-        // using atomic, so need to reset input
-        if(split_k > 1)
+        auto argument_ptr =
+            op_ptr->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
+                                        static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
+                                        static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
+                                        conv_param.N_,
+                                        conv_param.K_,
+                                        conv_param.C_,
+                                        conv_param.input_spatial_lengths_,
+                                        conv_param.filter_spatial_lengths_,
+                                        conv_param.output_spatial_lengths_,
+                                        conv_param.conv_filter_strides_,
+                                        conv_param.conv_filter_dilations_,
+                                        conv_param.input_left_pads_,
+                                        conv_param.input_right_pads_,
+                                        in_element_op,
+                                        wei_element_op,
+                                        out_element_op,
+                                        split_k);
+
+        if(op_ptr->IsSupportedArgument(argument_ptr.get()))
         {
+            // using atomic add, so need to reset input
             wei_device_buf.SetZero();
-        }
 
-        auto argument_ptr = conv_ptr->MakeArgumentPointer(
-            static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-            static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-            static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-            N,
-            K,
-            C,
-            input_spatial_lengths,
-            filter_spatial_lengths,
-            output_spatial_lengths,
-            conv_filter_strides,
-            conv_filter_dilations,
-            input_left_pads,
-            input_right_pads,
-            in_element_op,
-            wei_element_op,
-            out_element_op,
-            split_k);
+            std::string op_name = op_ptr->GetTypeString();
 
-        auto invoker_ptr = conv_ptr->MakeInvokerPointer();
+            auto invoker_ptr = op_ptr->MakeInvokerPointer();
 
-        if(conv_ptr->IsSupportedArgument(argument_ptr.get()))
-        {
-            std::string conv_name = conv_ptr->GetTypeString();
-
-            float ave_time =
+            float avg_time =
                 invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, time_kernel});
 
-            std::size_t flop = std::size_t(2) * N * K * Ho * Wo * C * Y * X;
+            std::size_t flop      = conv_param.GetFlops();
+            std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
 
-            std::size_t num_btype = sizeof(InDataType) * (N * C * Hi * Wi) +
-                                    sizeof(WeiDataType) * (K * C * Y * X) +
-                                    sizeof(OutDataType) * (N * K * Ho * Wo);
+            float tflops     = static_cast<float>(flop) / 1.E9 / avg_time;
+            float gb_per_sec = num_btype / 1.E6 / avg_time;
 
-            float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
-
-            float gb_per_sec = num_btype / 1.E6 / ave_time;
-
-            std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec
-                      << " GB/s, " << conv_name << std::endl;
+            std::cout << "Perf: " << std::setw(10) << avg_time << " ms, " << tflops << " TFlops, "
+                      << gb_per_sec << " GB/s, " << op_name << std::endl;
 
             if(tflops > best_tflops)
             {
-                best_conv_name  = conv_name;
+                best_op_name    = op_name;
                 best_tflops     = tflops;
-                best_ave_time   = ave_time;
+                best_avg_time   = avg_time;
                 best_gb_per_sec = gb_per_sec;
             }
 
             if(do_verification)
             {
-                wei_device_buf.FromDevice(wei_k_c_y_x_device_result.mData.data());
+                wei_device_buf.FromDevice(weight_device_result.mData.data());
 
-                pass = ck::utils::check_err(wei_k_c_y_x_host_result.mData,
-                                            wei_k_c_y_x_device_result.mData);
+                bool pass =
+                    ck::utils::check_err(weight_host_result.mData, weight_device_result.mData);
 
-                if(pass == false)
+                if(!pass)
                 {
-                    std::cout << "Fail info:" << conv_ptr->GetTypeString() << std::endl;
+                    std::cout << "Fail info:" << op_ptr->GetTypeString() << std::endl;
                 }
 
+                all_pass &= pass;
+
                 if(do_log)
                 {
-                    LogRangeAsType<float>(std::cout << "out: ", out_n_k_ho_wo.mData, ",")
-                        << std::endl;
-                    LogRangeAsType<float>(std::cout << "in : ", in_n_c_hi_wi.mData, ",")
-                        << std::endl;
-                    LogRangeAsType<float>(
-                        std::cout << "wei_host  : ", wei_k_c_y_x_host_result.mData, ",")
-                        << std::endl;
-                    LogRangeAsType<float>(
-                        std::cout << "wei_device: ", wei_k_c_y_x_device_result.mData, ",")
-                        << std::endl;
+                    std::cout << "in : ";
+                    show_data_nhwc_layout(output);
+                    std::cout << std::endl;
+
+                    std::cout << "wei: ";
+                    show_data_nhwc_layout(weight_host_result);
+                    std::cout << std::endl;
+
+                    std::cout << "out  : ";
+                    show_data_nhwc_layout(input);
+                    std::cout << std::endl;
+
+                    std::cout << "wei_device: ";
+                    show_data_nhwc_layout(weight_device_result);
+                    std::cout << std::endl;
                 }
             }
         }
+        else
+        {
+            std::cout << op_ptr->GetTypeString() << " does not support this problem" << std::endl;
+        }
     }
 
-    std::cout << "Best Perf: " << best_ave_time << " ms, " << best_tflops << " TFlops, "
-              << best_gb_per_sec << " GB/s, " << best_conv_name << std::endl;
+    std::cout << "Best configuration parameters:"
+              << "\nname: " << best_op_name << "\navg_time: " << best_avg_time
+              << "\ntflops: " << best_tflops << "\nGB/s: " << best_gb_per_sec << std::endl;
 
-    return pass;
+    return all_pass;
 }
 
 } // namespace profiler
diff --git a/profiler/include/profile_conv_fwd_bias_relu_add_impl.hpp b/profiler/include/profile_conv_fwd_bias_relu_add_impl.hpp
index 69bfe50a70..aad48946c8 100644
--- a/profiler/include/profile_conv_fwd_bias_relu_add_impl.hpp
+++ b/profiler/include/profile_conv_fwd_bias_relu_add_impl.hpp
@@ -9,9 +9,9 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp"
 
 namespace ck {
@@ -157,12 +157,12 @@ void profile_conv_fwd_bias_relu_add_impl(int do_verification,
         ref_invoker.Run(ref_argument);
     }
 
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpace());
+    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpaceSize());
     DeviceMem out_device_buf(sizeof(OutDataType) *
-                             out_n_k_ho_wo_device_result.mDesc.GetElementSpace());
-    DeviceMem bias_device_buf(sizeof(OutDataType) * bias_k.mDesc.GetElementSpace());
-    DeviceMem resi_device_buf(sizeof(OutDataType) * resi_n_k_ho_wo.mDesc.GetElementSpace());
+                             out_n_k_ho_wo_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem bias_device_buf(sizeof(OutDataType) * bias_k.mDesc.GetElementSpaceSize());
+    DeviceMem resi_device_buf(sizeof(OutDataType) * resi_n_k_ho_wo.mDesc.GetElementSpaceSize());
 
     in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
     wei_device_buf.ToDevice(wei_k_c_y_x.mData.data());
diff --git a/profiler/include/profile_conv_fwd_bias_relu_impl.hpp b/profiler/include/profile_conv_fwd_bias_relu_impl.hpp
index 166173ca89..f546606d67 100644
--- a/profiler/include/profile_conv_fwd_bias_relu_impl.hpp
+++ b/profiler/include/profile_conv_fwd_bias_relu_impl.hpp
@@ -9,9 +9,9 @@
 #include "ck/tensor_operation/gpu/device/device_conv_fwd_bias_activation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp"
 
 namespace ck {
@@ -149,11 +149,11 @@ void profile_conv_fwd_bias_relu_impl(int do_verification,
         ref_invoker.Run(ref_argument);
     }
 
-    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
-    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpace());
+    DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpaceSize());
     DeviceMem out_device_buf(sizeof(OutDataType) *
-                             out_n_k_ho_wo_device_result.mDesc.GetElementSpace());
-    DeviceMem bias_device_buf(sizeof(OutDataType) * bias_k.mDesc.GetElementSpace());
+                             out_n_k_ho_wo_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem bias_device_buf(sizeof(OutDataType) * bias_k.mDesc.GetElementSpaceSize());
 
     in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
     wei_device_buf.ToDevice(wei_k_c_y_x.mData.data());
diff --git a/profiler/include/profile_conv_fwd_impl.hpp b/profiler/include/profile_conv_fwd_impl.hpp
new file mode 100644
index 0000000000..4a91fede02
--- /dev/null
+++ b/profiler/include/profile_conv_fwd_impl.hpp
@@ -0,0 +1,221 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <iomanip>
+#include <iostream>
+#include <typeinfo>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_fwd.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/gpu/convolution_forward.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+
+namespace ck {
+namespace profiler {
+
+template <ck::index_t NDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+bool profile_conv_fwd_impl(int do_verification,
+                           int init_method,
+                           bool do_log,
+                           bool time_kernel,
+                           const ck::utils::conv::ConvParam& conv_param)
+{
+    using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+    using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+    using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    const auto in_g_n_c_wis_desc =
+        ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);
+
+    const auto wei_g_k_c_xs_desc =
+        ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(conv_param);
+
+    const auto out_g_n_k_wos_desc =
+        ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(conv_param);
+
+    Tensor<InDataType> input(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> weight(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> host_output(out_g_n_k_wos_desc);
+    Tensor<OutDataType> device_output(out_g_n_k_wos_desc);
+
+    std::cout << "input: " << input.mDesc << std::endl;
+    std::cout << "weight: " << weight.mDesc << std::endl;
+    std::cout << "output: " << host_output.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
+        weight.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        break;
+    default:
+        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
+        weight.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * weight.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpaceSize());
+
+    in_device_buf.ToDevice(input.mData.data());
+    wei_device_buf.ToDevice(weight.mData.data());
+
+    // run reference op
+    if(do_verification)
+    {
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvFwd<NDimSpatial,
+                                                                     InDataType,
+                                                                     WeiDataType,
+                                                                     OutDataType,
+                                                                     InElementOp,
+                                                                     WeiElementOp,
+                                                                     OutElementOp>{};
+
+        auto ref_invoker  = ref_conv.MakeInvoker();
+        auto ref_argument = ref_conv.MakeArgument(input,
+                                                  weight,
+                                                  host_output,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  in_element_op,
+                                                  wei_element_op,
+                                                  out_element_op);
+
+        // init host output to zero
+        host_output.SetZero();
+
+        ref_invoker.Run(ref_argument);
+    }
+
+    using DeviceOp = ck::tensor_operation::device::DeviceConvFwd<NDimSpatial,
+                                                                 InLayout,
+                                                                 WeiLayout,
+                                                                 OutLayout,
+                                                                 InDataType,
+                                                                 WeiDataType,
+                                                                 OutDataType,
+                                                                 InElementOp,
+                                                                 WeiElementOp,
+                                                                 OutElementOp>;
+
+    // get device op instances
+    const auto op_ptrs = ck::tensor_operation::device::instance::DeviceOperationInstanceFactory<
+        DeviceOp>::GetInstances();
+
+    std::cout << "found " << op_ptrs.size() << " instances" << std::endl;
+
+    std::string best_op_name;
+    float best_avg_time   = 0;
+    float best_tflops     = 0;
+    float best_gb_per_sec = 0;
+
+    // profile device op instances
+    bool pass = true;
+
+    for(auto& op_ptr : op_ptrs)
+    {
+        auto argument_ptr =
+            op_ptr->MakeArgumentPointer(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
+                                        static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
+                                        static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
+                                        conv_param.N_,
+                                        conv_param.K_,
+                                        conv_param.C_,
+                                        conv_param.input_spatial_lengths_,
+                                        conv_param.filter_spatial_lengths_,
+                                        conv_param.GetOutputSpatialLengths(),
+                                        conv_param.conv_filter_strides_,
+                                        conv_param.conv_filter_dilations_,
+                                        conv_param.input_left_pads_,
+                                        conv_param.input_right_pads_,
+                                        in_element_op,
+                                        wei_element_op,
+                                        out_element_op);
+
+        if(op_ptr->IsSupportedArgument(argument_ptr.get()))
+        {
+            // re-init output to zero before profiling next kernel
+            out_device_buf.SetZero();
+
+            std::string op_name = op_ptr->GetTypeString();
+
+            auto invoker_ptr = op_ptr->MakeInvokerPointer();
+
+            float avg_time =
+                invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, time_kernel});
+
+            std::size_t flop      = conv_param.GetFlops();
+            std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+            float tflops = static_cast<float>(flop) / 1.E9 / avg_time;
+
+            float gb_per_sec = num_btype / 1.E6 / avg_time;
+
+            std::cout << "Perf: " << std::setw(10) << avg_time << " ms, " << tflops << " TFlops, "
+                      << gb_per_sec << " GB/s, " << op_name << std::endl;
+
+            if(tflops > best_tflops)
+            {
+                best_op_name    = op_name;
+                best_tflops     = tflops;
+                best_avg_time   = avg_time;
+                best_gb_per_sec = gb_per_sec;
+            }
+
+            if(do_verification)
+            {
+                out_device_buf.FromDevice(device_output.mData.data());
+
+                pass = pass & ck::utils::check_err(device_output.mData, host_output.mData);
+
+                if(do_log)
+                {
+                    LogRangeAsType<float>(std::cout << "input : ", input.mData, ",") << std::endl;
+                    LogRangeAsType<float>(std::cout << "weight: ", weight.mData, ",") << std::endl;
+                    LogRangeAsType<float>(std::cout << "host_output  : ", host_output.mData, ",")
+                        << std::endl;
+                    LogRangeAsType<float>(std::cout << "device_output: ", device_output.mData, ",")
+                        << std::endl;
+                }
+            }
+        }
+        else
+        {
+            std::cout << op_ptr->GetTypeString() << " does not support this problem" << std::endl;
+        }
+    }
+
+    std::cout << "Best configuration parameters:"
+              << "\nname: " << best_op_name << "\navg_time: " << best_avg_time
+              << "\ntflops: " << best_tflops << "\nGB/s: " << best_gb_per_sec << std::endl;
+
+    return pass;
+}
+
+} // namespace profiler
+} // namespace ck
diff --git a/profiler/include/profile_gemm_add_add_fastgelu_impl.hpp b/profiler/include/profile_gemm_add_add_fastgelu_impl.hpp
index 849b6f3ea2..d4d37adae5 100644
--- a/profiler/include/profile_gemm_add_add_fastgelu_impl.hpp
+++ b/profiler/include/profile_gemm_add_add_fastgelu_impl.hpp
@@ -13,9 +13,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm_add_add_fastgelu.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -29,7 +29,9 @@ template <typename ADataType,
           typename EDataType,
           typename ALayout,
           typename BLayout,
-          typename DELayout> // assume Ds and E have same layout
+          typename D0Layout,
+          typename D1Layout,
+          typename ELayout>
 bool profile_gemm_add_add_fastgelu_impl(int do_verification,
                                         int init_method,
                                         bool /*do_log*/,
@@ -59,10 +61,10 @@ bool profile_gemm_add_add_fastgelu_impl(int do_verification,
 
     Tensor<ADataType> a_m_k(f_host_tensor_descriptor(M, K, StrideA, ALayout{}));
     Tensor<BDataType> b_k_n(f_host_tensor_descriptor(K, N, StrideB, BLayout{}));
-    Tensor<D0DataType> d0_m_n(f_host_tensor_descriptor(M, N, StrideD0, DELayout{}));
-    Tensor<D1DataType> d1_m_n(f_host_tensor_descriptor(M, N, StrideD1, DELayout{}));
-    Tensor<EDataType> e_m_n_device_result(f_host_tensor_descriptor(M, N, StrideE, DELayout{}));
-    Tensor<EDataType> e_m_n_host_result(f_host_tensor_descriptor(M, N, StrideE, DELayout{}));
+    Tensor<D0DataType> d0_m_n(f_host_tensor_descriptor(M, N, StrideD0, D0Layout{}));
+    Tensor<D1DataType> d1_m_n(f_host_tensor_descriptor(M, N, StrideD1, D1Layout{}));
+    Tensor<EDataType> e_m_n_device_result(f_host_tensor_descriptor(M, N, StrideE, ELayout{}));
+    Tensor<EDataType> e_m_n_host_result(f_host_tensor_descriptor(M, N, StrideE, ELayout{}));
 
     std::cout << "a_m_k: " << a_m_k.mDesc << std::endl;
     std::cout << "b_k_n: " << b_k_n.mDesc << std::endl;
@@ -100,7 +102,8 @@ bool profile_gemm_add_add_fastgelu_impl(int do_verification,
     using DeviceOp = ck::tensor_operation::device::DeviceGemmMultipleD<
         ALayout,
         BLayout,
-        DELayout,
+        ck::Tuple<D0Layout, D1Layout>,
+        ELayout,
         ADataType,
         BDataType,
         ck::Tuple<D0DataType, D1DataType>,
@@ -146,11 +149,11 @@ bool profile_gemm_add_add_fastgelu_impl(int do_verification,
         }
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem d0_m_n_device_buf(sizeof(D0DataType) * d0_m_n.mDesc.GetElementSpace());
-    DeviceMem d1_m_n_device_buf(sizeof(D1DataType) * d1_m_n.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem d0_m_n_device_buf(sizeof(D0DataType) * d0_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem d1_m_n_device_buf(sizeof(D1DataType) * d1_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/profiler/include/profile_gemm_bias_add_reduce_impl.hpp b/profiler/include/profile_gemm_bias_add_reduce_impl.hpp
index 34317c59a7..e59b283b0d 100644
--- a/profiler/include/profile_gemm_bias_add_reduce_impl.hpp
+++ b/profiler/include/profile_gemm_bias_add_reduce_impl.hpp
@@ -10,10 +10,10 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -217,15 +217,15 @@ void profile_gemm_bias_add_reduce_impl(int do_verification,
         }
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
-    DeviceMem bias_device_buf(sizeof(BiasDataType) * bias_n.mDesc.GetElementSpace());
-    DeviceMem d0_device_buf(sizeof(D0DataType) * d0_m_n.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
+    DeviceMem bias_device_buf(sizeof(BiasDataType) * bias_n.mDesc.GetElementSpaceSize());
+    DeviceMem d0_device_buf(sizeof(D0DataType) * d0_m_n.mDesc.GetElementSpaceSize());
     DeviceMem reduce0_device_buf(sizeof(ReduceDataType) *
-                                 reduce0_m_device_result.mDesc.GetElementSpace());
+                                 reduce0_m_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce1_device_buf(sizeof(ReduceDataType) *
-                                 reduce1_m_device_result.mDesc.GetElementSpace());
+                                 reduce1_m_device_result.mDesc.GetElementSpaceSize());
 
     std::array<void*, 2> p_reduces = {reduce0_device_buf.GetDeviceBuffer(),
                                       reduce1_device_buf.GetDeviceBuffer()};
diff --git a/profiler/include/profile_gemm_bilinear_impl.hpp b/profiler/include/profile_gemm_bilinear_impl.hpp
index f273ff4417..17d0553db8 100644
--- a/profiler/include/profile_gemm_bilinear_impl.hpp
+++ b/profiler/include/profile_gemm_bilinear_impl.hpp
@@ -13,9 +13,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm_bilinear.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -28,7 +28,8 @@ template <typename ADataType,
           typename EDataType,
           typename ALayout,
           typename BLayout,
-          typename DELayout> // assume Ds and E have same layout
+          typename DLayout,
+          typename ELayout>
 bool profile_gemm_bilinear_impl(int do_verification,
                                 int init_method,
                                 bool /*do_log*/,
@@ -59,9 +60,9 @@ bool profile_gemm_bilinear_impl(int do_verification,
 
     Tensor<ADataType> a_m_k(f_host_tensor_descriptor(M, K, StrideA, ALayout{}));
     Tensor<BDataType> b_k_n(f_host_tensor_descriptor(K, N, StrideB, BLayout{}));
-    Tensor<DDataType> d_m_n(f_host_tensor_descriptor(M, N, StrideD, DELayout{}));
-    Tensor<EDataType> e_m_n_device_result(f_host_tensor_descriptor(M, N, StrideE, DELayout{}));
-    Tensor<EDataType> e_m_n_host_result(f_host_tensor_descriptor(M, N, StrideE, DELayout{}));
+    Tensor<DDataType> d_m_n(f_host_tensor_descriptor(M, N, StrideD, DLayout{}));
+    Tensor<EDataType> e_m_n_device_result(f_host_tensor_descriptor(M, N, StrideE, ELayout{}));
+    Tensor<EDataType> e_m_n_host_result(f_host_tensor_descriptor(M, N, StrideE, ELayout{}));
 
     std::cout << "a_m_k: " << a_m_k.mDesc << std::endl;
     std::cout << "b_k_n: " << b_k_n.mDesc << std::endl;
@@ -96,7 +97,8 @@ bool profile_gemm_bilinear_impl(int do_verification,
     using DeviceOp = ck::tensor_operation::device::DeviceGemmMultipleD<
         ALayout,
         BLayout,
-        DELayout,
+        ck::Tuple<DLayout>,
+        ELayout,
         ADataType,
         BDataType,
         ck::Tuple<DDataType>,
@@ -142,10 +144,10 @@ bool profile_gemm_bilinear_impl(int do_verification,
         }
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem d_m_n_device_buf(sizeof(DDataType) * d_m_n.mDesc.GetElementSpace());
-    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem d_m_n_device_buf(sizeof(DDataType) * d_m_n.mDesc.GetElementSpaceSize());
+    DeviceMem e_device_buf(sizeof(EDataType) * e_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/profiler/include/profile_gemm_impl.hpp b/profiler/include/profile_gemm_impl.hpp
index 54b9e05c06..c15dcae691 100644
--- a/profiler/include/profile_gemm_impl.hpp
+++ b/profiler/include/profile_gemm_impl.hpp
@@ -15,21 +15,21 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
 namespace profiler {
 
-template <typename ADataType,
+template <typename ALayout,
+          typename BLayout,
+          typename CLayout,
+          typename ADataType,
           typename BDataType,
           typename AccDataType,
-          typename CDataType,
-          typename ALayout,
-          typename BLayout,
-          typename CLayout>
+          typename CDataType>
 int profile_gemm_impl(int do_verification,
                       int init_method,
                       bool do_log,
@@ -86,13 +86,12 @@ int profile_gemm_impl(int do_verification,
     const auto b_element_op = BElementOp{};
     const auto c_element_op = CElementOp{};
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
-    c_device_buf.ToDevice(c_m_n_device_result.mData.data());
 
     using DeviceOp = ck::tensor_operation::device::DeviceGemm<ALayout,
                                                               BLayout,
@@ -110,7 +109,7 @@ int profile_gemm_impl(int do_verification,
 
     std::cout << "found " << op_ptrs.size() << " instances" << std::endl;
 
-    // Run reference GEMM
+    // Run reference op
     if(do_verification)
     {
         using ReferenceGemmInstance = ck::tensor_operation::host::ReferenceGemm<ADataType,
@@ -131,11 +130,11 @@ int profile_gemm_impl(int do_verification,
     }
 
     std::string best_op_name;
-    float best_ave_time   = 0;
+    float best_avg_time   = 0;
     float best_tflops     = 0;
     float best_gb_per_sec = 0;
 
-    // profile device GEMM instances
+    // profile device op instances
     for(auto& op_ptr : op_ptrs)
     {
         auto argument_ptr =
@@ -161,7 +160,7 @@ int profile_gemm_impl(int do_verification,
 
             std::string op_name = op_ptr->GetTypeString();
 
-            float ave_time =
+            float avg_time =
                 invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, time_kernel});
 
             std::size_t flop = std::size_t(2) * M * N * K;
@@ -169,18 +168,18 @@ int profile_gemm_impl(int do_verification,
             std::size_t num_btype =
                 sizeof(ADataType) * M * K + sizeof(BDataType) * K * N + sizeof(CDataType) * M * N;
 
-            float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
+            float tflops = static_cast<float>(flop) / 1.E9 / avg_time;
 
-            float gb_per_sec = num_btype / 1.E6 / ave_time;
+            float gb_per_sec = num_btype / 1.E6 / avg_time;
 
-            std::cout << "Perf: " << std::setw(10) << ave_time << " ms, " << tflops << " TFlops, "
+            std::cout << "Perf: " << std::setw(10) << avg_time << " ms, " << tflops << " TFlops, "
                       << gb_per_sec << " GB/s, " << op_name << std::endl;
 
             if(tflops > best_tflops)
             {
                 best_op_name    = op_name;
                 best_tflops     = tflops;
-                best_ave_time   = ave_time;
+                best_avg_time   = avg_time;
                 best_gb_per_sec = gb_per_sec;
             }
 
@@ -244,7 +243,7 @@ int profile_gemm_impl(int do_verification,
     }
 
     std::cout << " M = " << M << " N = " << N << " K = " << K << " StrideA = " << StrideA
-              << " StrideB = " << StrideB << " StrideC = " << StrideC << " : " << best_ave_time
+              << " StrideB = " << StrideB << " StrideC = " << StrideC << " : " << best_avg_time
               << " ms, " << best_tflops << " TFlops, " << best_gb_per_sec << " GB/s, "
               << best_op_name << std::endl;
 
diff --git a/profiler/include/profile_gemm_reduce_impl.hpp b/profiler/include/profile_gemm_reduce_impl.hpp
index 0f891a7aee..fd4db3bce4 100644
--- a/profiler/include/profile_gemm_reduce_impl.hpp
+++ b/profiler/include/profile_gemm_reduce_impl.hpp
@@ -10,10 +10,10 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -189,13 +189,13 @@ bool profile_gemm_reduce_impl(int do_verification,
         }
     }
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce0_device_buf(sizeof(ReduceDataType) *
-                                 reduce0_m_device_result.mDesc.GetElementSpace());
+                                 reduce0_m_device_result.mDesc.GetElementSpaceSize());
     DeviceMem reduce1_device_buf(sizeof(ReduceDataType) *
-                                 reduce1_m_device_result.mDesc.GetElementSpace());
+                                 reduce1_m_device_result.mDesc.GetElementSpaceSize());
 
     std::array<void*, 2> p_reduces = {reduce0_device_buf.GetDeviceBuffer(),
                                       reduce1_device_buf.GetDeviceBuffer()};
diff --git a/profiler/include/profile_gemm_splitk_impl.hpp b/profiler/include/profile_gemm_splitk_impl.hpp
index 8be879dcbe..ba6ceb7514 100644
--- a/profiler/include/profile_gemm_splitk_impl.hpp
+++ b/profiler/include/profile_gemm_splitk_impl.hpp
@@ -15,9 +15,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm_splitk.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -87,9 +87,9 @@ bool profile_gemm_splitk_impl(int do_verification,
     const auto b_element_op = BElementOp{};
     const auto c_element_op = CElementOp{};
 
-    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(ADataType) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(BDataType) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(CDataType) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/profiler/include/profile_grouped_conv_fwd_impl.hpp b/profiler/include/profile_grouped_conv_fwd_impl.hpp
new file mode 100644
index 0000000000..8d7ebe0465
--- /dev/null
+++ b/profiler/include/profile_grouped_conv_fwd_impl.hpp
@@ -0,0 +1,250 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <iomanip>
+#include <iostream>
+#include <typeinfo>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp"
+
+#include "ck/library/utility/check_err.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+#include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
+
+namespace ck {
+namespace profiler {
+
+template <ck::index_t NDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+bool profile_grouped_conv_fwd_impl(int do_verification,
+                                   int init_method,
+                                   bool do_log,
+                                   bool time_kernel,
+                                   const ck::utils::conv::ConvParam& conv_param)
+{
+    using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+    using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+    using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+    const auto in_element_op  = InElementOp{};
+    const auto wei_element_op = WeiElementOp{};
+    const auto out_element_op = OutElementOp{};
+
+    const auto in_g_n_c_wis_desc =
+        ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);
+
+    const auto wei_g_k_c_xs_desc =
+        ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(conv_param);
+
+    const auto out_g_n_k_wos_desc =
+        ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(conv_param);
+
+    std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> a_g_n_c_wis_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> b_g_k_c_xs_strides{};
+    std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_lengths{};
+    std::array<ck::index_t, NDimSpatial + 3> e_g_n_k_wos_strides{};
+    std::array<ck::index_t, NDimSpatial> conv_filter_strides{};
+    std::array<ck::index_t, NDimSpatial> conv_filter_dilations{};
+    std::array<ck::index_t, NDimSpatial> input_left_pads{};
+    std::array<ck::index_t, NDimSpatial> input_right_pads{};
+
+    auto copy = [](auto& x, auto& y) { std::copy(x.begin(), x.end(), y.begin()); };
+
+    copy(in_g_n_c_wis_desc.GetLengths(), a_g_n_c_wis_lengths);
+    copy(in_g_n_c_wis_desc.GetStrides(), a_g_n_c_wis_strides);
+    copy(wei_g_k_c_xs_desc.GetLengths(), b_g_k_c_xs_lengths);
+    copy(wei_g_k_c_xs_desc.GetStrides(), b_g_k_c_xs_strides);
+    copy(out_g_n_k_wos_desc.GetLengths(), e_g_n_k_wos_lengths);
+    copy(out_g_n_k_wos_desc.GetStrides(), e_g_n_k_wos_strides);
+    copy(conv_param.conv_filter_strides_, conv_filter_strides);
+    copy(conv_param.conv_filter_dilations_, conv_filter_dilations);
+    copy(conv_param.input_left_pads_, input_left_pads);
+    copy(conv_param.input_right_pads_, input_right_pads);
+
+    Tensor<InDataType> input(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> weight(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> host_output(out_g_n_k_wos_desc);
+    Tensor<OutDataType> device_output(out_g_n_k_wos_desc);
+
+    std::cout << "input: " << input.mDesc << std::endl;
+    std::cout << "weight: " << weight.mDesc << std::endl;
+    std::cout << "output: " << host_output.mDesc << std::endl;
+
+    switch(init_method)
+    {
+    case 0: break;
+    case 1:
+        input.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
+        weight.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        break;
+    default:
+        input.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
+        weight.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+    }
+
+    DeviceMem in_device_buf(sizeof(InDataType) * input.mDesc.GetElementSpaceSize());
+    DeviceMem wei_device_buf(sizeof(WeiDataType) * weight.mDesc.GetElementSpaceSize());
+    DeviceMem out_device_buf(sizeof(OutDataType) * device_output.mDesc.GetElementSpaceSize());
+
+    in_device_buf.ToDevice(input.mData.data());
+    wei_device_buf.ToDevice(weight.mData.data());
+
+    // run reference op
+    if(do_verification)
+    {
+        auto ref_conv = ck::tensor_operation::host::ReferenceConvFwd<NDimSpatial,
+                                                                     InDataType,
+                                                                     WeiDataType,
+                                                                     OutDataType,
+                                                                     InElementOp,
+                                                                     WeiElementOp,
+                                                                     OutElementOp>{};
+
+        auto ref_invoker  = ref_conv.MakeInvoker();
+        auto ref_argument = ref_conv.MakeArgument(input,
+                                                  weight,
+                                                  host_output,
+                                                  conv_param.conv_filter_strides_,
+                                                  conv_param.conv_filter_dilations_,
+                                                  conv_param.input_left_pads_,
+                                                  conv_param.input_right_pads_,
+                                                  in_element_op,
+                                                  wei_element_op,
+                                                  out_element_op);
+
+        // init host output to zero
+        host_output.SetZero();
+
+        ref_invoker.Run(ref_argument);
+    }
+
+    using DeviceOp = ck::tensor_operation::device::DeviceGroupedConvFwdMultipleD<NDimSpatial,
+                                                                                 InLayout,
+                                                                                 WeiLayout,
+                                                                                 ck::Tuple<>,
+                                                                                 OutLayout,
+                                                                                 InDataType,
+                                                                                 WeiDataType,
+                                                                                 ck::Tuple<>,
+                                                                                 OutDataType,
+                                                                                 InElementOp,
+                                                                                 WeiElementOp,
+                                                                                 OutElementOp>;
+
+    // get device op instances
+    const auto op_ptrs = ck::tensor_operation::device::instance::DeviceOperationInstanceFactory<
+        DeviceOp>::GetInstances();
+
+    std::cout << "found " << op_ptrs.size() << " instances" << std::endl;
+
+    std::string best_op_name;
+    float best_avg_time   = 0;
+    float best_tflops     = 0;
+    float best_gb_per_sec = 0;
+
+    // profile device op instances
+    bool pass = true;
+
+    for(auto& op_ptr : op_ptrs)
+    {
+        auto argument_ptr =
+            op_ptr->MakeArgumentPointer(in_device_buf.GetDeviceBuffer(),
+                                        wei_device_buf.GetDeviceBuffer(),
+                                        std::array<const void*, 0>{},
+                                        out_device_buf.GetDeviceBuffer(),
+                                        a_g_n_c_wis_lengths,
+                                        a_g_n_c_wis_strides,
+                                        b_g_k_c_xs_lengths,
+                                        b_g_k_c_xs_strides,
+                                        std::array<std::array<ck::index_t, NDimSpatial + 3>, 0>{{}},
+                                        std::array<std::array<ck::index_t, NDimSpatial + 3>, 0>{{}},
+                                        e_g_n_k_wos_lengths,
+                                        e_g_n_k_wos_strides,
+                                        conv_filter_strides,
+                                        conv_filter_dilations,
+                                        input_left_pads,
+                                        input_right_pads,
+                                        in_element_op,
+                                        wei_element_op,
+                                        out_element_op);
+
+        if(op_ptr->IsSupportedArgument(argument_ptr.get()))
+        {
+            // re-init output to zero before profiling next kernel
+            out_device_buf.SetZero();
+
+            std::string op_name = op_ptr->GetTypeString();
+
+            auto invoker_ptr = op_ptr->MakeInvokerPointer();
+
+            float avg_time =
+                invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, time_kernel});
+
+            std::size_t flop      = conv_param.GetFlops();
+            std::size_t num_btype = conv_param.GetByte<InDataType, WeiDataType, OutDataType>();
+
+            float tflops = static_cast<float>(flop) / 1.E9 / avg_time;
+
+            float gb_per_sec = num_btype / 1.E6 / avg_time;
+
+            std::cout << "Perf: " << std::setw(10) << avg_time << " ms, " << tflops << " TFlops, "
+                      << gb_per_sec << " GB/s, " << op_name << std::endl;
+
+            if(tflops > best_tflops)
+            {
+                best_op_name    = op_name;
+                best_tflops     = tflops;
+                best_avg_time   = avg_time;
+                best_gb_per_sec = gb_per_sec;
+            }
+
+            if(do_verification)
+            {
+                out_device_buf.FromDevice(device_output.mData.data());
+
+                pass = pass & ck::utils::check_err(device_output.mData, host_output.mData);
+
+                if(do_log)
+                {
+                    LogRangeAsType<float>(std::cout << "input : ", input.mData, ",") << std::endl;
+                    LogRangeAsType<float>(std::cout << "weight: ", weight.mData, ",") << std::endl;
+                    LogRangeAsType<float>(std::cout << "host_output  : ", host_output.mData, ",")
+                        << std::endl;
+                    LogRangeAsType<float>(std::cout << "device_output: ", device_output.mData, ",")
+                        << std::endl;
+                }
+            }
+        }
+        else
+        {
+            std::cout << op_ptr->GetTypeString() << " does not support this problem" << std::endl;
+        }
+    }
+
+    std::cout << "Best configuration parameters:"
+              << "\nname: " << best_op_name << "\navg_time: " << best_avg_time
+              << "\ntflops: " << best_tflops << "\nGB/s: " << best_gb_per_sec << std::endl;
+
+    return pass;
+}
+
+} // namespace profiler
+} // namespace ck
diff --git a/profiler/include/profile_grouped_gemm_impl.hpp b/profiler/include/profile_grouped_gemm_impl.hpp
index ea2a503fbc..4853fc98f2 100644
--- a/profiler/include/profile_grouped_gemm_impl.hpp
+++ b/profiler/include/profile_grouped_gemm_impl.hpp
@@ -13,10 +13,10 @@
 #include "ck/library/tensor_operation_instance/gpu/grouped_gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -24,7 +24,7 @@ namespace profiler {
 
 template <typename ADataType,
           typename BDataType,
-          typename EDataType,
+          typename CDataType,
           typename AccDataType,
           typename ALayout,
           typename BLayout,
@@ -67,7 +67,7 @@ bool profile_grouped_gemm_impl(int do_verification,
 
     std::vector<Tensor<ADataType>> a_m_k;
     std::vector<Tensor<BDataType>> b_k_n;
-    std::vector<Tensor<EDataType>> c_m_n_device_results;
+    std::vector<Tensor<CDataType>> c_m_n_device_results;
 
     for(std::size_t i = 0; i < group_count; i++)
     {
@@ -77,7 +77,7 @@ bool profile_grouped_gemm_impl(int do_verification,
             Tensor<BDataType>(f_host_tensor_descriptor(Ks[i], Ns[i], StrideBs[i], BLayout{})));
 
         c_m_n_device_results.push_back(
-            Tensor<EDataType>(f_host_tensor_descriptor(Ms[i], Ns[i], StrideCs[i], CLayout{})));
+            Tensor<CDataType>(f_host_tensor_descriptor(Ms[i], Ns[i], StrideCs[i], CLayout{})));
 
         std::cout << "group: " << i << " a_m_k[" << i << "]:" << a_m_k[i].mDesc << ", b_k_n[" << i
                   << "]:" << b_k_n[i].mDesc << ", c_m_n_device_results[" << i
@@ -96,7 +96,7 @@ bool profile_grouped_gemm_impl(int do_verification,
             b_k_n[i].GenerateTensorValue(GeneratorTensor_3<BDataType>{-0.5, 0.5}, num_thread);
         }
 
-        c_m_n_device_results[i].GenerateTensorValue(GeneratorTensor_0<EDataType>{}, num_thread);
+        c_m_n_device_results[i].GenerateTensorValue(GeneratorTensor_0<CDataType>{}, num_thread);
     }
 
     using AElementOp = ck::tensor_operation::element_wise::PassThrough;
@@ -133,12 +133,12 @@ bool profile_grouped_gemm_impl(int do_verification,
     for(std::size_t i = 0; i < group_count; i++)
     {
         a_device_buf.emplace_back(
-            std::make_unique<DeviceMem>(sizeof(ADataType) * a_m_k[i].mDesc.GetElementSpace()));
+            std::make_unique<DeviceMem>(sizeof(ADataType) * a_m_k[i].mDesc.GetElementSpaceSize()));
         b_device_buf.emplace_back(
-            std::make_unique<DeviceMem>(sizeof(BDataType) * b_k_n[i].mDesc.GetElementSpace()));
+            std::make_unique<DeviceMem>(sizeof(BDataType) * b_k_n[i].mDesc.GetElementSpaceSize()));
 
         c_device_buf.emplace_back(std::make_unique<DeviceMem>(
-            sizeof(EDataType) * c_m_n_device_results[i].mDesc.GetElementSpace()));
+            sizeof(CDataType) * c_m_n_device_results[i].mDesc.GetElementSpaceSize()));
 
         a_device_buf[i]->ToDevice(a_m_k[i].mData.data());
         b_device_buf[i]->ToDevice(b_k_n[i].mData.data());
@@ -153,11 +153,12 @@ bool profile_grouped_gemm_impl(int do_verification,
 
     using DeviceOp = ck::tensor_operation::device::DeviceGroupedGemm<ALayout,
                                                                      BLayout,
+                                                                     ck::Tuple<>,
                                                                      CLayout,
                                                                      ADataType,
                                                                      BDataType,
                                                                      ck::Tuple<>,
-                                                                     EDataType,
+                                                                     CDataType,
                                                                      AElementOp,
                                                                      BElementOp,
                                                                      CElementOp>;
@@ -209,7 +210,7 @@ bool profile_grouped_gemm_impl(int do_verification,
                 flop += std::size_t(2) * Ms[i] * Ns[i] * Ks[i];
 
                 num_btype += sizeof(ADataType) * Ms[i] * Ks[i] + sizeof(BDataType) * Ks[i] * Ns[i] +
-                             sizeof(EDataType) * Ms[i] * Ns[i];
+                             sizeof(CDataType) * Ms[i] * Ns[i];
             }
 
             float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
@@ -233,13 +234,13 @@ bool profile_grouped_gemm_impl(int do_verification,
 
                     c_device_buf[i]->FromDevice(c_m_n_device_results[i].mData.data());
 
-                    Tensor<EDataType> c_m_n_host_result(
+                    Tensor<CDataType> c_m_n_host_result(
                         f_host_tensor_descriptor(Ms[i], Ns[i], StrideCs[i], CLayout{}));
 
                     using ReferenceGemmInstance =
                         ck::tensor_operation::host::ReferenceGemm<ADataType,
                                                                   BDataType,
-                                                                  EDataType,
+                                                                  CDataType,
                                                                   AccDataType,
                                                                   AElementOp,
                                                                   BElementOp,
diff --git a/profiler/include/profile_normalization_impl.hpp b/profiler/include/profile_normalization_impl.hpp
index 6e864698c1..77a2c32d18 100644
--- a/profiler/include/profile_normalization_impl.hpp
+++ b/profiler/include/profile_normalization_impl.hpp
@@ -9,10 +9,10 @@
 #include "ck/tensor_operation/gpu/device/device_softmax.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_softmax.hpp"
 
 namespace ck {
@@ -92,8 +92,8 @@ void profile_normalization_impl(int do_verification,
 
     Tensor<OutDataType> out_ref(out);
 
-    DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpace());
-    DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpace());
+    DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+    DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
     in_dev.ToDevice(in.mData.data());
     out_dev.ToDevice(out.mData.data());
 
diff --git a/profiler/include/profile_reduce_impl.hpp b/profiler/include/profile_reduce_impl.hpp
index a88b4bcd07..2d06ec22c5 100644
--- a/profiler/include/profile_reduce_impl.hpp
+++ b/profiler/include/profile_reduce_impl.hpp
@@ -8,10 +8,10 @@
 
 #include "ck/library/utility/check_err.hpp"
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_reduction.hpp"
-#include "ck/library/host_tensor/host_common_util.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_reduction.hpp"
+#include "ck/library/utility/host_common_util.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 namespace ck {
 namespace tensor_operation {
@@ -245,13 +245,13 @@ bool profile_reduce_impl_impl(bool do_verification,
             }
 
             if(beta != 0.0f)
-                for(size_t i = 0; i < out_ref.mDesc.GetElementSpace(); i++)
+                for(size_t i = 0; i < out_ref.mDesc.GetElementSpaceSize(); i++)
                     out.mData[i] = out_ref.mData[i];
         };
 
         // these buffers are usually provided by the user application
-        DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpace());
-        DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpace());
+        DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+        DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
 
         in_dev.ToDevice(in.mData.data());
 
diff --git a/profiler/src/profile_batched_gemm_reduce.cpp b/profiler/src/profile_batched_gemm_reduce.cpp
index 7c518e979b..d734b5d87b 100644
--- a/profiler/src/profile_batched_gemm_reduce.cpp
+++ b/profiler/src/profile_batched_gemm_reduce.cpp
@@ -24,9 +24,9 @@ int profile_batched_gemm_reduce(int argc, char* argv[])
         F16_F16_F16_F32_F32, // 1
     };
 
-    if(!(argc == 15 || argc == 16))
+    if(argc != 15)
     {
-        printf("arg1: tensor operation (batched_gemm: BatchedGEMM+Reduce)\n");
+        printf("arg1: tensor operation (batched_gemm_reduce: BatchedGEMM+Reduce)\n");
         printf("arg2: data type (0: fp32; 1: fp16)\n");
         printf("arg3: matrix layout (0: A[m, k] * B[k, n] = C[m, n];\n");
         printf("                     1: A[m, k] * B[n, k] = C[m, n];\n");
@@ -37,7 +37,6 @@ int profile_batched_gemm_reduce(int argc, char* argv[])
         printf("arg6: print tensor value (0: no; 1: yes)\n");
         printf("arg7: time kernel (0=n0, 1=yes)\n");
         printf("arg8 to 14: M, N, K, StrideA, StrideB, StrideC, BatchCount\n");
-        printf("arg15: split k into  mulitiple batch\n");
         exit(1);
     }
 
diff --git a/profiler/src/profile_conv_bwd_data.cpp b/profiler/src/profile_conv_bwd_data.cpp
new file mode 100644
index 0000000000..cf42afd2aa
--- /dev/null
+++ b/profiler/src/profile_conv_bwd_data.cpp
@@ -0,0 +1,184 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+
+#include "profiler/include/profile_conv_bwd_data_impl.hpp"
+
+namespace {
+
+enum struct ConvLayout
+{
+    NCHW_KCYX_NKHW, // 0
+    NHWC_KYXC_NHWK, // 1
+};
+
+enum struct ConvDataType
+{
+    F32_F32_F32,    // 0
+    F16_F16_F16,    // 1
+    BF16_BF16_BF16, // 2
+    INT8_INT8_INT8, // 3
+};
+
+static void print_helper_msg()
+{
+    std::cout
+        << "arg1: tensor operation (conv_bwd_data: Convolution Backward Data)\n"
+        << "arg2: data type (0: Input fp32, Weight fp32, Output fp32\n"
+        << "                 1: Input fp16, Weight fp16, Output fp16\n"
+        << "                 2: Input bf16, Weight bf16, Output bf16\n"
+        << "                 3: Input int8, Weight int8, Output int8)\n"
+        << "arg3: tensor layout (0: Input[N, C, Hi, Wi], Weight[K, C, Y, X], Output[N, K, Ho, Wo]\n"
+        << "                     1: Input[N, Hi, Wi, C], Weight[K, Y, X, C], Output[N, Ho, Wo, "
+           "K])\n"
+        << "arg4: verification (0: no, 1: yes)\n"
+        << "arg5: initialization (0: no init, 1: integer value, 2: decimal value)\n"
+        << "arg6: print tensor value (0: no; 1: yes)\n"
+        << "arg7: time kernel (0: no, 1: yes)\n"
+        << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+}
+
+} // namespace
+
+int profile_conv_bwd_data(int argc, char* argv[])
+{
+    // 8 for control, 1 for num_dim_spatial
+    if(argc < 9)
+    {
+        print_helper_msg();
+        return 1;
+    }
+
+    const auto data_type       = static_cast<ConvDataType>(std::stoi(argv[2]));
+    const auto layout          = static_cast<ConvLayout>(std::stoi(argv[3]));
+    const bool do_verification = std::stoi(argv[4]);
+    const int init_method      = std::stoi(argv[5]);
+    const bool do_log          = std::stoi(argv[6]);
+    const bool time_kernel     = std::stoi(argv[7]);
+    const int num_dim_spatial  = std::stoi(argv[8]);
+
+    // 8 for control, 1 for num_dim_spatial, 4 for G/N/K/C, and 6 * num_dim_spatial
+    if(argc != 8 + 1 + 4 + 6 * num_dim_spatial)
+    {
+        print_helper_msg();
+        return 1;
+    }
+
+    const auto params = ck::utils::conv::parse_conv_param(num_dim_spatial, 9, argv);
+
+    using F32  = float;
+    using F16  = ck::half_t;
+    using BF16 = ck::bhalf_t;
+    using INT8 = int8_t;
+
+    using NWC   = ck::tensor_layout::convolution::NWC;
+    using NHWC  = ck::tensor_layout::convolution::NHWC;
+    using NDHWC = ck::tensor_layout::convolution::NDHWC;
+
+    using KXC   = ck::tensor_layout::convolution::KXC;
+    using KYXC  = ck::tensor_layout::convolution::KYXC;
+    using KZYXC = ck::tensor_layout::convolution::KZYXC;
+
+    using NWK   = ck::tensor_layout::convolution::NWK;
+    using NHWK  = ck::tensor_layout::convolution::NHWK;
+    using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+    constexpr auto I1 = ck::Number<1>{};
+    constexpr auto I2 = ck::Number<2>{};
+    constexpr auto I3 = ck::Number<3>{};
+
+    auto profile = [&](auto num_dim_spatial_tmp,
+                       auto in_layout,
+                       auto wei_layout,
+                       auto out_layout,
+                       auto in_type,
+                       auto wei_type,
+                       auto out_type) {
+        constexpr ck::index_t NDimSpatial = num_dim_spatial_tmp.value;
+
+        using InLayout  = decltype(in_layout);
+        using WeiLayout = decltype(wei_layout);
+        using OutLayout = decltype(out_layout);
+
+        using InDataType  = decltype(in_type);
+        using WeiDataType = decltype(wei_type);
+        using OutDataType = decltype(out_type);
+
+        bool pass = ck::profiler::profile_conv_bwd_data_impl<NDimSpatial,
+                                                             InLayout,
+                                                             WeiLayout,
+                                                             OutLayout,
+                                                             InDataType,
+                                                             WeiDataType,
+                                                             OutDataType>(
+            do_verification, init_method, do_log, time_kernel, params);
+
+        return pass ? 0 : 1;
+    };
+
+    if(num_dim_spatial == 1 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 2 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 3 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+
+    std::cout << "this data_type & layout is not implemented" << std::endl;
+
+    return 1;
+}
diff --git a/profiler/src/profile_conv_bwd_weight.cpp b/profiler/src/profile_conv_bwd_weight.cpp
index 989c480886..5ff5031eab 100644
--- a/profiler/src/profile_conv_bwd_weight.cpp
+++ b/profiler/src/profile_conv_bwd_weight.cpp
@@ -8,141 +8,168 @@
 
 #include "profiler/include/profile_conv_bwd_weight_impl.hpp"
 
+namespace {
+
+enum struct ConvLayout
+{
+    NCHW_KCYX_NKHW, // 0
+    NHWC_KYXC_NHWK, // 1
+};
+
 enum struct ConvDataType
 {
-    F32_F32_F32,    // 0
-    F16_F16_F16,    // 1
-    BF16_BF16_BF16, // 2
-    INT8_INT8_INT8, // 3
+    F32_F32_F32,   // 0
+    F16_F16_F16,   // 1
+    BF16_F32_BF16, // 2
 };
 
-enum struct ConvInputLayout
+static void print_helper_msg()
 {
-    NCHW, // 0
-    NHWC, // 1
-};
+    std::cout
+        << "arg1: tensor operation (conv_bwd_weight: Convolution Backward Weight\n"
+        << "arg2: data type (0: Input fp32, Weight fp32, Output fp32\n"
+        << "                 1: Input fp16, Weight fp16, Output fp16\n"
+        << "                 2: Input bf16, Weight fp32, Output bf16)\n"
+        << "arg3: tensor layout (0: Input[N, C, Hi, Wi], Weight[K, C, Y, X], Output[N, K, Ho, Wo]\n"
+        << "                     1: Input[N, Hi, Wi, C], Weight[K, Y, X, C], Output[N, Ho, Wo, K]\n"
+        << "arg4: verification (0: no, 1: yes)\n"
+        << "arg5: initialization (0: no init, 1: integer value, 2: decimal value)\n"
+        << "arg6: print tensor value (0: no; 1: yes)\n"
+        << "arg7: time kernel (0: no, 1: yes)\n"
+        << ck::utils::conv::get_conv_param_parser_helper_msg() << " SplitK\n"
+        << std::endl;
+}
 
-enum struct ConvWeightLayout
-{
-    KCYX, // 0
-    KYXC, // 1
-};
-
-enum struct ConvOutputLayout
-{
-    NKHW, // 0
-    NHWK, // 1
-};
+} // namespace
 
 int profile_conv_bwd_weight(int argc, char* argv[])
 {
-    if(argc != 26)
+    // 8 for control, 1 for num_dim_spatial
+    if(argc < 9)
     {
-        printf("arg1: tensor operation (conv_fwd: ForwardConvolution)\n");
-        printf("arg2: data type (0: fp32; 1: fp16)\n");
-        printf("arg3: input tensor layout (0: NCHW; 1: NHWC)\n");
-        printf("arg4: weight tensor layout (0: KCYX; 1: KYXC)\n");
-        printf("arg5: output tensor layout (0: NKHW; 1: NHWK)\n");
-        printf("arg6: verification (0: no; 1: yes)\n");
-        printf("arg7: initialization (0: no init; 1: integer value; 2: decimal value)\n");
-        printf("arg8: print tensor value (0: no; 1: yes)\n");
-        printf("arg9: run kernel # of times (>1)\n");
-        printf("arg10 to 24: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-               "RightPx\n");
-        printf("arg25: split k (>=1)\n");
-        exit(1);
+        print_helper_msg();
+        return 1;
     }
 
     const auto data_type       = static_cast<ConvDataType>(std::stoi(argv[2]));
-    const auto in_layout       = static_cast<ConvInputLayout>(std::stoi(argv[3]));
-    const auto wei_layout      = static_cast<ConvWeightLayout>(std::stoi(argv[4]));
-    const auto out_layout      = static_cast<ConvOutputLayout>(std::stoi(argv[5]));
-    const bool do_verification = std::stoi(argv[6]);
-    const int init_method      = std::stoi(argv[7]);
-    const bool do_log          = std::stoi(argv[8]);
-    const bool time_kernel     = std::stoi(argv[9]);
+    const auto layout          = static_cast<ConvLayout>(std::stoi(argv[3]));
+    const bool do_verification = std::stoi(argv[4]);
+    const int init_method      = std::stoi(argv[5]);
+    const bool do_log          = std::stoi(argv[6]);
+    const bool time_kernel     = std::stoi(argv[7]);
+    const int num_dim_spatial  = std::stoi(argv[8]);
 
-    const ck::index_t N  = std::stoi(argv[10]);
-    const ck::index_t K  = std::stoi(argv[11]);
-    const ck::index_t C  = std::stoi(argv[12]);
-    const ck::index_t Y  = std::stoi(argv[13]);
-    const ck::index_t X  = std::stoi(argv[14]);
-    const ck::index_t Hi = std::stoi(argv[15]);
-    const ck::index_t Wi = std::stoi(argv[16]);
-
-    const ck::index_t conv_stride_h   = std::stoi(argv[17]);
-    const ck::index_t conv_stride_w   = std::stoi(argv[18]);
-    const ck::index_t conv_dilation_h = std::stoi(argv[19]);
-    const ck::index_t conv_dilation_w = std::stoi(argv[20]);
-    const ck::index_t in_left_pad_h   = std::stoi(argv[21]);
-    const ck::index_t in_left_pad_w   = std::stoi(argv[22]);
-    const ck::index_t in_right_pad_h  = std::stoi(argv[23]);
-    const ck::index_t in_right_pad_w  = std::stoi(argv[24]);
-    ck::index_t split_k               = std::stoi(argv[25]);
-    split_k                           = std::max(1, split_k);
-
-    const ck::index_t YEff = (Y - 1) * conv_dilation_h + 1;
-    const ck::index_t XEff = (X - 1) * conv_dilation_w + 1;
-
-    const ck::index_t Ho = (Hi + in_left_pad_h + in_right_pad_h - YEff) / conv_stride_h + 1;
-    const ck::index_t Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
-
-    if(data_type == ConvDataType::F32_F32_F32 && in_layout == ConvInputLayout::NHWC &&
-       wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
+    // 8 for control, 1 for num_dim_spatial, 4 for G/N/K/C, and 6 * num_dim_spatial, 1 for split-K
+    if(argc != 8 + 1 + 4 + 6 * num_dim_spatial + 1)
     {
-        ck::profiler::profile_conv_bwd_weight_impl<2,
-                                                   float,
-                                                   float,
-                                                   float,
-                                                   ck::tensor_layout::convolution::NHWC,
-                                                   ck::tensor_layout::convolution::KYXC,
-                                                   ck::tensor_layout::convolution::NHWK>(
-            do_verification,
-            init_method,
-            do_log,
-            time_kernel,
-            N,
-            K,
-            C,
-            std::vector<ck::index_t>{Hi, Wi},
-            std::vector<ck::index_t>{Y, X},
-            std::vector<ck::index_t>{Ho, Wo},
-            std::vector<ck::index_t>{conv_stride_h, conv_stride_w},
-            std::vector<ck::index_t>{conv_dilation_h, conv_dilation_w},
-            std::vector<ck::index_t>{in_left_pad_h, in_left_pad_w},
-            std::vector<ck::index_t>{in_right_pad_h, in_right_pad_w},
-            split_k);
-    }
-    else if(data_type == ConvDataType::F16_F16_F16 && in_layout == ConvInputLayout::NHWC &&
-            wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        ck::profiler::profile_conv_bwd_weight_impl<2,
-                                                   ck::half_t,
-                                                   ck::half_t,
-                                                   ck::half_t,
-                                                   ck::tensor_layout::convolution::NHWC,
-                                                   ck::tensor_layout::convolution::KYXC,
-                                                   ck::tensor_layout::convolution::NHWK>(
-            do_verification,
-            init_method,
-            do_log,
-            time_kernel,
-            N,
-            K,
-            C,
-            std::vector<ck::index_t>{Hi, Wi},
-            std::vector<ck::index_t>{Y, X},
-            std::vector<ck::index_t>{Ho, Wo},
-            std::vector<ck::index_t>{conv_stride_h, conv_stride_w},
-            std::vector<ck::index_t>{conv_dilation_h, conv_dilation_w},
-            std::vector<ck::index_t>{in_left_pad_h, in_left_pad_w},
-            std::vector<ck::index_t>{in_right_pad_h, in_right_pad_w},
-            split_k);
-    }
-    else
-    {
-        throw std::runtime_error("wrong! this Conv data_type & layout is not implemented");
+        print_helper_msg();
+        return 1;
     }
 
-    return 0;
+    const auto params = ck::utils::conv::parse_conv_param(num_dim_spatial, 9, argv);
+
+    ck::index_t split_k = std::stoi(argv[8 + 1 + 4 + 6 * num_dim_spatial]);
+    split_k             = std::max(1, split_k);
+
+    using F32  = float;
+    using F16  = ck::half_t;
+    using BF16 = ck::bhalf_t;
+
+    using NWC   = ck::tensor_layout::convolution::NWC;
+    using NHWC  = ck::tensor_layout::convolution::NHWC;
+    using NDHWC = ck::tensor_layout::convolution::NDHWC;
+
+    using KXC   = ck::tensor_layout::convolution::KXC;
+    using KYXC  = ck::tensor_layout::convolution::KYXC;
+    using KZYXC = ck::tensor_layout::convolution::KZYXC;
+
+    using NWK   = ck::tensor_layout::convolution::NWK;
+    using NHWK  = ck::tensor_layout::convolution::NHWK;
+    using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+    constexpr auto I1 = ck::Number<1>{};
+    constexpr auto I2 = ck::Number<2>{};
+    constexpr auto I3 = ck::Number<3>{};
+
+    auto profile = [&](auto num_dim_spatial_tmp,
+                       auto in_layout,
+                       auto wei_layout,
+                       auto out_layout,
+                       auto in_type,
+                       auto wei_type,
+                       auto out_type) {
+        constexpr ck::index_t NDimSpatial = num_dim_spatial_tmp.value;
+
+        using InLayout  = decltype(in_layout);
+        using WeiLayout = decltype(wei_layout);
+        using OutLayout = decltype(out_layout);
+
+        using InDataType  = decltype(in_type);
+        using WeiDataType = decltype(wei_type);
+        using OutDataType = decltype(out_type);
+
+        bool pass = ck::profiler::profile_conv_bwd_weight_impl<NDimSpatial,
+                                                               InLayout,
+                                                               WeiLayout,
+                                                               OutLayout,
+                                                               InDataType,
+                                                               WeiDataType,
+                                                               OutDataType>(
+            do_verification, init_method, do_log, time_kernel, params, split_k);
+
+        return pass ? 0 : 1;
+    };
+
+    if(num_dim_spatial == 1 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_F32_BF16)
+        {
+            // fp32 atomic add is used for weight tensor in bf16 kernel
+            return profile(I1, NWC{}, KXC{}, NWK{}, BF16{}, F32{}, BF16{});
+        }
+    }
+    else if(num_dim_spatial == 2 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_F32_BF16)
+        {
+            // fp32 atomic add is used for weight tensor in bf16 kernel
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, BF16{}, F32{}, BF16{});
+        }
+    }
+    else if(num_dim_spatial == 3 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_F32_BF16)
+        {
+            // fp32 atomic add is used for weight tensor in bf16 kernel
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, BF16{}, F32{}, BF16{});
+        }
+    }
+
+    std::cout << "this data_type & layout is not implemented" << std::endl;
+
+    return 1;
 }
diff --git a/profiler/src/profile_conv_fwd.cpp b/profiler/src/profile_conv_fwd.cpp
new file mode 100644
index 0000000000..72b6a6b629
--- /dev/null
+++ b/profiler/src/profile_conv_fwd.cpp
@@ -0,0 +1,186 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+
+#include "profiler/include/profile_conv_fwd_impl.hpp"
+
+namespace {
+
+enum struct ConvLayout
+{
+    NCHW_KCYX_NKHW, // 0
+    NHWC_KYXC_NHWK, // 1
+};
+
+enum struct ConvDataType
+{
+    F32_F32_F32,    // 0
+    F16_F16_F16,    // 1
+    BF16_BF16_BF16, // 2
+    INT8_INT8_INT8, // 3
+};
+
+static void print_helper_msg()
+{
+    std::cout
+        // clang-format-off
+        << "arg1: tensor operation (conv_fwd: Convolution Forward)\n"
+        << "arg2: data type (0: Input fp32, Weight fp32, Output fp32\n"
+        << "                 1: Input fp16, Weight fp16, Output fp16\n"
+        << "                 2: Input bf16, Weight bf16, Output bf16\n"
+        << "                 3: Input int8, Weight int8, Output int8)\n"
+        << "arg3: tensor layout (0: Input[N, C, Hi, Wi], Weight[K, C, Y, X], Output[N, K, Ho, Wo]\n"
+        << "                     1: Input[N, Hi, Wi, C], Weight[K, Y, X, C], Output[N, Ho, Wo, "
+           "K])\n"
+        << "arg4: verification (0: no, 1: yes)\n"
+        << "arg5: initialization (0: no init, 1: integer value, 2: decimal value)\n"
+        << "arg6: print tensor value (0: no; 1: yes)\n"
+        << "arg7: time kernel (0: no, 1: yes)\n"
+        << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+    // clang-format-on
+}
+
+} // namespace
+
+int profile_conv_fwd(int argc, char* argv[])
+{
+    // 8 for control, 1 for num_dim_spatial
+    if(argc < 9)
+    {
+        print_helper_msg();
+        return 1;
+    }
+
+    const auto data_type       = static_cast<ConvDataType>(std::stoi(argv[2]));
+    const auto layout          = static_cast<ConvLayout>(std::stoi(argv[3]));
+    const bool do_verification = std::stoi(argv[4]);
+    const int init_method      = std::stoi(argv[5]);
+    const bool do_log          = std::stoi(argv[6]);
+    const bool time_kernel     = std::stoi(argv[7]);
+    const int num_dim_spatial  = std::stoi(argv[8]);
+
+    // 8 for control, 1 for num_dim_spatial, 4 for G/N/K/C, and 6 * num_dim_spatial
+    if(argc != 8 + 1 + 4 + 6 * num_dim_spatial)
+    {
+        print_helper_msg();
+        return 1;
+    }
+
+    const auto params = ck::utils::conv::parse_conv_param(num_dim_spatial, 9, argv);
+
+    using F32  = float;
+    using F16  = ck::half_t;
+    using BF16 = ck::bhalf_t;
+    using INT8 = int8_t;
+
+    using NWC   = ck::tensor_layout::convolution::NWC;
+    using NHWC  = ck::tensor_layout::convolution::NHWC;
+    using NDHWC = ck::tensor_layout::convolution::NDHWC;
+
+    using KXC   = ck::tensor_layout::convolution::KXC;
+    using KYXC  = ck::tensor_layout::convolution::KYXC;
+    using KZYXC = ck::tensor_layout::convolution::KZYXC;
+
+    using NWK   = ck::tensor_layout::convolution::NWK;
+    using NHWK  = ck::tensor_layout::convolution::NHWK;
+    using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+    constexpr auto I1 = ck::Number<1>{};
+    constexpr auto I2 = ck::Number<2>{};
+    constexpr auto I3 = ck::Number<3>{};
+
+    auto profile = [&](auto num_dim_spatial_tmp,
+                       auto in_layout,
+                       auto wei_layout,
+                       auto out_layout,
+                       auto in_type,
+                       auto wei_type,
+                       auto out_type) {
+        constexpr ck::index_t NDimSpatial = num_dim_spatial_tmp.value;
+
+        using InLayout  = decltype(in_layout);
+        using WeiLayout = decltype(wei_layout);
+        using OutLayout = decltype(out_layout);
+
+        using InDataType  = decltype(in_type);
+        using WeiDataType = decltype(wei_type);
+        using OutDataType = decltype(out_type);
+
+        bool pass = ck::profiler::profile_conv_fwd_impl<NDimSpatial,
+                                                        InLayout,
+                                                        WeiLayout,
+                                                        OutLayout,
+                                                        InDataType,
+                                                        WeiDataType,
+                                                        OutDataType>(
+            do_verification, init_method, do_log, time_kernel, params);
+
+        return pass ? 0 : 1;
+    };
+
+    if(num_dim_spatial == 1 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I1, NWC{}, KXC{}, NWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 2 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I2, NHWC{}, KYXC{}, NHWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 3 && layout == ConvLayout::NHWC_KYXC_NHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I3, NDHWC{}, KZYXC{}, NDHWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+
+    std::cout << "this data_type & layout is not implemented" << std::endl;
+
+    return 1;
+}
diff --git a/profiler/src/profile_convnd_bwd_data.cpp b/profiler/src/profile_convnd_bwd_data.cpp
deleted file mode 100644
index 7c387d375e..0000000000
--- a/profiler/src/profile_convnd_bwd_data.cpp
+++ /dev/null
@@ -1,229 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "profiler/include/profile_convnd_bwd_data_impl.hpp"
-
-namespace {
-
-enum struct ConvDataType
-{
-    F32_F32_F32,    // 0
-    F16_F16_F16,    // 1
-    BF16_BF16_BF16, // 2
-    INT8_INT8_INT8, // 3
-};
-
-enum struct ConvInputLayout
-{
-    NCHW, // 0
-    NHWC, // 1
-};
-
-enum struct ConvWeightLayout
-{
-    KCYX, // 0
-    KYXC, // 1
-};
-
-enum struct ConvOutputLayout
-{
-    NKHW, // 0
-    NHWK, // 1
-};
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, char* argv[], int arg_idx)
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    ck::utils::conv::ConvParams params;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // namespace
-
-int profile_convnd_bwd_data(int argc, char* argv[], int num_dim_spatial)
-{
-    const int preParams = 10;
-    int conv_args       = 3 + num_dim_spatial * 6;
-    int cmdline_nargs   = conv_args + preParams;
-    if(cmdline_nargs != argc)
-    {
-        printf("arg1: tensor operation (conv[1|2|3]d_bwd_data: BackwardConvolution)\n");
-        printf("arg2: data type (0: fp32; 1: fp16)\n");
-        printf("arg3: input tensor layout (0: NCHW; 1: NHWC)\n");
-        printf("arg4: weight tensor layout (0: KCYX; 1: KYXC)\n");
-        printf("arg5: output tensor layout (0: NKHW; 1: NHWK)\n");
-        printf("arg6: verification (0: no; 1: yes)\n");
-        printf("arg7: initialization (0: no init; 1: integer value; 2: decimal value)\n");
-        printf("arg8: print tensor value (0: no; 1: yes)\n");
-        printf("arg9: time kernel (0=n0, 1=yes)\n");
-        printf("arg10 to 24: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-               "RightPx\n");
-        return 1;
-    }
-
-    const auto data_type       = static_cast<ConvDataType>(std::stoi(argv[2]));
-    const auto in_layout       = static_cast<ConvInputLayout>(std::stoi(argv[3]));
-    const auto wei_layout      = static_cast<ConvWeightLayout>(std::stoi(argv[4]));
-    const auto out_layout      = static_cast<ConvOutputLayout>(std::stoi(argv[5]));
-    const bool do_verification = std::stoi(argv[6]);
-    const int init_method      = std::stoi(argv[7]);
-    const bool do_log          = std::stoi(argv[8]);
-    const bool time_kernel     = std::stoi(argv[9]);
-
-    ck::utils::conv::ConvParams params = parse_conv_params(num_dim_spatial, argv, preParams);
-
-    auto Run = [&](auto input_type, auto wei_type, auto out_type, auto acc_type) {
-        using InDataType  = decltype(input_type);
-        using WeiDataType = decltype(wei_type);
-        using OutDataType = decltype(out_type);
-        using AccDataType = decltype(acc_type);
-
-        switch(num_dim_spatial)
-        {
-        case 1:
-            ck::profiler::profile_convnd_bwd_data_impl<1,
-                                                       InDataType,
-                                                       WeiDataType,
-                                                       OutDataType,
-                                                       AccDataType,
-                                                       ck::tensor_layout::convolution::NWC,
-                                                       ck::tensor_layout::convolution::KXC,
-                                                       ck::tensor_layout::convolution::NWK>(
-                do_verification,
-                init_method,
-                do_log,
-                time_kernel,
-                params.N_,
-                params.K_,
-                params.C_,
-                params.input_spatial_lengths_,
-                params.filter_spatial_lengths_,
-                params.GetOutputSpatialLengths(),
-                params.conv_filter_strides_,
-                params.conv_filter_dilations_,
-                params.input_left_pads_,
-                params.input_right_pads_);
-            break;
-
-        case 2:
-            ck::profiler::profile_convnd_bwd_data_impl<2,
-                                                       InDataType,
-                                                       WeiDataType,
-                                                       OutDataType,
-                                                       AccDataType,
-                                                       ck::tensor_layout::convolution::NHWC,
-                                                       ck::tensor_layout::convolution::KYXC,
-                                                       ck::tensor_layout::convolution::NHWK>(
-                do_verification,
-                init_method,
-                do_log,
-                time_kernel,
-                params.N_,
-                params.K_,
-                params.C_,
-                params.input_spatial_lengths_,
-                params.filter_spatial_lengths_,
-                params.GetOutputSpatialLengths(),
-                params.conv_filter_strides_,
-                params.conv_filter_dilations_,
-                params.input_left_pads_,
-                params.input_right_pads_);
-            break;
-
-        case 3:
-            ck::profiler::profile_convnd_bwd_data_impl<3,
-                                                       InDataType,
-                                                       WeiDataType,
-                                                       OutDataType,
-                                                       AccDataType,
-                                                       ck::tensor_layout::convolution::NDHWC,
-                                                       ck::tensor_layout::convolution::KZYXC,
-                                                       ck::tensor_layout::convolution::NDHWK>(
-                do_verification,
-                init_method,
-                do_log,
-                time_kernel,
-                params.N_,
-                params.K_,
-                params.C_,
-                params.input_spatial_lengths_,
-                params.filter_spatial_lengths_,
-                params.GetOutputSpatialLengths(),
-                params.conv_filter_strides_,
-                params.conv_filter_dilations_,
-                params.input_left_pads_,
-                params.input_right_pads_);
-            break;
-
-        default: break;
-        }
-    };
-    if(data_type == ConvDataType::F32_F32_F32 && in_layout == ConvInputLayout::NHWC &&
-       wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(float{}, float{}, float{}, float{});
-    }
-    else if(data_type == ConvDataType::F16_F16_F16 && in_layout == ConvInputLayout::NHWC &&
-            wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(ck::half_t{}, ck::half_t{}, ck::half_t{}, float{});
-    }
-    else if(data_type == ConvDataType::BF16_BF16_BF16 && in_layout == ConvInputLayout::NHWC &&
-            wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(ck::bhalf_t{}, ck::bhalf_t{}, ck::bhalf_t{}, float{});
-    }
-    else if(data_type == ConvDataType::INT8_INT8_INT8 && in_layout == ConvInputLayout::NHWC &&
-            wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(int8_t{}, int8_t{}, int8_t{}, int32_t{});
-    }
-    else
-    {
-        std::cout << "wrong! this Conv data_type & layout is not implemented" << std::endl;
-        return 1;
-    }
-
-    return 0;
-}
diff --git a/profiler/src/profile_convnd_bwd_weight.cpp b/profiler/src/profile_convnd_bwd_weight.cpp
deleted file mode 100644
index 741d9ac656..0000000000
--- a/profiler/src/profile_convnd_bwd_weight.cpp
+++ /dev/null
@@ -1,226 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-
-#include "profiler/include/profile_convnd_bwd_weight_impl.hpp"
-
-namespace {
-
-enum struct ConvDataType
-{
-    F32_F32_F32,    // 0
-    F16_F16_F16,    // 1
-    BF16_BF16_BF16, // 2
-};
-
-enum struct ConvInputLayout
-{
-    NCHW, // 0
-    NHWC, // 1
-};
-
-enum struct ConvWeightLayout
-{
-    KCYX, // 0
-    KYXC, // 1
-};
-
-enum struct ConvOutputLayout
-{
-    NKHW, // 0
-    NHWK, // 1
-};
-ck::utils::conv::ConvParams parse_conv_params(int num_dim_spatial, char* argv[], int arg_idx)
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    ck::utils::conv::ConvParams params;
-
-    params.num_dim_spatial_ = num_dim_spatial;
-    params.N_               = std::stoi(argv[arg_idx++]);
-    params.K_               = std::stoi(argv[arg_idx++]);
-    params.C_               = std::stoi(argv[arg_idx++]);
-
-    params.filter_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.filter_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_spatial_lengths_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_spatial_lengths_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_strides_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_strides_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.conv_filter_dilations_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.conv_filter_dilations_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_left_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_left_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-    params.input_right_pads_.resize(num_dim_spatial);
-    for(int i = 0; i < num_dim_spatial; ++i)
-    {
-        params.input_right_pads_[i] = std::stoi(argv[arg_idx++]);
-    }
-
-    return params;
-}
-
-} // namespace
-
-int profile_convnd_bwd_weight(int argc, char* argv[], int num_dim_spatial)
-{
-    const int preParams = 11;
-    int conv_args       = 3 + num_dim_spatial * 6;
-    int cmdline_nargs   = conv_args + preParams;
-    if(cmdline_nargs != argc)
-    {
-        printf("arg1: tensor operation (convnd[1|2|3]d_bwd_weight: BackwardConvolution)\n");
-        printf("arg2: data type (0: fp32; 1: fp16, 2: bf16)\n");
-        printf("arg3: input tensor layout (0: NCHW; 1: NHWC)\n");
-        printf("arg4: weight tensor layout (0: KCYX; 1: KYXC)\n");
-        printf("arg5: output tensor layout (0: NKHW; 1: NHWK)\n");
-        printf("arg6: verification (0: no; 1: yes)\n");
-        printf("arg7: initialization (0: no init; 1: integer value; 2: decimal value)\n");
-        printf("arg8: print tensor value (0: no; 1: yes)\n");
-        printf("arg9: time kernel (0=n0, 1=yes)\n");
-        printf("arg10: splitk\n");
-        printf("arg11 to 25: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-               "RightPx\n");
-        return 1;
-    }
-
-    const auto data_type       = static_cast<ConvDataType>(std::stoi(argv[2]));
-    const auto in_layout       = static_cast<ConvInputLayout>(std::stoi(argv[3]));
-    const auto wei_layout      = static_cast<ConvWeightLayout>(std::stoi(argv[4]));
-    const auto out_layout      = static_cast<ConvOutputLayout>(std::stoi(argv[5]));
-    const bool do_verification = std::stoi(argv[6]);
-    const int init_method      = std::stoi(argv[7]);
-    const bool do_log          = std::stoi(argv[8]);
-    const bool time_kernel     = std::stoi(argv[9]);
-
-    ck::index_t split_k = std::stoi(argv[10]);
-    split_k             = std::max(1, split_k);
-
-    ck::utils::conv::ConvParams params = parse_conv_params(num_dim_spatial, argv, preParams);
-
-    auto Run = [&](auto input_type, auto wei_type, auto out_type) {
-        using InDataType  = decltype(input_type);
-        using WeiDataType = decltype(wei_type);
-        using OutDataType = decltype(out_type);
-
-        switch(num_dim_spatial)
-        {
-        case 1:
-            ck::profiler::profile_convnd_bwd_weight_impl<1,
-                                                         InDataType,
-                                                         WeiDataType,
-                                                         OutDataType,
-                                                         ck::tensor_layout::convolution::NWC,
-                                                         ck::tensor_layout::convolution::KXC,
-                                                         ck::tensor_layout::convolution::NWK>(
-                do_verification,
-                init_method,
-                do_log,
-                time_kernel,
-                params.N_,
-                params.K_,
-                params.C_,
-                params.input_spatial_lengths_,
-                params.filter_spatial_lengths_,
-                params.GetOutputSpatialLengths(),
-                params.conv_filter_strides_,
-                params.conv_filter_dilations_,
-                params.input_left_pads_,
-                params.input_right_pads_,
-                split_k);
-            break;
-
-        case 2:
-            ck::profiler::profile_convnd_bwd_weight_impl<2,
-                                                         InDataType,
-                                                         WeiDataType,
-                                                         OutDataType,
-                                                         ck::tensor_layout::convolution::NHWC,
-                                                         ck::tensor_layout::convolution::KYXC,
-                                                         ck::tensor_layout::convolution::NHWK>(
-                do_verification,
-                init_method,
-                do_log,
-                time_kernel,
-                params.N_,
-                params.K_,
-                params.C_,
-                params.input_spatial_lengths_,
-                params.filter_spatial_lengths_,
-                params.GetOutputSpatialLengths(),
-                params.conv_filter_strides_,
-                params.conv_filter_dilations_,
-                params.input_left_pads_,
-                params.input_right_pads_,
-                split_k);
-            break;
-
-        case 3:
-            ck::profiler::profile_convnd_bwd_weight_impl<3,
-                                                         InDataType,
-                                                         WeiDataType,
-                                                         OutDataType,
-                                                         ck::tensor_layout::convolution::NDHWC,
-                                                         ck::tensor_layout::convolution::KZYXC,
-                                                         ck::tensor_layout::convolution::NDHWK>(
-                do_verification,
-                init_method,
-                do_log,
-                time_kernel,
-                params.N_,
-                params.K_,
-                params.C_,
-                params.input_spatial_lengths_,
-                params.filter_spatial_lengths_,
-                params.GetOutputSpatialLengths(),
-                params.conv_filter_strides_,
-                params.conv_filter_dilations_,
-                params.input_left_pads_,
-                params.input_right_pads_,
-                split_k);
-            break;
-
-        default: break;
-        }
-    };
-    if(data_type == ConvDataType::F32_F32_F32 && in_layout == ConvInputLayout::NHWC &&
-       wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(float{}, float{}, float{});
-    }
-    else if(data_type == ConvDataType::F16_F16_F16 && in_layout == ConvInputLayout::NHWC &&
-            wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(ck::half_t{}, ck::half_t{}, ck::half_t{});
-    }
-    else if(data_type == ConvDataType::BF16_BF16_BF16 && in_layout == ConvInputLayout::NHWC &&
-            wei_layout == ConvWeightLayout::KYXC && out_layout == ConvOutputLayout::NHWK)
-    {
-        Run(ck::bhalf_t{}, ck::bhalf_t{}, ck::bhalf_t{});
-    }
-    else
-    {
-        std::cout << "wrong! this Conv data_type & layout is not implemented" << std::endl;
-        return 1;
-    }
-
-    return 0;
-}
diff --git a/profiler/src/profile_convnd_fwd.cpp b/profiler/src/profile_convnd_fwd.cpp
deleted file mode 100644
index 8223be160e..0000000000
--- a/profiler/src/profile_convnd_fwd.cpp
+++ /dev/null
@@ -1,359 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <functional>
-#include <iostream>
-#include <memory>
-#include <string>
-#include <vector>
-
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/conv_util.hpp"
-#include "ck/library/utility/fill.hpp"
-
-namespace {
-
-enum struct ConvDataType
-{
-    F32_F32_F32,    // 0
-    F16_F16_F16,    // 1
-    BF16_BF16_BF16, // 2
-    INT8_INT8_INT8, // 3
-};
-
-enum struct ConvDataLayout
-{
-    NCHW, // 0
-    NHWC, // 1
-};
-
-namespace ctl = ck::tensor_layout::convolution;
-
-template <int NDim, ConvDataLayout DataLayout>
-struct ConvolutionLayouts;
-
-template <>
-struct ConvolutionLayouts<1, ConvDataLayout::NHWC>
-{
-    typedef ctl::NWC Input;
-    typedef ctl::KXC Weight;
-    typedef ctl::NWK Output;
-};
-template <>
-struct ConvolutionLayouts<2, ConvDataLayout::NHWC>
-{
-    typedef ctl::NHWC Input;
-    typedef ctl::KYXC Weight;
-    typedef ctl::NHWK Output;
-};
-template <>
-struct ConvolutionLayouts<3, ConvDataLayout::NHWC>
-{
-    typedef ctl::NDHWC Input;
-    typedef ctl::KZYXC Weight;
-    typedef ctl::NDHWK Output;
-};
-template <>
-struct ConvolutionLayouts<1, ConvDataLayout::NCHW>
-{
-    typedef ctl::NCW Input;
-    typedef ctl::KCX Weight;
-    typedef ctl::NKW Output;
-};
-template <>
-struct ConvolutionLayouts<2, ConvDataLayout::NCHW>
-{
-    typedef ctl::NCHW Input;
-    typedef ctl::KCYX Weight;
-    typedef ctl::NKHW Output;
-};
-template <>
-struct ConvolutionLayouts<3, ConvDataLayout::NCHW>
-{
-    typedef ctl::NCDHW Input;
-    typedef ctl::KCZYX Weight;
-    typedef ctl::NKDHW Output;
-};
-
-void print_use_msg()
-{
-    std::cout << "arg1: tensor operation (conv_fwd: ForwardConvolution)\n"
-              << "arg2: data type (0: fp32; 1: fp16, 2: bf16, 3: int8)\n"
-              << "arg3: data layout (0: NCHW; 1: NHWC)\n"
-              << "arg4: verification (0=no, 1=yes)\n"
-              << "arg5: initialization (0=no init, 1=integer value, 2=decimal value)\n"
-              << "arg6: print tensor value (0: no; 1: yes)\n"
-              << "arg7: run kernel # of times (>1)\n"
-              << "arg8: N spatial dimensions (default 2)\n"
-              << "Following arguments (depending on number of spatial dims):\n"
-              << " N, K, C, \n"
-              << " <filter spatial dimensions>, (ie Y, X for 2D)\n"
-              << " <input image spatial dimensions>, (ie Hi, Wi for 2D)\n"
-              << " <strides>, (ie Sy, Sx for 2D)\n"
-              << " <dilations>, (ie Dy, Dx for 2D)\n"
-              << " <left padding>, (ie LeftPy, LeftPx for 2D)\n"
-              << " <right padding>, (ie RightPy, RightPx for 2D)\n"
-              << std::endl;
-}
-
-ck::utils::conv::ConvParams parse_params(int num_dim_spatial, int argc, char* argv[])
-{
-    // (N, K, C) + num_dim_spatial * 6 (filter, input, strides, dilations, pad left, pad right)
-    int conv_args     = 3 + num_dim_spatial * 6;
-    int cmdline_nargs = conv_args + 9;
-    if(cmdline_nargs != argc)
-    {
-        print_use_msg();
-        exit(1);
-    }
-    int arg_idx = 9;
-
-    return ck::utils::conv::parse_conv_params(num_dim_spatial, arg_idx, argv);
-}
-
-template <int NDim,
-          typename InDataType,
-          typename WeiDataType,
-          typename OutDataType,
-          typename ConvLayouts>
-void profile_convnd_instances_impl(const ck::utils::conv::ConvParams& params,
-                                   bool do_verification,
-                                   bool do_log,
-                                   bool time_kernel,
-                                   int init_method,
-                                   ConvLayouts)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-
-    std::unique_ptr<OpInstance<OutDataType, InDataType, WeiDataType>> conv_instance;
-
-    switch(init_method)
-    {
-    case 0:
-        conv_instance =
-            std::make_unique<conv::ConvFwdOpInstance<InDataType,
-                                                     WeiDataType,
-                                                     OutDataType,
-                                                     typename ConvLayouts::Input,
-                                                     typename ConvLayouts::Weight,
-                                                     typename ConvLayouts::Output>>(params, false);
-        break;
-    case 1:
-        conv_instance = std::make_unique<
-            conv::ConvFwdOpInstance<InDataType,
-                                    WeiDataType,
-                                    OutDataType,
-                                    typename ConvLayouts::Input,
-                                    typename ConvLayouts::Weight,
-                                    typename ConvLayouts::Output,
-                                    ck::tensor_operation::element_wise::PassThrough,
-                                    ck::tensor_operation::element_wise::PassThrough,
-                                    ck::tensor_operation::element_wise::PassThrough,
-                                    ck::utils::FillUniformDistributionIntegerValue<int>,
-                                    ck::utils::FillUniformDistributionIntegerValue<int>>>(
-            params,
-            true,
-            ck::utils::FillUniformDistributionIntegerValue<int>{},
-            ck::utils::FillUniformDistributionIntegerValue<int>{});
-        break;
-    case 2:
-        conv_instance = std::make_unique<
-            conv::ConvFwdOpInstance<InDataType,
-                                    WeiDataType,
-                                    OutDataType,
-                                    typename ConvLayouts::Input,
-                                    typename ConvLayouts::Weight,
-                                    typename ConvLayouts::Output,
-                                    ck::tensor_operation::element_wise::PassThrough,
-                                    ck::tensor_operation::element_wise::PassThrough,
-                                    ck::tensor_operation::element_wise::PassThrough,
-                                    ck::utils::FillUniformDistribution<InDataType>,
-                                    ck::utils::FillUniformDistribution<WeiDataType>>>(
-            params,
-            true,
-            ck::utils::FillUniformDistribution<InDataType>{},
-            ck::utils::FillUniformDistribution<WeiDataType>{});
-        break;
-    default: throw std::runtime_error("Unsupported init method!");
-    }
-
-    auto reference_conv_fwd_fun = std::bind(
-        conv::run_reference_convolution_forward<NDim, InDataType, WeiDataType, OutDataType>,
-        params,
-        _1,
-        _2,
-        _3);
-
-    OpInstanceRunEngine<InDataType, WeiDataType, OutDataType> run_engine(
-        *conv_instance, reference_conv_fwd_fun, do_verification);
-
-    auto best_conf = run_engine.Profile(
-        conv::ConvolutionFwdInstances<InDataType, WeiDataType, OutDataType>::template Get<NDim>(),
-        time_kernel,
-        do_verification,
-        do_log);
-
-    std::cout << "Best configuration parameters:"
-              << "\nname: " << best_conf.best_op_name << "\navg_time: " << best_conf.best_avg_time
-              << "\ntflops: " << best_conf.best_tflops << "\nGB/s: " << best_conf.best_gb_per_sec
-              << std::endl;
-}
-
-template <int NDim>
-void profile_convnd_instances(ConvDataType data_type,
-                              ConvDataLayout data_layout,
-                              const ck::utils::conv::ConvParams& params,
-                              bool do_verification,
-                              bool do_log,
-                              bool time_kernel,
-                              int init_method)
-{
-    switch(data_layout)
-    {
-    case ConvDataLayout::NHWC: {
-        switch(data_type)
-        {
-        case ConvDataType::F32_F32_F32:
-            profile_convnd_instances_impl<NDim, float, float, float>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NHWC>{});
-            break;
-        case ConvDataType::F16_F16_F16:
-            profile_convnd_instances_impl<NDim, ck::half_t, ck::half_t, ck::half_t>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NHWC>{});
-            break;
-        case ConvDataType::BF16_BF16_BF16:
-            profile_convnd_instances_impl<NDim, ck::bhalf_t, ck::bhalf_t, ck::bhalf_t>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NHWC>{});
-            break;
-        case ConvDataType::INT8_INT8_INT8:
-            profile_convnd_instances_impl<NDim, int8_t, int8_t, int8_t>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NHWC>{});
-            break;
-        }
-        break;
-    }
-    case ConvDataLayout::NCHW: {
-        switch(data_type)
-        {
-        case ConvDataType::F32_F32_F32:
-            profile_convnd_instances_impl<NDim, float, float, float>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NCHW>{});
-            break;
-        case ConvDataType::F16_F16_F16:
-            profile_convnd_instances_impl<NDim, ck::half_t, ck::half_t, ck::half_t>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NCHW>{});
-            break;
-        case ConvDataType::BF16_BF16_BF16:
-            profile_convnd_instances_impl<NDim, ck::bhalf_t, ck::bhalf_t, ck::bhalf_t>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NCHW>{});
-            break;
-        case ConvDataType::INT8_INT8_INT8:
-            profile_convnd_instances_impl<NDim, int8_t, int8_t, int8_t>(
-                params,
-                do_verification,
-                do_log,
-                time_kernel,
-                init_method,
-                ConvolutionLayouts<NDim, ConvDataLayout::NCHW>{});
-            break;
-        }
-        break;
-    }
-    }
-}
-
-} // namespace
-
-int profile_convnd_fwd(int argc, char* argv[])
-{
-    using namespace ck::utils::conv;
-
-    ConvDataType data_type{ConvDataType::F32_F32_F32};
-    ConvDataLayout data_layout{ConvDataLayout::NHWC};
-    bool do_verification{true};
-    int init_method{2};
-    bool do_log{false};
-    bool time_kernel{false};
-    int num_dim_spatial{2};
-    ConvParams params;
-
-    if(argc >= 4)
-    {
-        data_type   = static_cast<ConvDataType>(std::stoi(argv[2]));
-        data_layout = static_cast<ConvDataLayout>(std::stoi(argv[3]));
-    }
-    if(argc >= 9)
-    {
-        do_verification = std::stoi(argv[4]);
-        init_method     = std::stoi(argv[5]);
-        do_log          = std::stoi(argv[6]);
-        time_kernel     = std::stoi(argv[7]);
-        num_dim_spatial = std::stoi(argv[8]);
-    }
-    if(argc >= 10)
-    {
-        params = parse_params(num_dim_spatial, argc, argv);
-    }
-
-    // TODO Print nice message what is being profiled.
-
-    switch(num_dim_spatial)
-    {
-    case 1:
-        profile_convnd_instances<1>(
-            data_type, data_layout, params, do_verification, do_log, time_kernel, init_method);
-        break;
-    case 2:
-        profile_convnd_instances<2>(
-            data_type, data_layout, params, do_verification, do_log, time_kernel, init_method);
-        break;
-    case 3:
-        profile_convnd_instances<3>(
-            data_type, data_layout, params, do_verification, do_log, time_kernel, init_method);
-        break;
-    default:
-        throw std::runtime_error("profile_conv_fwd: unsupported num_dim_spatial value: " +
-                                 std::to_string(num_dim_spatial));
-    }
-
-    return 0;
-}
diff --git a/profiler/src/profile_gemm.cpp b/profiler/src/profile_gemm.cpp
index 624f3dbf61..f53f478197 100644
--- a/profiler/src/profile_gemm.cpp
+++ b/profiler/src/profile_gemm.cpp
@@ -24,21 +24,27 @@ enum struct GemmDataType
     INT8_INT8_INT8, // 3
 };
 
+static void print_helper_msg()
+{
+    std::cout << "arg1: tensor operation (gemm: GEMM)\n"
+              << "arg2: data type (0: fp32; 1: fp16; 2: bf16; 3: int8)\n"
+              << "arg3: matrix layout (0: A[m, k] * B[k, n] = C[m, n];\n"
+              << "                     1: A[m, k] * B[n, k] = C[m, n];\n"
+              << "                     2: A[k, m] * B[k, n] = C[m, n];\n"
+              << "                     3: A[k, m] * B[n, k] = C[m, n])\n"
+              << "arg4: verification (0: no; 1: yes)\n"
+              << "arg5: initialization (0: no init; 1: integer value; 2: decimal value)\n"
+              << "arg6: print tensor value (0: no; 1: yes)\n"
+              << "arg7: time kernel (0: no, 1: yes)\n"
+              << "arg8 to 13: M, N, K, StrideA, StrideB, StrideC\n"
+              << std::endl;
+}
+
 int profile_gemm(int argc, char* argv[])
 {
     if(argc != 14)
     {
-        printf("arg1: tensor operation (gemm: GEMM)\n");
-        printf("arg2: data type (0: fp32; 1: fp16; 2: bf16; 3: int8)\n");
-        printf("arg3: matrix layout (0: A[m, k] * B[k, n] = C[m, n];\n");
-        printf("                     1: A[m, k] * B[n, k] = C[m, n];\n");
-        printf("                     2: A[k, m] * B[k, n] = C[m, n];\n");
-        printf("                     3: A[k, m] * B[n, k] = C[m, n])\n");
-        printf("arg4: verification (0: no; 1: yes)\n");
-        printf("arg5: initialization (0: no init; 1: integer value; 2: decimal value)\n");
-        printf("arg6: print tensor value (0: no; 1: yes)\n");
-        printf("arg7: time kernel (0=no, 1=yes)\n");
-        printf("arg8 to 13: M, N, K, StrideA, StrideB, StrideC\n");
+        print_helper_msg();
         exit(1);
     }
 
@@ -109,67 +115,67 @@ int profile_gemm(int argc, char* argv[])
 
     if(data_type == GemmDataType::F32_F32_F32 && layout == GemmMatrixLayout::MK_KN_MN)
     {
-        return profile(F32{}, F32{}, F32{}, F32{}, Row{}, Row{}, Row{});
+        return profile(Row{}, Row{}, Row{}, F32{}, F32{}, F32{}, F32{});
     }
     else if(data_type == GemmDataType::F32_F32_F32 && layout == GemmMatrixLayout::MK_NK_MN)
     {
-        return profile(F32{}, F32{}, F32{}, F32{}, Row{}, Col{}, Row{});
+        return profile(Row{}, Col{}, Row{}, F32{}, F32{}, F32{}, F32{});
     }
     else if(data_type == GemmDataType::F32_F32_F32 && layout == GemmMatrixLayout::KM_KN_MN)
     {
-        return profile(F32{}, F32{}, F32{}, F32{}, Col{}, Row{}, Row{});
+        return profile(Col{}, Row{}, Row{}, F32{}, F32{}, F32{}, F32{});
     }
     else if(data_type == GemmDataType::F32_F32_F32 && layout == GemmMatrixLayout::KM_NK_MN)
     {
-        return profile(F32{}, F32{}, F32{}, F32{}, Col{}, Col{}, Row{});
+        return profile(Col{}, Col{}, Row{}, F32{}, F32{}, F32{}, F32{});
     }
     else if(data_type == GemmDataType::F16_F16_F16 && layout == GemmMatrixLayout::MK_KN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, Row{}, Row{}, Row{});
+        return profile(Row{}, Row{}, Row{}, F16{}, F16{}, F32{}, F16{});
     }
     else if(data_type == GemmDataType::F16_F16_F16 && layout == GemmMatrixLayout::MK_NK_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, Row{}, Col{}, Row{});
+        return profile(Row{}, Col{}, Row{}, F16{}, F16{}, F32{}, F16{});
     }
     else if(data_type == GemmDataType::F16_F16_F16 && layout == GemmMatrixLayout::KM_KN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, Col{}, Row{}, Row{});
+        return profile(Col{}, Row{}, Row{}, F16{}, F16{}, F32{}, F16{});
     }
     else if(data_type == GemmDataType::F16_F16_F16 && layout == GemmMatrixLayout::KM_NK_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, Col{}, Col{}, Row{});
+        return profile(Col{}, Col{}, Row{}, F16{}, F16{}, F32{}, F16{});
     }
     else if(data_type == GemmDataType::BF16_BF16_BF16 && layout == GemmMatrixLayout::MK_KN_MN)
     {
-        return profile(BF16{}, BF16{}, F32{}, BF16{}, Row{}, Row{}, Row{});
+        return profile(Row{}, Row{}, Row{}, BF16{}, BF16{}, F32{}, BF16{});
     }
     else if(data_type == GemmDataType::BF16_BF16_BF16 && layout == GemmMatrixLayout::MK_NK_MN)
     {
-        return profile(BF16{}, BF16{}, F32{}, BF16{}, Row{}, Col{}, Row{});
+        return profile(Row{}, Col{}, Row{}, BF16{}, BF16{}, F32{}, BF16{});
     }
     else if(data_type == GemmDataType::BF16_BF16_BF16 && layout == GemmMatrixLayout::KM_KN_MN)
     {
-        return profile(BF16{}, BF16{}, F32{}, BF16{}, Col{}, Row{}, Row{});
+        return profile(Col{}, Row{}, Row{}, BF16{}, BF16{}, F32{}, BF16{});
     }
     else if(data_type == GemmDataType::BF16_BF16_BF16 && layout == GemmMatrixLayout::KM_NK_MN)
     {
-        return profile(BF16{}, BF16{}, F32{}, BF16{}, Col{}, Col{}, Row{});
+        return profile(Col{}, Col{}, Row{}, BF16{}, BF16{}, F32{}, BF16{});
     }
     else if(data_type == GemmDataType::INT8_INT8_INT8 && layout == GemmMatrixLayout::MK_KN_MN)
     {
-        return profile(INT8{}, INT8{}, INT32{}, INT8{}, Row{}, Row{}, Row{});
+        return profile(Row{}, Row{}, Row{}, INT8{}, INT8{}, INT32{}, INT8{});
     }
     else if(data_type == GemmDataType::INT8_INT8_INT8 && layout == GemmMatrixLayout::MK_NK_MN)
     {
-        return profile(INT8{}, INT8{}, INT32{}, INT8{}, Row{}, Col{}, Row{});
+        return profile(Row{}, Col{}, Row{}, INT8{}, INT8{}, INT32{}, INT8{});
     }
     else if(data_type == GemmDataType::INT8_INT8_INT8 && layout == GemmMatrixLayout::KM_KN_MN)
     {
-        return profile(INT8{}, INT8{}, INT32{}, INT8{}, Col{}, Row{}, Row{});
+        return profile(Col{}, Row{}, Row{}, INT8{}, INT8{}, INT32{}, INT8{});
     }
     else if(data_type == GemmDataType::INT8_INT8_INT8 && layout == GemmMatrixLayout::KM_NK_MN)
     {
-        return profile(INT8{}, INT8{}, INT32{}, INT8{}, Col{}, Col{}, Row{});
+        return profile(Col{}, Col{}, Row{}, INT8{}, INT8{}, INT32{}, INT8{});
     }
     else
     {
diff --git a/profiler/src/profile_gemm_add_add_fastgelu.cpp b/profiler/src/profile_gemm_add_add_fastgelu.cpp
index a381222cbc..8d3d280d7b 100644
--- a/profiler/src/profile_gemm_add_add_fastgelu.cpp
+++ b/profiler/src/profile_gemm_add_add_fastgelu.cpp
@@ -75,7 +75,9 @@ int profile_gemm_add_add_fastgelu(int argc, char* argv[])
                        auto e_type,
                        auto a_layout,
                        auto b_layout,
-                       auto de_layout) {
+                       auto d0_layout,
+                       auto d1_layout,
+                       auto e_layout) {
         using ADataType   = decltype(a_type);
         using BDataType   = decltype(b_type);
         using AccDataType = decltype(acc_type);
@@ -85,13 +87,15 @@ int profile_gemm_add_add_fastgelu(int argc, char* argv[])
 
         using ALayout  = decltype(a_layout);
         using BLayout  = decltype(b_layout);
-        using DELayout = decltype(de_layout);
+        using D0Layout = decltype(d0_layout);
+        using D1Layout = decltype(d1_layout);
+        using ELayout  = decltype(e_layout);
 
         const int DefaultStrideA  = ck::is_same_v<ALayout, Row> ? K : M;
         const int DefaultStrideB  = ck::is_same_v<BLayout, Row> ? N : K;
-        const int DefaultStrideD0 = ck::is_same_v<DELayout, Row> ? N : M;
-        const int DefaultStrideD1 = ck::is_same_v<DELayout, Row> ? N : M;
-        const int DefaultStrideE  = ck::is_same_v<DELayout, Row> ? N : M;
+        const int DefaultStrideD0 = ck::is_same_v<D0Layout, Row> ? N : M;
+        const int DefaultStrideD1 = ck::is_same_v<D1Layout, Row> ? N : M;
+        const int DefaultStrideE  = ck::is_same_v<ELayout, Row> ? N : M;
 
         bool pass = ck::profiler::profile_gemm_add_add_fastgelu_impl<ADataType,
                                                                      BDataType,
@@ -101,7 +105,9 @@ int profile_gemm_add_add_fastgelu(int argc, char* argv[])
                                                                      EDataType,
                                                                      ALayout,
                                                                      BLayout,
-                                                                     DELayout>(
+                                                                     D0Layout,
+                                                                     D1Layout,
+                                                                     ELayout>(
             do_verification,
             init_method,
             do_log,
@@ -120,22 +126,22 @@ int profile_gemm_add_add_fastgelu(int argc, char* argv[])
 
     if(data_type == MatrixDataType::F16_F16_F16_F16_F16 && layout == MatrixLayout::MK_KN_MN_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Row{}, Row{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Row{}, Row{}, Row{}, Row{}, Row{});
     }
     else if(data_type == MatrixDataType::F16_F16_F16_F16_F16 &&
             layout == MatrixLayout::MK_NK_MN_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Row{}, Col{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Row{}, Col{}, Row{}, Row{}, Row{});
     }
     else if(data_type == MatrixDataType::F16_F16_F16_F16_F16 &&
             layout == MatrixLayout::KM_KN_MN_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Col{}, Row{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Col{}, Row{}, Row{}, Row{}, Row{});
     }
     else if(data_type == MatrixDataType::F16_F16_F16_F16_F16 &&
             layout == MatrixLayout::KM_NK_MN_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Col{}, Col{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, F16{}, Col{}, Col{}, Row{}, Row{}, Row{});
     }
     else
     {
diff --git a/profiler/src/profile_gemm_bilinear.cpp b/profiler/src/profile_gemm_bilinear.cpp
index 14c577897c..4f7e5a800d 100644
--- a/profiler/src/profile_gemm_bilinear.cpp
+++ b/profiler/src/profile_gemm_bilinear.cpp
@@ -77,21 +77,23 @@ int profile_gemm_bilinear(int argc, char* argv[])
                        auto e_type,
                        auto a_layout,
                        auto b_layout,
-                       auto de_layout) {
+                       auto d_layout,
+                       auto e_layout) {
         using ADataType   = decltype(a_type);
         using BDataType   = decltype(b_type);
         using AccDataType = decltype(acc_type);
         using DDataType   = decltype(d_type);
         using EDataType   = decltype(e_type);
 
-        using ALayout  = decltype(a_layout);
-        using BLayout  = decltype(b_layout);
-        using DELayout = decltype(de_layout);
+        using ALayout = decltype(a_layout);
+        using BLayout = decltype(b_layout);
+        using DLayout = decltype(d_layout);
+        using ELayout = decltype(e_layout);
 
         const int DefaultStrideA = ck::is_same_v<ALayout, Row> ? K : M;
         const int DefaultStrideB = ck::is_same_v<BLayout, Row> ? N : K;
-        const int DefaultStrideD = ck::is_same_v<DELayout, Row> ? N : M;
-        const int DefaultStrideE = ck::is_same_v<DELayout, Row> ? N : M;
+        const int DefaultStrideD = ck::is_same_v<DLayout, Row> ? N : M;
+        const int DefaultStrideE = ck::is_same_v<ELayout, Row> ? N : M;
 
         bool pass = ck::profiler::profile_gemm_bilinear_impl<ADataType,
                                                              BDataType,
@@ -100,7 +102,8 @@ int profile_gemm_bilinear(int argc, char* argv[])
                                                              EDataType,
                                                              ALayout,
                                                              BLayout,
-                                                             DELayout>(
+                                                             DLayout,
+                                                             ELayout>(
             do_verification,
             init_method,
             do_log,
@@ -120,19 +123,19 @@ int profile_gemm_bilinear(int argc, char* argv[])
 
     if(data_type == MatrixDataType::F16_F16_F16_F16 && layout == MatrixLayout::MK_KN_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Row{}, Row{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Row{}, Row{}, Row{}, Row{});
     }
     else if(data_type == MatrixDataType::F16_F16_F16_F16 && layout == MatrixLayout::MK_NK_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Row{}, Col{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Row{}, Col{}, Row{}, Row{});
     }
     else if(data_type == MatrixDataType::F16_F16_F16_F16 && layout == MatrixLayout::KM_KN_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Col{}, Row{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Col{}, Row{}, Row{}, Row{});
     }
     else if(data_type == MatrixDataType::F16_F16_F16_F16 && layout == MatrixLayout::KM_NK_MN_MN)
     {
-        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Col{}, Col{}, Row{});
+        return profile(F16{}, F16{}, F32{}, F16{}, F16{}, Col{}, Col{}, Row{}, Row{});
     }
     else
     {
diff --git a/profiler/src/profile_grouped_conv_fwd.cpp b/profiler/src/profile_grouped_conv_fwd.cpp
new file mode 100644
index 0000000000..5873fb676e
--- /dev/null
+++ b/profiler/src/profile_grouped_conv_fwd.cpp
@@ -0,0 +1,258 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+#include <numeric>
+#include <initializer_list>
+#include <cstdlib>
+
+#include "profiler/include/profile_grouped_conv_fwd_impl.hpp"
+
+namespace {
+
+enum struct ConvLayout
+{
+    GNHWC_GKYXC_GNHWK, // 0
+    NHWGC_KYXGC_NHWGK, // 1
+};
+
+enum struct ConvDataType
+{
+    F32_F32_F32,    // 0
+    F16_F16_F16,    // 1
+    BF16_BF16_BF16, // 2
+    INT8_INT8_INT8, // 3
+};
+
+static void print_helper_msg()
+{
+    std::cout
+        // clang-format off
+        << "arg1: tensor operation (grouped_conv_fwd: Grouped Convolution Forward)\n"
+        << "arg2: data type (0: Input fp32, Weight fp32, Output fp32\n"
+        << "                 1: Input fp16, Weight fp16, Output fp16\n"
+        << "                 2: Input bf16, Weight bf16, Output bf16\n"
+        << "                 3: Input int8, Weight int8, Output int8)\n"
+        << "arg3: tensor layout (0: Input[G, N, Hi, Wi, C], Weight[G, K, Y, X, C], Output[G, N, Ho, Wo, K]\n"
+        << "                     1: Input[N, Hi, Wi, G, C], Weight[K, Y, X, G, C], Output[N, Ho, Wo, G, K])\n"
+        << "arg4: verification (0: no, 1: yes)\n"
+        << "arg5: initialization (0: no init, 1: integer value, 2: decimal value)\n"
+        << "arg6: print tensor value (0: no; 1: yes)\n"
+        << "arg7: time kernel (0: no, 1: yes)\n"
+        << ck::utils::conv::get_conv_param_parser_helper_msg() << std::endl;
+    // clang-format on
+}
+
+} // namespace
+
+int profile_grouped_conv_fwd(int argc, char* argv[])
+{
+    // 8 for control, 1 for num_dim_spatial
+    if(argc < 9)
+    {
+        print_helper_msg();
+        return 1;
+    }
+
+    const auto data_type       = static_cast<ConvDataType>(std::stoi(argv[2]));
+    const auto layout          = static_cast<ConvLayout>(std::stoi(argv[3]));
+    const bool do_verification = std::stoi(argv[4]);
+    const int init_method      = std::stoi(argv[5]);
+    const bool do_log          = std::stoi(argv[6]);
+    const bool time_kernel     = std::stoi(argv[7]);
+    const int num_dim_spatial  = std::stoi(argv[8]);
+
+    // 8 for control, 1 for num_dim_spatial, 4 for G/N/K/C, and 6 * num_dim_spatial
+    if(argc != 8 + 1 + 4 + 6 * num_dim_spatial)
+    {
+        print_helper_msg();
+        return 1;
+    }
+
+    const auto params = ck::utils::conv::parse_conv_param(num_dim_spatial, 9, argv);
+
+    using F32  = float;
+    using F16  = ck::half_t;
+    using BF16 = ck::bhalf_t;
+    using INT8 = int8_t;
+
+    //
+    using GNWC   = ck::tensor_layout::convolution::GNWC;
+    using GNHWC  = ck::tensor_layout::convolution::GNHWC;
+    using GNDHWC = ck::tensor_layout::convolution::GNDHWC;
+
+    using GKXC   = ck::tensor_layout::convolution::GKXC;
+    using GKYXC  = ck::tensor_layout::convolution::GKYXC;
+    using GKZYXC = ck::tensor_layout::convolution::GKZYXC;
+
+    using GNWK   = ck::tensor_layout::convolution::GNWK;
+    using GNHWK  = ck::tensor_layout::convolution::GNHWK;
+    using GNDHWK = ck::tensor_layout::convolution::GNDHWK;
+
+    //
+    using NWGC   = ck::tensor_layout::convolution::NWGC;
+    using NHWGC  = ck::tensor_layout::convolution::NHWGC;
+    using NDHWGC = ck::tensor_layout::convolution::NDHWGC;
+
+    using KXGC   = ck::tensor_layout::convolution::KXGC;
+    using KYXGC  = ck::tensor_layout::convolution::KYXGC;
+    using KZYXGC = ck::tensor_layout::convolution::KZYXGC;
+
+    using NWGK   = ck::tensor_layout::convolution::NWGK;
+    using NHWGK  = ck::tensor_layout::convolution::NHWGK;
+    using NDHWGK = ck::tensor_layout::convolution::NDHWGK;
+
+    constexpr auto I1 = ck::Number<1>{};
+    constexpr auto I2 = ck::Number<2>{};
+    constexpr auto I3 = ck::Number<3>{};
+
+    auto profile = [&](auto num_dim_spatial_tmp,
+                       auto in_layout,
+                       auto wei_layout,
+                       auto out_layout,
+                       auto in_type,
+                       auto wei_type,
+                       auto out_type) {
+        constexpr ck::index_t NDimSpatial = num_dim_spatial_tmp.value;
+
+        using InLayout  = decltype(in_layout);
+        using WeiLayout = decltype(wei_layout);
+        using OutLayout = decltype(out_layout);
+
+        using InDataType  = decltype(in_type);
+        using WeiDataType = decltype(wei_type);
+        using OutDataType = decltype(out_type);
+
+        bool pass = ck::profiler::profile_grouped_conv_fwd_impl<NDimSpatial,
+                                                                InLayout,
+                                                                WeiLayout,
+                                                                OutLayout,
+                                                                InDataType,
+                                                                WeiDataType,
+                                                                OutDataType>(
+            do_verification, init_method, do_log, time_kernel, params);
+
+        return pass ? 0 : 1;
+    };
+
+    // GNHWC_GKYXC_GNHWK
+    if(num_dim_spatial == 1 && layout == ConvLayout::GNHWC_GKYXC_GNHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I1, GNWC{}, GKXC{}, GNWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I1, GNWC{}, GKXC{}, GNWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I1, GNWC{}, GKXC{}, GNWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I1, GNWC{}, GKXC{}, GNWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 2 && layout == ConvLayout::GNHWC_GKYXC_GNHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I2, GNHWC{}, GKYXC{}, GNHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I2, GNHWC{}, GKYXC{}, GNHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I2, GNHWC{}, GKYXC{}, GNHWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I2, GNHWC{}, GKYXC{}, GNHWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 3 && layout == ConvLayout::GNHWC_GKYXC_GNHWK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I3, GNDHWC{}, GKZYXC{}, GNDHWK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I3, GNDHWC{}, GKZYXC{}, GNDHWK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I3, GNDHWC{}, GKZYXC{}, GNDHWK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I3, GNDHWC{}, GKZYXC{}, GNDHWK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    // NHWGC_KYXGC_NHWGK
+    else if(num_dim_spatial == 1 && layout == ConvLayout::NHWGC_KYXGC_NHWGK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I1, NWGC{}, KXGC{}, NWGK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I1, NWGC{}, KXGC{}, NWGK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I1, NWGC{}, KXGC{}, NWGK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I1, NWGC{}, KXGC{}, NWGK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 2 && layout == ConvLayout::NHWGC_KYXGC_NHWGK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I2, NHWGC{}, KYXGC{}, NHWGK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I2, NHWGC{}, KYXGC{}, NHWGK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I2, NHWGC{}, KYXGC{}, NHWGK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I2, NHWGC{}, KYXGC{}, NHWGK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+    else if(num_dim_spatial == 3 && layout == ConvLayout::NHWGC_KYXGC_NHWGK)
+    {
+        if(data_type == ConvDataType::F32_F32_F32)
+        {
+            return profile(I3, NDHWGC{}, KZYXGC{}, NDHWGK{}, F32{}, F32{}, F32{});
+        }
+        else if(data_type == ConvDataType::F16_F16_F16)
+        {
+            return profile(I3, NDHWGC{}, KZYXGC{}, NDHWGK{}, F16{}, F16{}, F16{});
+        }
+        else if(data_type == ConvDataType::BF16_BF16_BF16)
+        {
+            return profile(I3, NDHWGC{}, KZYXGC{}, NDHWGK{}, BF16{}, BF16{}, BF16{});
+        }
+        else if(data_type == ConvDataType::INT8_INT8_INT8)
+        {
+            return profile(I3, NDHWGC{}, KZYXGC{}, NDHWGK{}, INT8{}, INT8{}, INT8{});
+        }
+    }
+
+    std::cout << "this data_type & layout is not implemented" << std::endl;
+
+    return 1;
+}
diff --git a/profiler/src/profile_grouped_gemm.cpp b/profiler/src/profile_grouped_gemm.cpp
index a51505ae9c..1e24c6091b 100644
--- a/profiler/src/profile_grouped_gemm.cpp
+++ b/profiler/src/profile_grouped_gemm.cpp
@@ -83,7 +83,7 @@ int profile_grouped_gemm(int argc, char* argv[])
         ck::profiler::profile_grouped_gemm_impl<ck::half_t,
                                                 ck::half_t,
                                                 ck::half_t,
-                                                ck::half_t,
+                                                float,
                                                 ck::tensor_layout::gemm::RowMajor,
                                                 ck::tensor_layout::gemm::RowMajor,
                                                 ck::tensor_layout::gemm::RowMajor>(do_verification,
@@ -102,7 +102,7 @@ int profile_grouped_gemm(int argc, char* argv[])
         ck::profiler::profile_grouped_gemm_impl<ck::half_t,
                                                 ck::half_t,
                                                 ck::half_t,
-                                                ck::half_t,
+                                                float,
                                                 ck::tensor_layout::gemm::RowMajor,
                                                 ck::tensor_layout::gemm::ColumnMajor,
                                                 ck::tensor_layout::gemm::RowMajor>(do_verification,
@@ -121,7 +121,7 @@ int profile_grouped_gemm(int argc, char* argv[])
         ck::profiler::profile_grouped_gemm_impl<ck::half_t,
                                                 ck::half_t,
                                                 ck::half_t,
-                                                ck::half_t,
+                                                float,
                                                 ck::tensor_layout::gemm::ColumnMajor,
                                                 ck::tensor_layout::gemm::RowMajor,
                                                 ck::tensor_layout::gemm::RowMajor>(do_verification,
@@ -140,7 +140,7 @@ int profile_grouped_gemm(int argc, char* argv[])
         ck::profiler::profile_grouped_gemm_impl<ck::half_t,
                                                 ck::half_t,
                                                 ck::half_t,
-                                                ck::half_t,
+                                                float,
                                                 ck::tensor_layout::gemm::ColumnMajor,
                                                 ck::tensor_layout::gemm::ColumnMajor,
                                                 ck::tensor_layout::gemm::RowMajor>(do_verification,
diff --git a/profiler/src/profile_reduce.cpp b/profiler/src/profile_reduce.cpp
index d31cdb74d8..1ec2a6d6e6 100644
--- a/profiler/src/profile_reduce.cpp
+++ b/profiler/src/profile_reduce.cpp
@@ -11,7 +11,7 @@
 
 #include "ck/utility/reduction_enums.hpp"
 
-#include "ck/library/host_tensor/host_common_util.hpp"
+#include "ck/library/utility/host_common_util.hpp"
 
 #include "profiler/include/profile_reduce_impl.hpp"
 #include "profiler/include/data_type_enum.hpp"
diff --git a/profiler/src/profiler.cpp b/profiler/src/profiler.cpp
index 5dbfc547f8..0b1602acc2 100644
--- a/profiler/src/profiler.cpp
+++ b/profiler/src/profiler.cpp
@@ -15,12 +15,11 @@ int profile_grouped_gemm(int, char*[]);
 int profile_conv_fwd(int, char*[]);
 int profile_conv_fwd_bias_relu(int, char*[]);
 int profile_conv_fwd_bias_relu_add(int, char*[]);
-int profile_convnd_fwd(int argc, char* argv[]);
-int profile_convnd_bwd_data(int, char*[], int);
+int profile_conv_bwd_data(int, char*[]);
 int profile_conv_bwd_weight(int, char*[]);
+int profile_grouped_conv_fwd(int, char*[]);
 int profile_normalization(int, char*[]);
 int profile_reduce(int, char*[]);
-int profile_convnd_bwd_weight(int, char*[], int);
 
 static void print_helper_message()
 {
@@ -34,13 +33,12 @@ static void print_helper_message()
            "                        batched_gemm: Batched GEMM\n"
            "                        batched_gemm_reduce: Batched GEMM+Reduce\n"
            "                        grouped_gemm: Grouped GEMM\n"
-           "                        conv_fwd: ForwardConvolution\n"
+           "                        conv_fwd: Convolution Forward\n"
            "                        conv_fwd_bias_relu: ForwardConvolution+Bias+ReLU\n"
            "                        conv_fwd_bias_relu_add: ForwardConvolution+Bias+ReLU+Add\n"
-           "                        conv1d_bwd_data: BackwardConvolution data 1 dim\n"
-           "                        conv2d_bwd_data: BackwardConvolution data 2 dim\n"
-           "                        conv3d_bwd_data: BackwardConvolution data 3 dim\n"
-           "                        conv2d_bwd_weight: Backward Weight Convolution 2d\n"
+           "                        conv_bwd_data: Convolution Backward Data\n"
+           "                        conv_bwd_weight: Convolution Backward Weight\n"
+           "                        grouped_conv_fwd: Grouped Convolution Forward\n"
            "                        reduce: Reduce\n");
     // clang-format on
 }
@@ -53,8 +51,7 @@ int main(int argc, char* argv[])
 
         return 0;
     }
-
-    if(strcmp(argv[1], "gemm") == 0)
+    else if(strcmp(argv[1], "gemm") == 0)
     {
         return profile_gemm(argc, argv);
     }
@@ -92,7 +89,7 @@ int main(int argc, char* argv[])
     }
     else if(strcmp(argv[1], "conv_fwd") == 0)
     {
-        return profile_convnd_fwd(argc, argv);
+        return profile_conv_fwd(argc, argv);
     }
     else if(strcmp(argv[1], "conv_fwd_bias_relu") == 0)
     {
@@ -102,33 +99,17 @@ int main(int argc, char* argv[])
     {
         return profile_conv_fwd_bias_relu_add(argc, argv);
     }
-    else if(strcmp(argv[1], "conv1d_bwd_data") == 0)
+    else if(strcmp(argv[1], "conv_bwd_data") == 0)
     {
-        return profile_convnd_bwd_data(argc, argv, 1);
+        return profile_conv_bwd_data(argc, argv);
     }
-    else if(strcmp(argv[1], "conv2d_bwd_data") == 0)
-    {
-        return profile_convnd_bwd_data(argc, argv, 2);
-    }
-    else if(strcmp(argv[1], "conv3d_bwd_data") == 0)
-    {
-        return profile_convnd_bwd_data(argc, argv, 3);
-    }
-    else if(strcmp(argv[1], "conv2d_bwd_weight") == 0)
+    else if(strcmp(argv[1], "conv_bwd_weight") == 0)
     {
         return profile_conv_bwd_weight(argc, argv);
     }
-    else if(strcmp(argv[1], "convnd1d_bwd_weight") == 0)
+    else if(strcmp(argv[1], "grouped_conv_fwd") == 0)
     {
-        return profile_convnd_bwd_weight(argc, argv, 1);
-    }
-    else if(strcmp(argv[1], "convnd2d_bwd_weight") == 0)
-    {
-        return profile_convnd_bwd_weight(argc, argv, 2);
-    }
-    else if(strcmp(argv[1], "convnd3d_bwd_weight") == 0)
-    {
-        return profile_convnd_bwd_weight(argc, argv, 3);
+        return profile_grouped_conv_fwd(argc, argv);
     }
     else if(strcmp(argv[1], "reduce") == 0)
     {
diff --git a/test/CMakeLists.txt b/test/CMakeLists.txt
index 3df4c9b844..eca4df2c8f 100644
--- a/test/CMakeLists.txt
+++ b/test/CMakeLists.txt
@@ -41,11 +41,11 @@ add_subdirectory(gemm_reduce)
 add_subdirectory(batched_gemm)
 add_subdirectory(batched_gemm_reduce)
 add_subdirectory(grouped_gemm)
-add_subdirectory(convnd_fwd)
 add_subdirectory(reduce)
-add_subdirectory(conv2d_bwd_weight)
+add_subdirectory(convnd_fwd)
 add_subdirectory(convnd_bwd_weight)
 add_subdirectory(convnd_bwd_data)
+add_subdirectory(grouped_convnd_fwd)
 add_subdirectory(block_to_ctile_map)
 add_subdirectory(softmax)
 add_subdirectory(layernorm)
diff --git a/test/batched_gemm/CMakeLists.txt b/test/batched_gemm/CMakeLists.txt
index b70e3aae9b..338c460762 100644
--- a/test/batched_gemm/CMakeLists.txt
+++ b/test/batched_gemm/CMakeLists.txt
@@ -1,4 +1,4 @@
 add_test_executable(test_batched_gemm_fp16 batched_gemm_fp16.cpp)
-target_link_libraries(test_batched_gemm_fp16 PRIVATE host_tensor)
+target_link_libraries(test_batched_gemm_fp16 PRIVATE utility)
 target_link_libraries(test_batched_gemm_fp16 PRIVATE device_batched_gemm_instance)
 
diff --git a/test/batched_gemm_reduce/CMakeLists.txt b/test/batched_gemm_reduce/CMakeLists.txt
index fa1a2bf87f..4dc0b08257 100644
--- a/test/batched_gemm_reduce/CMakeLists.txt
+++ b/test/batched_gemm_reduce/CMakeLists.txt
@@ -1,3 +1,3 @@
 add_test_executable(test_batched_gemm_reduce_fp16 batched_gemm_reduce_fp16.cpp)
-target_link_libraries(test_batched_gemm_reduce_fp16 PRIVATE host_tensor)
+target_link_libraries(test_batched_gemm_reduce_fp16 PRIVATE utility)
 target_link_libraries(test_batched_gemm_reduce_fp16 PRIVATE device_batched_gemm_reduce_instance)
diff --git a/test/conv2d_bwd_data/CMakeLists.txt b/test/conv2d_bwd_data/CMakeLists.txt
deleted file mode 100644
index 1b5c03afa3..0000000000
--- a/test/conv2d_bwd_data/CMakeLists.txt
+++ /dev/null
@@ -1,3 +0,0 @@
-add_test_executable(test_conv2d_bwd_data conv2d_bwd_data.cpp)
-target_link_libraries(test_conv2d_bwd_data PRIVATE host_tensor)
-target_link_libraries(test_conv2d_bwd_data PRIVATE device_conv2d_bwd_data_instance)
diff --git a/test/conv2d_bwd_data/conv2d_bwd_data.cpp b/test/conv2d_bwd_data/conv2d_bwd_data.cpp
deleted file mode 100644
index cb9245387a..0000000000
--- a/test/conv2d_bwd_data/conv2d_bwd_data.cpp
+++ /dev/null
@@ -1,330 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include "config.hpp"
-#include "device.hpp"
-#include "host_tensor.hpp"
-#include "host_tensor_generator.hpp"
-#include "host_conv.hpp"
-#include "tensor_layout.hpp"
-#include "device_tensor.hpp"
-#include "device_conv_bwd_data.hpp"
-#include "element_wise_operation.hpp"
-#include "reference_conv_bwd_data.hpp"
-
-using F16  = ck::half_t;
-using F32  = float;
-using BF16 = ck::bhalf_t;
-using INT8 = int8_t;
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using DeviceConvBwdDataNoOpPtr =
-    DeviceConvBwdDataPtr<ck::tensor_operation::element_wise::PassThrough,
-                         ck::tensor_operation::element_wise::PassThrough,
-                         ck::tensor_operation::element_wise::PassThrough>;
-
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances(
-    std::vector<DeviceConvBwdDataNoOpPtr>&);
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances(
-    std::vector<DeviceConvBwdDataNoOpPtr>&);
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvBwdDataNoOpPtr>&);
-void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(
-    std::vector<DeviceConvBwdDataNoOpPtr>&);
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-
-template <typename T>
-static bool check_out(const Tensor<T>& ref, const Tensor<T>& result)
-{
-    float max_diff = 1e-6;
-
-    for(int i = 0; i < ref.mData.size(); ++i)
-    {
-        float diff = std::abs(double(ref.mData[i]) - double(result.mData[i]));
-        if(max_diff < diff)
-        {
-            return false;
-        }
-    }
-
-    return true;
-}
-
-int main(int argc, char* argv[])
-{
-    int data_type   = 0;
-    int init_method = 0;
-
-    // Conv shape
-    ck::index_t N               = 128;
-    ck::index_t K               = 256;
-    ck::index_t C               = 192;
-    ck::index_t Y               = 3;
-    ck::index_t X               = 3;
-    ck::index_t Hi              = 71;
-    ck::index_t Wi              = 71;
-    ck::index_t conv_stride_h   = 2;
-    ck::index_t conv_stride_w   = 2;
-    ck::index_t conv_dilation_h = 1;
-    ck::index_t conv_dilation_w = 1;
-    ck::index_t in_left_pad_h   = 1;
-    ck::index_t in_left_pad_w   = 1;
-    ck::index_t in_right_pad_h  = 1;
-    ck::index_t in_right_pad_w  = 1;
-
-    if(argc == 1)
-    {
-        data_type   = 1;
-        init_method = 1;
-    }
-    else if(argc == 3)
-    {
-        data_type   = std::stoi(argv[1]);
-        init_method = std::stoi(argv[2]);
-    }
-    else if(argc == 18)
-    {
-        data_type   = std::stoi(argv[1]);
-        init_method = std::stoi(argv[2]);
-
-        N               = std::stoi(argv[3]);
-        K               = std::stoi(argv[4]);
-        C               = std::stoi(argv[5]);
-        Y               = std::stoi(argv[6]);
-        X               = std::stoi(argv[7]);
-        Hi              = std::stoi(argv[8]);
-        Wi              = std::stoi(argv[9]);
-        conv_stride_h   = std::stoi(argv[10]);
-        conv_stride_w   = std::stoi(argv[11]);
-        conv_dilation_h = std::stoi(argv[12]);
-        conv_dilation_w = std::stoi(argv[13]);
-        in_left_pad_h   = std::stoi(argv[14]);
-        in_left_pad_w   = std::stoi(argv[15]);
-        in_right_pad_h  = std::stoi(argv[16]);
-        in_right_pad_w  = std::stoi(argv[17]);
-    }
-    else
-    {
-        printf("arg1: data type (0=fp32, 1=fp16, 2= bfp16, 3= int8_t )\n");
-        printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
-        printf("arg3 to 17: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-               "RightPx\n");
-        exit(1);
-    }
-
-    auto Run = [&](auto input_type, auto wei_type, auto out_type, auto acc_type) {
-        using InDataType  = decltype(input_type);
-        using WeiDataType = decltype(wei_type);
-        using OutDataType = decltype(out_type);
-        using AccDataType = decltype(acc_type);
-
-        using ReferenceConvBwdInstance =
-            ck::tensor_operation::host::ReferenceConvBwdData<InDataType,
-                                                             WeiDataType,
-                                                             OutDataType,
-                                                             AccDataType,
-                                                             InElementOp,
-                                                             WeiElementOp,
-                                                             OutElementOp>;
-
-        const ck::index_t YEff = (Y - 1) * conv_dilation_h + 1;
-        const ck::index_t XEff = (X - 1) * conv_dilation_w + 1;
-
-        const ck::index_t Ho = (Hi + in_left_pad_h + in_right_pad_h - YEff) / conv_stride_h + 1;
-        const ck::index_t Wo = (Wi + in_left_pad_w + in_right_pad_w - XEff) / conv_stride_w + 1;
-
-        const std::vector<ck::index_t> input_spatial_lengths{{Hi, Wi}};
-        const std::vector<ck::index_t> filter_spatial_lengths{{Y, X}};
-        const std::vector<ck::index_t> output_spatial_lengths{{Ho, Wo}};
-        const std::vector<ck::index_t> conv_filter_strides{{conv_stride_h, conv_stride_w}};
-        const std::vector<ck::index_t> conv_filter_dilations{{conv_dilation_h, conv_dilation_w}};
-        const std::vector<ck::index_t> input_left_pads{{in_left_pad_h, in_left_pad_w}};
-        const std::vector<ck::index_t> input_right_pads{{in_right_pad_h, in_right_pad_w}};
-
-        auto f_host_tensor_descriptor =
-            [](std::size_t N_, std::size_t C_, std::size_t H, std::size_t W) {
-                return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
-                                            std::vector<std::size_t>({C_ * H * W, 1, W * C_, C_}));
-            };
-
-        Tensor<OutDataType> out_n_k_ho_wo(f_host_tensor_descriptor(N, K, Ho, Wo));
-        Tensor<WeiDataType> wei_k_c_y_x(f_host_tensor_descriptor(K, C, Y, X));
-        Tensor<InDataType> in_n_c_hi_wi_host_result(f_host_tensor_descriptor(N, C, Hi, Wi));
-        Tensor<InDataType> in_n_c_hi_wi_device_result(f_host_tensor_descriptor(N, C, Hi, Wi));
-
-        std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi_host_result.mDesc << std::endl;
-        std::cout << "wei_k_c_y_x: " << wei_k_c_y_x.mDesc << std::endl;
-        std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo.mDesc << std::endl;
-
-        switch(init_method)
-        {
-        case 0: break;
-        case 1:
-            out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_2<OutDataType>{-5, 5});
-            wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
-            break;
-        default:
-            out_n_k_ho_wo.GenerateTensorValue(GeneratorTensor_1<OutDataType>{1});
-            wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_1<WeiDataType>{1});
-        }
-
-        DeviceMem in_device_buf(sizeof(InDataType) *
-                                in_n_c_hi_wi_device_result.mDesc.GetElementSpace());
-        DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpace());
-        DeviceMem out_device_buf(sizeof(OutDataType) * out_n_k_ho_wo.mDesc.GetElementSpace());
-
-        out_device_buf.ToDevice(out_n_k_ho_wo.mData.data());
-        wei_device_buf.ToDevice(wei_k_c_y_x.mData.data());
-        // reset input to zero
-        in_n_c_hi_wi_device_result.GenerateTensorValue(GeneratorTensor_1<InDataType>{0});
-        in_device_buf.ToDevice(in_n_c_hi_wi_device_result.mData.data());
-
-        // get host result
-        {
-            auto ref_conv    = ReferenceConvBwdInstance{};
-            auto ref_invoker = ref_conv.MakeInvoker();
-
-            auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi_host_result,
-                                                      wei_k_c_y_x,
-                                                      out_n_k_ho_wo,
-                                                      conv_filter_strides,
-                                                      conv_filter_dilations,
-                                                      input_left_pads,
-                                                      input_right_pads,
-                                                      InElementOp{},
-                                                      WeiElementOp{},
-                                                      OutElementOp{});
-            ref_invoker.Run(ref_argument);
-        }
-
-        using PassThrough              = ck::tensor_operation::element_wise::PassThrough;
-        using DeviceConvBwdDataNoOpPtr = ck::tensor_operation::device::
-            DeviceConvBwdDataPtr<PassThrough, PassThrough, PassThrough>;
-
-        // add device Conv instances
-        std::vector<DeviceConvBwdDataNoOpPtr> conv_ptrs;
-
-        if constexpr(ck::is_same_v<ck::remove_cv_t<InDataType>, float> &&
-                     ck::is_same_v<ck::remove_cv_t<WeiDataType>, float> &&
-                     ck::is_same_v<ck::remove_cv_t<OutDataType>, float>)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instances(conv_ptrs);
-        }
-        else if constexpr(ck::is_same_v<ck::remove_cv_t<InDataType>, ck::half_t> &&
-                          ck::is_same_v<ck::remove_cv_t<WeiDataType>, ck::half_t> &&
-                          ck::is_same_v<ck::remove_cv_t<OutDataType>, ck::half_t>)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instances(conv_ptrs);
-        }
-        else if constexpr(ck::is_same_v<ck::remove_cv_t<InDataType>, ck::bhalf_t> &&
-                          ck::is_same_v<ck::remove_cv_t<WeiDataType>, ck::bhalf_t> &&
-                          ck::is_same_v<ck::remove_cv_t<OutDataType>, ck::bhalf_t>)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instances(conv_ptrs);
-        }
-        else if constexpr(ck::is_same_v<ck::remove_cv_t<InDataType>, int8_t> &&
-                          ck::is_same_v<ck::remove_cv_t<WeiDataType>, int8_t> &&
-                          ck::is_same_v<ck::remove_cv_t<OutDataType>, int8_t>)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(conv_ptrs);
-        }
-
-        if(conv_ptrs.size() <= 0)
-        {
-            throw std::runtime_error("wrong! no device Conv instance found");
-        }
-
-        // profile device Conv instances
-        bool success = true;
-        for(auto& conv_ptr : conv_ptrs)
-        {
-            auto argument_ptr = conv_ptr->MakeArgumentPointer(
-                static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
-                static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
-                static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
-                N,
-                K,
-                C,
-                input_spatial_lengths,
-                filter_spatial_lengths,
-                output_spatial_lengths,
-                conv_filter_strides,
-                conv_filter_dilations,
-                input_left_pads,
-                input_right_pads,
-                InElementOp{},
-                WeiElementOp{},
-                OutElementOp{});
-
-            if(conv_ptr->IsSupportedArgument(argument_ptr.get()))
-            {
-                auto invoker_ptr = conv_ptr->MakeInvokerPointer();
-                invoker_ptr->Run(argument_ptr.get(), 1);
-
-                in_device_buf.FromDevice(in_n_c_hi_wi_device_result.mData.data());
-
-                if(!check_out(in_n_c_hi_wi_host_result, in_n_c_hi_wi_device_result))
-                {
-                    std::cout << "Fail Info: " << conv_ptr->GetTypeString() << std::endl;
-                    success = false;
-                }
-                else
-                {
-                    std::cout << "Pass Info: " << conv_ptr->GetTypeString() << std::endl;
-                }
-            }
-            else
-            {
-                std::cout << "Not support Info: " << conv_ptr->GetTypeString() << std::endl;
-            }
-        }
-
-        if(success)
-        {
-            std::cout << "test conv2d bwd : Pass" << std::endl;
-            return 0;
-        }
-        else
-        {
-            std::cout << "test conv2d bwd: Fail " << std::endl;
-            return -1;
-        }
-    };
-
-    if(data_type == 0)
-    {
-        return Run(F32(), F32(), F32(), F32());
-    }
-    else if(data_type == 1)
-    {
-        return Run(F16(), F16(), F16(), F32());
-    }
-    else if(data_type == 2)
-    {
-        return Run(BF16(), BF16(), BF16(), F32());
-    }
-    else if(data_type == 3)
-    {
-        return Run(INT8(), INT8(), INT8(), int());
-    }
-    else
-    {
-        return 1;
-    }
-}
diff --git a/test/conv2d_bwd_weight/CMakeLists.txt b/test/conv2d_bwd_weight/CMakeLists.txt
deleted file mode 100644
index 0acd546830..0000000000
--- a/test/conv2d_bwd_weight/CMakeLists.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-#add_test_executable(test_conv2d_bwd_weight conv2d_bwd_weight.cpp)
-#target_link_libraries(test_conv2d_bwd_weight PRIVATE host_tensor device_conv2d_bwd_weight_instance conv_util)
diff --git a/test/conv2d_bwd_weight/conv2d_bwd_weight.cpp b/test/conv2d_bwd_weight/conv2d_bwd_weight.cpp
deleted file mode 100644
index 7af0fa3d82..0000000000
--- a/test/conv2d_bwd_weight/conv2d_bwd_weight.cpp
+++ /dev/null
@@ -1,217 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
-#include <cstdlib>
-#include <vector>
-
-#include "test/convnd_fwd/conv_util.hpp"
-#include "profiler/include/profile_conv_bwd_weight_impl.hpp"
-
-int test_self()
-{
-    bool pass = true;
-    std::vector<ck::utils::conv::ConvParams> params;
-
-    params.push_back({2, 128, 256, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
-    params.push_back({2, 128, 256, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
-    params.push_back({2, 128, 256, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
-
-    for(auto& param : params)
-    {
-        // f32
-        pass &= ck::profiler::profile_conv_bwd_weight_impl<2,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NHWC,
-                                                           ck::tensor_layout::convolution::KYXC,
-                                                           ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // fp16
-        pass &= ck::profiler::profile_conv_bwd_weight_impl<2,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           ck::tensor_layout::convolution::NHWC,
-                                                           ck::tensor_layout::convolution::KYXC,
-                                                           ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-    }
-    return pass;
-}
-int main(int argc, char* argv[])
-{
-    int data_type   = 1;
-    int init_method = 1;
-
-    // Conv shape
-    ck::index_t N               = 128;
-    ck::index_t K               = 256;
-    ck::index_t C               = 192;
-    ck::index_t Y               = 3;
-    ck::index_t X               = 3;
-    ck::index_t Hi              = 71;
-    ck::index_t Wi              = 71;
-    ck::index_t conv_stride_h   = 2;
-    ck::index_t conv_stride_w   = 2;
-    ck::index_t conv_dilation_h = 1;
-    ck::index_t conv_dilation_w = 1;
-    ck::index_t in_left_pad_h   = 1;
-    ck::index_t in_left_pad_w   = 1;
-    ck::index_t in_right_pad_h  = 1;
-    ck::index_t in_right_pad_w  = 1;
-    ck::index_t split_k         = 1;
-
-    bool pass = true;
-    if(argc == 1)
-    {
-        pass = test_self();
-    }
-    else
-    {
-        if(argc == 3)
-        {
-            data_type   = std::stoi(argv[1]);
-            init_method = std::stoi(argv[2]);
-        }
-        else if(argc == 19)
-        {
-            data_type   = std::stoi(argv[1]);
-            init_method = std::stoi(argv[2]);
-
-            N               = std::stoi(argv[3]);
-            K               = std::stoi(argv[4]);
-            C               = std::stoi(argv[5]);
-            Y               = std::stoi(argv[6]);
-            X               = std::stoi(argv[7]);
-            Hi              = std::stoi(argv[8]);
-            Wi              = std::stoi(argv[9]);
-            conv_stride_h   = std::stoi(argv[10]);
-            conv_stride_w   = std::stoi(argv[11]);
-            conv_dilation_h = std::stoi(argv[12]);
-            conv_dilation_w = std::stoi(argv[13]);
-            in_left_pad_h   = std::stoi(argv[14]);
-            in_left_pad_w   = std::stoi(argv[15]);
-            in_right_pad_h  = std::stoi(argv[16]);
-            in_right_pad_w  = std::stoi(argv[17]);
-            split_k         = std::stoi(argv[18]);
-        }
-        else
-        {
-            printf("arg1: data type (0=fp32, 1=fp16, 2= bfp16, 3= int8_t )\n");
-            printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
-            printf("arg3 to 17: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
-                   "RightPx\n");
-            exit(1);
-        }
-
-        ck::utils::conv::ConvParams param{2,
-                                          N,
-                                          K,
-                                          C,
-                                          {Y, X},
-                                          {Hi, Wi},
-                                          {conv_stride_h, conv_stride_w},
-                                          {conv_dilation_h, conv_dilation_w},
-                                          {in_left_pad_h, in_left_pad_w},
-                                          {in_right_pad_h, in_right_pad_w}};
-        if(data_type == 0)
-        {
-            pass = ck::profiler::profile_conv_bwd_weight_impl<2,
-                                                              float,
-                                                              float,
-                                                              float,
-                                                              ck::tensor_layout::convolution::NHWC,
-                                                              ck::tensor_layout::convolution::KYXC,
-                                                              ck::tensor_layout::convolution::NHWK>(
-                true, // do_verification
-                init_method,
-                false, // do_log
-                false, // time_kernel
-                param.N_,
-                param.K_,
-                param.C_,
-                param.input_spatial_lengths_,
-                param.filter_spatial_lengths_,
-                param.GetOutputSpatialLengths(),
-                param.conv_filter_strides_,
-                param.conv_filter_dilations_,
-                param.input_left_pads_,
-                param.input_right_pads_,
-                split_k);
-        }
-        else if(data_type == 1)
-        {
-            pass = ck::profiler::profile_conv_bwd_weight_impl<2,
-                                                              ck::half_t,
-                                                              ck::half_t,
-                                                              ck::half_t,
-                                                              ck::tensor_layout::convolution::NHWC,
-                                                              ck::tensor_layout::convolution::KYXC,
-                                                              ck::tensor_layout::convolution::NHWK>(
-                true, // do_verification
-                init_method,
-                false, // do_log
-                false, // time_kernel
-                param.N_,
-                param.K_,
-                param.C_,
-                param.input_spatial_lengths_,
-                param.filter_spatial_lengths_,
-                param.GetOutputSpatialLengths(),
-                param.conv_filter_strides_,
-                param.conv_filter_dilations_,
-                param.input_left_pads_,
-                param.input_right_pads_,
-                split_k);
-        }
-        else
-        {
-            std::cout << "Not support data type" << std::endl;
-            return 1;
-        }
-    }
-
-    if(pass)
-    {
-        std::cout << "test conv2d bwd weight : Pass" << std::endl;
-        return 0;
-    }
-    else
-    {
-        std::cout << "test conv2d bwd weight: Fail " << std::endl;
-        return -1;
-    }
-}
diff --git a/test/conv_util/CMakeLists.txt b/test/conv_util/CMakeLists.txt
index 795c9ec0ac..7a46039f15 100644
--- a/test/conv_util/CMakeLists.txt
+++ b/test/conv_util/CMakeLists.txt
@@ -1,2 +1,2 @@
 add_gtest_executable(test_conv_util conv_util.cpp)
-target_link_libraries(test_conv_util PRIVATE host_tensor conv_util)
+target_link_libraries(test_conv_util PRIVATE utility)
diff --git a/test/conv_util/conv_util.cpp b/test/conv_util/conv_util.cpp
index 293d94542c..73797a7169 100644
--- a/test/conv_util/conv_util.cpp
+++ b/test/conv_util/conv_util.cpp
@@ -10,198 +10,147 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
 
 namespace {
 
 class TestConvUtil : public ::testing::Test
 {
     public:
-    void SetNDParams(std::size_t ndims)
+    void SetNDParams(std::size_t ndims, std::size_t s, std::size_t d, std::size_t p)
     {
-        conv_params.num_dim_spatial_        = ndims;
-        conv_params.filter_spatial_lengths_ = std::vector<ck::index_t>(ndims, 3);
-        conv_params.input_spatial_lengths_  = std::vector<ck::index_t>(ndims, 71);
-        conv_params.conv_filter_strides_    = std::vector<ck::index_t>(ndims, 2);
-        conv_params.conv_filter_dilations_  = std::vector<ck::index_t>(ndims, 1);
-        conv_params.input_left_pads_        = std::vector<ck::index_t>(ndims, 1);
-        conv_params.input_right_pads_       = std::vector<ck::index_t>(ndims, 1);
+        conv_params = ck::utils::conv::ConvParam(ndims,
+                                                 2,
+                                                 128,
+                                                 192,
+                                                 256,
+                                                 std::vector<ck::index_t>(ndims, 3),
+                                                 std::vector<ck::index_t>(ndims, 71),
+                                                 std::vector<ck::index_t>(ndims, s),
+                                                 std::vector<ck::index_t>(ndims, d),
+                                                 std::vector<ck::index_t>(ndims, p),
+                                                 std::vector<ck::index_t>(ndims, p));
     }
 
     protected:
     // -------  default 2D -------
-    // input NCHW {128,192,71,71},
-    // weights KCYX {256,192,3,3},
-    // stride {2,2},
-    // dilations {1,1},
-    // padding {{1,1}, {1,1}}
-    ck::utils::conv::ConvParams conv_params;
+    // input GNCHW {2, 128, 192, 71, 71},
+    // weights GKCYX {2, 256, 192, 3, 3},
+    // stride {s, s},
+    // dilations {d, d},
+    // padding {{p, p}, {p, p}
+    ck::utils::conv::ConvParam conv_params;
 };
 
 } // namespace
 
-TEST_F(TestConvUtil, ConvParamsGetOutputSpatialLengths2D)
-{
-    ck::utils::conv::ConvParams conv_params;
-    std::vector<ck::index_t> out_spatial_len = conv_params.GetOutputSpatialLengths();
-    EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
-                                     std::vector<ck::index_t>{36, 36},
-                                     "Error: ConvParams 2D default constructor."));
-
-    conv_params.conv_filter_strides_ = std::vector<ck::index_t>{1, 1};
-    out_spatial_len                  = conv_params.GetOutputSpatialLengths();
-    EXPECT_TRUE(ck::utils::check_err(
-        out_spatial_len, std::vector<ck::index_t>{71, 71}, "Error: ConvParams 2D stride {1,1}."));
-
-    conv_params.conv_filter_strides_ = std::vector<ck::index_t>{2, 2};
-    conv_params.input_left_pads_     = std::vector<ck::index_t>{2, 2};
-    conv_params.input_right_pads_    = std::vector<ck::index_t>{2, 2};
-    out_spatial_len                  = conv_params.GetOutputSpatialLengths();
-    EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
-                                     std::vector<ck::index_t>{37, 37},
-                                     "Error: ConvParams 2D padding left/right {2,2}."));
-
-    conv_params.conv_filter_dilations_ = std::vector<ck::index_t>{2, 2};
-    out_spatial_len                    = conv_params.GetOutputSpatialLengths();
-    EXPECT_TRUE(ck::utils::check_err(
-        out_spatial_len, std::vector<ck::index_t>{36, 36}, "Error: ConvParams 2D dilation {2,2}."));
-
-    conv_params.conv_filter_strides_   = std::vector<ck::index_t>{3, 3};
-    conv_params.input_left_pads_       = std::vector<ck::index_t>{1, 1};
-    conv_params.input_right_pads_      = std::vector<ck::index_t>{1, 1};
-    conv_params.conv_filter_dilations_ = std::vector<ck::index_t>{2, 2};
-    out_spatial_len                    = conv_params.GetOutputSpatialLengths();
-    EXPECT_TRUE(
-        ck::utils::check_err(out_spatial_len,
-                             std::vector<ck::index_t>{23, 23},
-                             "Error: ConvParams 2D strides{3,3}, padding {1,1}, dilations {2,2}."));
-}
-
 TEST_F(TestConvUtil, ConvParamsGetOutputSpatialLengths1D)
 {
-    SetNDParams(1);
-
+    // stride 2, dilation 1, pad 1
+    SetNDParams(1, 2, 1, 1);
     std::vector<ck::index_t> out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(
         out_spatial_len, std::vector<ck::index_t>{36}, "Error: ConvParams 1D."));
 
-    conv_params.conv_filter_strides_ = std::vector<ck::index_t>{1};
-    out_spatial_len                  = conv_params.GetOutputSpatialLengths();
+    // stride 1, dilation 1, pad 1
+    SetNDParams(1, 1, 1, 1);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(
         out_spatial_len, std::vector<ck::index_t>{71}, "Error: ConvParams 1D stride {1}."));
 
-    conv_params.conv_filter_strides_ = std::vector<ck::index_t>{2};
-    conv_params.input_left_pads_     = std::vector<ck::index_t>{2};
-    conv_params.input_right_pads_    = std::vector<ck::index_t>{2};
-    out_spatial_len                  = conv_params.GetOutputSpatialLengths();
+    // stride 2, dilation 1, pad 2
+    SetNDParams(1, 2, 1, 2);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
                                      std::vector<ck::index_t>{37},
                                      "Error: ConvParams 1D padding left/right {2}."));
 
-    conv_params.conv_filter_dilations_ = std::vector<ck::index_t>{2};
-    out_spatial_len                    = conv_params.GetOutputSpatialLengths();
+    // stride 2, dilation 2, pad 2
+    SetNDParams(1, 2, 2, 2);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(
         out_spatial_len, std::vector<ck::index_t>{36}, "Error: ConvParams 1D dilation {2}."));
 
-    conv_params.conv_filter_strides_   = std::vector<ck::index_t>{3};
-    conv_params.input_left_pads_       = std::vector<ck::index_t>{1};
-    conv_params.input_right_pads_      = std::vector<ck::index_t>{1};
-    conv_params.conv_filter_dilations_ = std::vector<ck::index_t>{2};
-    out_spatial_len                    = conv_params.GetOutputSpatialLengths();
+    // stride 3, dilation 2, pad 1
+    SetNDParams(1, 3, 2, 1);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(
         ck::utils::check_err(out_spatial_len,
                              std::vector<ck::index_t>{23},
                              "Error: ConvParams 1D strides{3}, padding {1}, dilations {2}."));
 }
 
+TEST_F(TestConvUtil, ConvParamsGetOutputSpatialLengths2D)
+{
+    // stride 2, dilation 1, pad 1
+    SetNDParams(2, 2, 1, 1);
+    std::vector<ck::index_t> out_spatial_len = conv_params.GetOutputSpatialLengths();
+    EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
+                                     std::vector<ck::index_t>{36, 36},
+                                     "Error: ConvParams 2D default constructor."));
+
+    // stride 1, dilation 1, pad 1
+    SetNDParams(2, 1, 1, 1);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
+    EXPECT_TRUE(ck::utils::check_err(
+        out_spatial_len, std::vector<ck::index_t>{71, 71}, "Error: ConvParams 2D stride {1,1}."));
+
+    // stride 2, dilation 1, pad 2
+    SetNDParams(2, 2, 1, 2);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
+    EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
+                                     std::vector<ck::index_t>{37, 37},
+                                     "Error: ConvParams 2D padding left/right {2,2}."));
+
+    // stride 2, dilation 2, pad 2
+    SetNDParams(2, 2, 2, 2);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
+    EXPECT_TRUE(ck::utils::check_err(
+        out_spatial_len, std::vector<ck::index_t>{36, 36}, "Error: ConvParams 2D dilation {2,2}."));
+
+    // stride 3, dilation 2, pad 1
+    SetNDParams(2, 3, 2, 1);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
+    EXPECT_TRUE(
+        ck::utils::check_err(out_spatial_len,
+                             std::vector<ck::index_t>{23, 23},
+                             "Error: ConvParams 2D strides{3,3}, padding {1,1}, dilations {2,2}."));
+}
+
 TEST_F(TestConvUtil, ConvParamsGetOutputSpatialLengths3D)
 {
-    SetNDParams(3);
-
+    // stride 2, dilation 1, pad 1
+    SetNDParams(3, 2, 1, 1);
     std::vector<ck::index_t> out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(
         out_spatial_len, std::vector<ck::index_t>{36, 36, 36}, "Error: ConvParams 3D."));
 
-    conv_params.conv_filter_strides_ = std::vector<ck::index_t>{1, 1, 1};
-    out_spatial_len                  = conv_params.GetOutputSpatialLengths();
+    // stride 1, dilation 1, pad 1
+    SetNDParams(3, 1, 1, 1);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
                                      std::vector<ck::index_t>{71, 71, 71},
                                      "Error: ConvParams 3D stride {1, 1, 1}."));
 
-    conv_params.conv_filter_strides_ = std::vector<ck::index_t>{2, 2, 2};
-    conv_params.input_left_pads_     = std::vector<ck::index_t>{2, 2, 2};
-    conv_params.input_right_pads_    = std::vector<ck::index_t>{2, 2, 2};
-    out_spatial_len                  = conv_params.GetOutputSpatialLengths();
+    // stride 2, dilation 1, pad 2
+    SetNDParams(3, 2, 1, 2);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
                                      std::vector<ck::index_t>{37, 37, 37},
                                      "Error: ConvParams 3D padding left/right {2, 2, 2}."));
 
-    conv_params.conv_filter_dilations_ = std::vector<ck::index_t>{2, 2, 2};
-    out_spatial_len                    = conv_params.GetOutputSpatialLengths();
+    // stride 2, dilation 2, pad 2
+    SetNDParams(3, 2, 2, 2);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(out_spatial_len,
                                      std::vector<ck::index_t>{36, 36, 36},
                                      "Error: ConvParams 3D dilation {2, 2, 2}."));
 
-    conv_params.conv_filter_strides_   = std::vector<ck::index_t>{3, 3, 3};
-    conv_params.input_left_pads_       = std::vector<ck::index_t>{1, 1, 1};
-    conv_params.input_right_pads_      = std::vector<ck::index_t>{1, 1, 1};
-    conv_params.conv_filter_dilations_ = std::vector<ck::index_t>{2, 2, 2};
-    out_spatial_len                    = conv_params.GetOutputSpatialLengths();
+    // stride 3, dilation 2, pad 1
+    SetNDParams(3, 3, 2, 1);
+    out_spatial_len = conv_params.GetOutputSpatialLengths();
     EXPECT_TRUE(ck::utils::check_err(
         out_spatial_len,
         std::vector<ck::index_t>{23, 23, 23},
         "Error: ConvParams 3D strides{3, 3, 3}, padding {1, 1, 1}, dilations {2, 2, 2}."));
 }
-
-TEST(ConvUtil, GetHostTensorDescriptor)
-{
-    namespace tl = ck::tensor_layout::convolution;
-    std::vector<std::size_t> dims{2, 3, 4, 5};
-    HostTensorDescriptor h = ck::utils::conv::get_host_tensor_descriptor(dims, tl::NHWC{});
-    EXPECT_TRUE(ck::utils::check_err(
-        h.GetLengths(), {2, 3, 4, 5}, "Error: wrong NHWC dimensions lengths!"));
-    EXPECT_TRUE(ck::utils::check_err(
-        h.GetStrides(), {3 * 4 * 5, 1, 3 * 5, 3}, "Error: wrong NHWC dimensions strides!"));
-
-    h = ck::utils::conv::get_host_tensor_descriptor(dims, tl::NCHW{});
-    EXPECT_TRUE(ck::utils::check_err(
-        h.GetLengths(), {2, 3, 4, 5}, "Error: wrong NCHW dimensions lengths!"));
-    EXPECT_TRUE(ck::utils::check_err(
-        h.GetStrides(), {3 * 4 * 5, 4 * 5, 5, 1}, "Error: wrong NCHW dimensions strides!"));
-
-    dims = std::vector<std::size_t>{2, 3, 4};
-    h    = ck::utils::conv::get_host_tensor_descriptor(dims, tl::NWC{});
-    EXPECT_TRUE(
-        ck::utils::check_err(h.GetLengths(), {2, 3, 4}, "Error: wrong NWC dimensions lengths!"));
-    EXPECT_TRUE(ck::utils::check_err(
-        h.GetStrides(), {3 * 4, 1, 3}, "Error: wrong NWC dimensions strides!"));
-
-    h = ck::utils::conv::get_host_tensor_descriptor(dims, tl::NCW{});
-    EXPECT_TRUE(
-        ck::utils::check_err(h.GetLengths(), {2, 3, 4}, "Error: wrong NCW dimensions lengths!"));
-    EXPECT_TRUE(ck::utils::check_err(
-        h.GetStrides(), {3 * 4, 4, 1}, "Error: wrong NCW dimensions strides!"));
-
-    dims = std::vector<std::size_t>{2, 3, 4, 5, 6};
-    h    = ck::utils::conv::get_host_tensor_descriptor(dims, tl::NDHWC{});
-    EXPECT_TRUE(
-        ck::utils::check_err(h.GetLengths(), dims, "Error: wrong NDHWC dimensions lengths!"));
-    EXPECT_TRUE(ck::utils::check_err(h.GetStrides(),
-                                     {3 * 4 * 5 * 6, // N
-                                      1,             // C
-                                      3 * 5 * 6,     // D
-                                      3 * 6,         // H
-                                      3},            // W
-                                     "Error: wrong NDHWC dimensions strides!"));
-
-    h = ck::utils::conv::get_host_tensor_descriptor(dims, tl::NCDHW{});
-    EXPECT_TRUE(
-        ck::utils::check_err(h.GetLengths(), dims, "Error: wrong NCDHW dimensions lengths!"));
-    EXPECT_TRUE(ck::utils::check_err(h.GetStrides(),
-                                     {3 * 4 * 5 * 6, // N
-                                      4 * 5 * 6,     // C
-                                      5 * 6,         // D
-                                      6,             // H
-                                      1},            // W
-                                     "Error: wrong NCDHW dimensions strides!"));
-}
diff --git a/test/convnd_bwd_data/CMakeLists.txt b/test/convnd_bwd_data/CMakeLists.txt
index 554bcd18fb..16ca4de872 100644
--- a/test/convnd_bwd_data/CMakeLists.txt
+++ b/test/convnd_bwd_data/CMakeLists.txt
@@ -1,2 +1,2 @@
-add_test_executable(test_convnd_bwd_data convnd_bwd_data.cpp)
-target_link_libraries(test_convnd_bwd_data PRIVATE host_tensor device_convnd_bwd_data_instance conv_util)
+add_gtest_executable(test_convnd_bwd_data convnd_bwd_data.cpp)
+target_link_libraries(test_convnd_bwd_data PRIVATE utility device_conv1d_bwd_data_instance device_conv2d_bwd_data_instance device_conv3d_bwd_data_instance)
diff --git a/test/convnd_bwd_data/convnd_bwd_data.cpp b/test/convnd_bwd_data/convnd_bwd_data.cpp
index a5b83b9eed..cc555faf68 100644
--- a/test/convnd_bwd_data/convnd_bwd_data.cpp
+++ b/test/convnd_bwd_data/convnd_bwd_data.cpp
@@ -1,331 +1,241 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
 #include <cstdlib>
+#include <iostream>
+#include <initializer_list>
 #include <vector>
+#include <gtest/gtest.h>
 
-#include "profiler/include/profile_convnd_bwd_data_impl.hpp"
+#include "profiler/include/profile_conv_bwd_data_impl.hpp"
 
-int main()
+class TestConvndBwdData : public ::testing::Test
 {
-    bool pass = true;
-    // check 1d
-    std::vector<ck::utils::conv::ConvParams> params;
-    params.push_back({1, 128, 128, 256, {1}, {14}, {2}, {1}, {0}, {0}});
-    params.push_back({1, 128, 128, 256, {3}, {28}, {1}, {1}, {1}, {1}});
-    params.push_back({1, 128, 128, 256, {1}, {3}, {1}, {1}, {0}, {0}});
+    protected:
+    std::vector<ck::utils::conv::ConvParam> conv_params;
+};
 
-    for(auto& param : params)
+// 1d
+TEST_F(TestConvndBwdData, Conv1dBwdData)
+{
+    conv_params.clear();
+    conv_params.push_back({1, 1, 128, 128, 256, {1}, {14}, {2}, {1}, {0}, {0}});
+    conv_params.push_back({1, 1, 128, 128, 256, {3}, {28}, {1}, {1}, {1}, {1}});
+    conv_params.push_back({1, 1, 128, 128, 256, {1}, {3}, {1}, {1}, {0}, {0}});
+
+    for(auto& param : conv_params)
     {
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<1,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NWC,
-                                                           ck::tensor_layout::convolution::KXC,
-                                                           ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
+        bool pass;
 
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<1,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NWC,
-                                                           ck::tensor_layout::convolution::KXC,
-                                                           ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
+        // fp32
+        pass = ck::profiler::profile_conv_bwd_data_impl<1,
+                                                        ck::tensor_layout::convolution::NWC,
+                                                        ck::tensor_layout::convolution::KXC,
+                                                        ck::tensor_layout::convolution::NWK,
+                                                        float,
+                                                        float,
+                                                        float>(true,  // do_verification
+                                                               1,     // init_method
+                                                               false, // do_log
+                                                               false, // time_kernel
+                                                               param);
 
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<1,
-                                                           ck::bhalf_t,
-                                                           ck::bhalf_t,
-                                                           ck::bhalf_t,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NWC,
-                                                           ck::tensor_layout::convolution::KXC,
-                                                           ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
+        EXPECT_TRUE(pass);
 
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<1,
-                                                           int8_t,
-                                                           int8_t,
-                                                           int8_t,
-                                                           int,
-                                                           ck::tensor_layout::convolution::NWC,
-                                                           ck::tensor_layout::convolution::KXC,
-                                                           ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
-    }
+        // fp16
+        pass = ck::profiler::profile_conv_bwd_data_impl<1,
+                                                        ck::tensor_layout::convolution::NWC,
+                                                        ck::tensor_layout::convolution::KXC,
+                                                        ck::tensor_layout::convolution::NWK,
+                                                        ck::half_t,
+                                                        ck::half_t,
+                                                        ck::half_t>(true,  // do_verification
+                                                                    1,     // init_method
+                                                                    false, // do_log
+                                                                    false, // time_kernel
+                                                                    param);
 
-    // check 2d
-    params.clear();
-    params.push_back({2, 128, 128, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
-    params.push_back({2, 128, 128, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
-    params.push_back({2, 128, 128, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
+        EXPECT_TRUE(pass);
 
-    for(auto& param : params)
-    {
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<2,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NHWC,
-                                                           ck::tensor_layout::convolution::KYXC,
-                                                           ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
+        // bf16
+        pass = ck::profiler::profile_conv_bwd_data_impl<1,
+                                                        ck::tensor_layout::convolution::NWC,
+                                                        ck::tensor_layout::convolution::KXC,
+                                                        ck::tensor_layout::convolution::NWK,
+                                                        ck::bhalf_t,
+                                                        ck::bhalf_t,
+                                                        ck::bhalf_t>(true,  // do_verification
+                                                                     1,     // init_method
+                                                                     false, // do_log
+                                                                     false, // time_kernel
+                                                                     param);
 
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<2,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NHWC,
-                                                           ck::tensor_layout::convolution::KYXC,
-                                                           ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
+        EXPECT_TRUE(pass);
 
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<2,
-                                                           ck::bhalf_t,
-                                                           ck::bhalf_t,
-                                                           ck::bhalf_t,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NHWC,
-                                                           ck::tensor_layout::convolution::KYXC,
-                                                           ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
+        // int8
+        pass = ck::profiler::profile_conv_bwd_data_impl<1,
+                                                        ck::tensor_layout::convolution::NWC,
+                                                        ck::tensor_layout::convolution::KXC,
+                                                        ck::tensor_layout::convolution::NWK,
+                                                        int8_t,
+                                                        int8_t,
+                                                        int8_t>(true,  // do_verification
+                                                                1,     // init_method
+                                                                false, // do_log
+                                                                false, // time_kernel
+                                                                param);
 
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<2,
-                                                           int8_t,
-                                                           int8_t,
-                                                           int8_t,
-                                                           int,
-                                                           ck::tensor_layout::convolution::NHWC,
-                                                           ck::tensor_layout::convolution::KYXC,
-                                                           ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
-    }
-
-    // check 3d
-    params.clear();
-    params.push_back(
-        {3, 128, 128, 256, {1, 1, 1}, {7, 7, 7}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
-    params.push_back(
-        {3, 128, 128, 256, {3, 3, 3}, {14, 14, 14}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}});
-    params.push_back(
-        {3, 128, 128, 256, {1, 1, 1}, {3, 3, 3}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
-
-    for(auto& param : params)
-    {
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<3,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NDHWC,
-                                                           ck::tensor_layout::convolution::KZYXC,
-                                                           ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
-
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<3,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           ck::half_t,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NDHWC,
-                                                           ck::tensor_layout::convolution::KZYXC,
-                                                           ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
-
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<3,
-                                                           ck::bhalf_t,
-                                                           ck::bhalf_t,
-                                                           ck::bhalf_t,
-                                                           float,
-                                                           ck::tensor_layout::convolution::NDHWC,
-                                                           ck::tensor_layout::convolution::KZYXC,
-                                                           ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
-
-        pass &= ck::profiler::profile_convnd_bwd_data_impl<3,
-                                                           int8_t,
-                                                           int8_t,
-                                                           int8_t,
-                                                           int,
-                                                           ck::tensor_layout::convolution::NDHWC,
-                                                           ck::tensor_layout::convolution::KZYXC,
-                                                           ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            false, // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_);
-    }
-
-    if(pass)
-    {
-        std::cout << "test convnd bwd : Pass" << std::endl;
-        return 0;
-    }
-    else
-    {
-        std::cout << "test convnd bwd: Fail " << std::endl;
-        return -1;
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 2d
+TEST_F(TestConvndBwdData, Conv2dBwdData)
+{
+    conv_params.clear();
+    conv_params.push_back({2, 1, 128, 128, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
+    conv_params.push_back({2, 1, 128, 128, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
+    conv_params.push_back({2, 1, 128, 128, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_bwd_data_impl<2,
+                                                        ck::tensor_layout::convolution::NHWC,
+                                                        ck::tensor_layout::convolution::KYXC,
+                                                        ck::tensor_layout::convolution::NHWK,
+                                                        float,
+                                                        float,
+                                                        float>(true,  // do_verification
+                                                               1,     // init_method
+                                                               false, // do_log
+                                                               false, // time_kernel
+                                                               param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_bwd_data_impl<2,
+                                                        ck::tensor_layout::convolution::NHWC,
+                                                        ck::tensor_layout::convolution::KYXC,
+                                                        ck::tensor_layout::convolution::NHWK,
+                                                        ck::half_t,
+                                                        ck::half_t,
+                                                        ck::half_t>(true,  // do_verification
+                                                                    1,     // init_method
+                                                                    false, // do_log
+                                                                    false, // time_kernel
+                                                                    param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_bwd_data_impl<2,
+                                                        ck::tensor_layout::convolution::NHWC,
+                                                        ck::tensor_layout::convolution::KYXC,
+                                                        ck::tensor_layout::convolution::NHWK,
+                                                        ck::bhalf_t,
+                                                        ck::bhalf_t,
+                                                        ck::bhalf_t>(true,  // do_verification
+                                                                     1,     // init_method
+                                                                     false, // do_log
+                                                                     false, // time_kernel
+                                                                     param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_conv_bwd_data_impl<2,
+                                                        ck::tensor_layout::convolution::NHWC,
+                                                        ck::tensor_layout::convolution::KYXC,
+                                                        ck::tensor_layout::convolution::NHWK,
+                                                        int8_t,
+                                                        int8_t,
+                                                        int8_t>(true,  // do_verification
+                                                                1,     // init_method
+                                                                false, // do_log
+                                                                false, // time_kernel
+                                                                param);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 3d
+TEST_F(TestConvndBwdData, Conv3dBwdData)
+{
+    conv_params.clear();
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {1, 1, 1}, {7, 7, 7}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {3, 3, 3}, {14, 14, 3}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}});
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {1, 1, 1}, {3, 3, 3}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_bwd_data_impl<3,
+                                                        ck::tensor_layout::convolution::NDHWC,
+                                                        ck::tensor_layout::convolution::KZYXC,
+                                                        ck::tensor_layout::convolution::NDHWK,
+                                                        float,
+                                                        float,
+                                                        float>(true,  // do_verification
+                                                               1,     // init_method
+                                                               false, // do_log
+                                                               false, // time_kernel
+                                                               param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_bwd_data_impl<3,
+                                                        ck::tensor_layout::convolution::NDHWC,
+                                                        ck::tensor_layout::convolution::KZYXC,
+                                                        ck::tensor_layout::convolution::NDHWK,
+                                                        ck::half_t,
+                                                        ck::half_t,
+                                                        ck::half_t>(true,  // do_verification
+                                                                    1,     // init_method
+                                                                    false, // do_log
+                                                                    false, // time_kernel
+                                                                    param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_bwd_data_impl<3,
+                                                        ck::tensor_layout::convolution::NDHWC,
+                                                        ck::tensor_layout::convolution::KZYXC,
+                                                        ck::tensor_layout::convolution::NDHWK,
+                                                        ck::bhalf_t,
+                                                        ck::bhalf_t,
+                                                        ck::bhalf_t>(true,  // do_verification
+                                                                     1,     // init_method
+                                                                     false, // do_log
+                                                                     false, // time_kernel
+                                                                     param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_conv_bwd_data_impl<3,
+                                                        ck::tensor_layout::convolution::NDHWC,
+                                                        ck::tensor_layout::convolution::KZYXC,
+                                                        ck::tensor_layout::convolution::NDHWK,
+                                                        int8_t,
+                                                        int8_t,
+                                                        int8_t>(true,  // do_verification
+                                                                1,     // init_method
+                                                                false, // do_log
+                                                                false, // time_kernel
+                                                                param);
+
+        EXPECT_TRUE(pass);
     }
 }
diff --git a/test/convnd_bwd_weight/CMakeLists.txt b/test/convnd_bwd_weight/CMakeLists.txt
index e76c28bf4f..cfbbf1bb41 100644
--- a/test/convnd_bwd_weight/CMakeLists.txt
+++ b/test/convnd_bwd_weight/CMakeLists.txt
@@ -1,2 +1,2 @@
-add_test_executable(test_convnd_bwd_weight convnd_bwd_weight.cpp) 
-target_link_libraries(test_convnd_bwd_weight PRIVATE host_tensor device_convnd_bwd_weight_instance conv_util)
+add_gtest_executable(test_convnd_bwd_weight convnd_bwd_weight.cpp) 
+target_link_libraries(test_convnd_bwd_weight PRIVATE utility device_conv1d_bwd_weight_instance device_conv2d_bwd_weight_instance  device_conv3d_bwd_weight_instance)
diff --git a/test/convnd_bwd_weight/convnd_bwd_weight.cpp b/test/convnd_bwd_weight/convnd_bwd_weight.cpp
index febcef16c0..af27282f19 100644
--- a/test/convnd_bwd_weight/convnd_bwd_weight.cpp
+++ b/test/convnd_bwd_weight/convnd_bwd_weight.cpp
@@ -1,283 +1,205 @@
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
-#include <iostream>
-#include <numeric>
-#include <initializer_list>
 #include <cstdlib>
+#include <iostream>
+#include <initializer_list>
 #include <vector>
+#include <gtest/gtest.h>
 
-#include "test/convnd_fwd/conv_util.hpp"
-#include "profiler/include/profile_convnd_bwd_weight_impl.hpp"
+#include "profiler/include/profile_conv_bwd_weight_impl.hpp"
 
-int test_self()
+class TestConvndBwdWeight : public ::testing::Test
 {
-    bool pass = true;
-    std::vector<ck::utils::conv::ConvParams> params;
+    protected:
+    std::vector<ck::utils::conv::ConvParam> conv_params;
+};
 
-    params.push_back({1, 128, 256, 256, {1}, {7}, {2}, {1}, {0}, {0}});
-    params.push_back({1, 128, 256, 256, {3}, {14}, {1}, {1}, {1}, {1}});
-    params.push_back({1, 128, 256, 256, {1}, {3}, {1}, {1}, {0}, {0}});
-
-    for(auto& param : params)
-    {
-        // f32
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<1,
-                                                             float,
-                                                             float,
-                                                             float,
-                                                             ck::tensor_layout::convolution::NWC,
-                                                             ck::tensor_layout::convolution::KXC,
-                                                             ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // fp16
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<1,
-                                                             ck::half_t,
-                                                             ck::half_t,
-                                                             ck::half_t,
-                                                             ck::tensor_layout::convolution::NWC,
-                                                             ck::tensor_layout::convolution::KXC,
-                                                             ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // bf16
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<1,
-                                                             ck::bhalf_t,
-                                                             ck::bhalf_t,
-                                                             ck::bhalf_t,
-                                                             ck::tensor_layout::convolution::NWC,
-                                                             ck::tensor_layout::convolution::KXC,
-                                                             ck::tensor_layout::convolution::NWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-    }
-
-    // check 2d
-    params.clear();
-    params.push_back({2, 128, 256, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
-    params.push_back({2, 128, 256, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
-    params.push_back({2, 128, 256, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
-
-    for(auto& param : params)
-    {
-        // f32
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<2,
-                                                             float,
-                                                             float,
-                                                             float,
-                                                             ck::tensor_layout::convolution::NHWC,
-                                                             ck::tensor_layout::convolution::KYXC,
-                                                             ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // fp16
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<2,
-                                                             ck::half_t,
-                                                             ck::half_t,
-                                                             ck::half_t,
-                                                             ck::tensor_layout::convolution::NHWC,
-                                                             ck::tensor_layout::convolution::KYXC,
-                                                             ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // bf16
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<2,
-                                                             ck::bhalf_t,
-                                                             ck::bhalf_t,
-                                                             ck::bhalf_t,
-                                                             ck::tensor_layout::convolution::NHWC,
-                                                             ck::tensor_layout::convolution::KYXC,
-                                                             ck::tensor_layout::convolution::NHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-    }
-
-    // check 2d
-    params.clear();
-    params.push_back(
-        {3, 128, 256, 256, {1, 1, 1}, {4, 4, 4}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
-    params.push_back(
-        {3, 128, 256, 256, {3, 3, 3}, {4, 4, 8}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}});
-    params.push_back(
-        {3, 128, 256, 256, {1, 1, 1}, {3, 3, 3}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
-
-    for(auto& param : params)
-    {
-        // f32
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<3,
-                                                             float,
-                                                             float,
-                                                             float,
-                                                             ck::tensor_layout::convolution::NDHWC,
-                                                             ck::tensor_layout::convolution::KZYXC,
-                                                             ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // fp16
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<3,
-                                                             ck::half_t,
-                                                             ck::half_t,
-                                                             ck::half_t,
-                                                             ck::tensor_layout::convolution::NDHWC,
-                                                             ck::tensor_layout::convolution::KZYXC,
-                                                             ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-
-        // bf16
-        pass &= ck::profiler::profile_convnd_bwd_weight_impl<3,
-                                                             ck::bhalf_t,
-                                                             ck::bhalf_t,
-                                                             ck::bhalf_t,
-                                                             ck::tensor_layout::convolution::NDHWC,
-                                                             ck::tensor_layout::convolution::KZYXC,
-                                                             ck::tensor_layout::convolution::NDHWK>(
-            true,  // do_verification
-            1,     // init_method
-            false, // do_log
-            true,  // time_kernel
-            param.N_,
-            param.K_,
-            param.C_,
-            param.input_spatial_lengths_,
-            param.filter_spatial_lengths_,
-            param.GetOutputSpatialLengths(),
-            param.conv_filter_strides_,
-            param.conv_filter_dilations_,
-            param.input_left_pads_,
-            param.input_right_pads_,
-            2);
-    }
-
-    return pass;
-}
-int main()
+// 1d
+TEST_F(TestConvndBwdWeight, Conv1dBwdWeight)
 {
-    // int data_type   = 1;
-    // int init_method = 1;
+    conv_params.clear();
+    conv_params.push_back({1, 1, 128, 128, 256, {1}, {14}, {2}, {1}, {0}, {0}});
+    conv_params.push_back({1, 1, 128, 128, 256, {3}, {28}, {1}, {1}, {1}, {1}});
+    conv_params.push_back({1, 1, 128, 128, 256, {1}, {3}, {1}, {1}, {0}, {0}});
 
-    bool pass = true;
-
-    pass = test_self();
-
-    if(pass)
+    for(auto& param : conv_params)
     {
-        std::cout << "test conv2d bwd weight : Pass" << std::endl;
-        return 0;
-    }
-    else
-    {
-        std::cout << "test conv2d bwd weight: Fail " << std::endl;
-        return -1;
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_bwd_weight_impl<1,
+                                                          ck::tensor_layout::convolution::NWC,
+                                                          ck::tensor_layout::convolution::KXC,
+                                                          ck::tensor_layout::convolution::NWK,
+                                                          float,
+                                                          float,
+                                                          float>(true,  // do_verification
+                                                                 1,     // init_method
+                                                                 false, // do_log
+                                                                 false, // time_kernel
+                                                                 param,
+                                                                 2);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_bwd_weight_impl<1,
+                                                          ck::tensor_layout::convolution::NWC,
+                                                          ck::tensor_layout::convolution::KXC,
+                                                          ck::tensor_layout::convolution::NWK,
+                                                          ck::half_t,
+                                                          ck::half_t,
+                                                          ck::half_t>(true,  // do_verification
+                                                                      1,     // init_method
+                                                                      false, // do_log
+                                                                      false, // time_kernel
+                                                                      param,
+                                                                      2);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_bwd_weight_impl<1,
+                                                          ck::tensor_layout::convolution::NWC,
+                                                          ck::tensor_layout::convolution::KXC,
+                                                          ck::tensor_layout::convolution::NWK,
+                                                          ck::bhalf_t,
+                                                          ck::bhalf_t,
+                                                          ck::bhalf_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param,
+                                                                       2);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 2d
+TEST_F(TestConvndBwdWeight, Conv2dBwdWeight)
+{
+    conv_params.clear();
+    conv_params.push_back({2, 1, 128, 128, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
+    conv_params.push_back({2, 1, 32, 128, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
+    conv_params.push_back({2, 1, 128, 128, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_bwd_weight_impl<2,
+                                                          ck::tensor_layout::convolution::NHWC,
+                                                          ck::tensor_layout::convolution::KYXC,
+                                                          ck::tensor_layout::convolution::NHWK,
+                                                          float,
+                                                          float,
+                                                          float>(true,  // do_verification
+                                                                 1,     // init_method
+                                                                 false, // do_log
+                                                                 false, // time_kernel
+                                                                 param,
+                                                                 2);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_bwd_weight_impl<2,
+                                                          ck::tensor_layout::convolution::NHWC,
+                                                          ck::tensor_layout::convolution::KYXC,
+                                                          ck::tensor_layout::convolution::NHWK,
+                                                          ck::half_t,
+                                                          ck::half_t,
+                                                          ck::half_t>(true,  // do_verification
+                                                                      1,     // init_method
+                                                                      false, // do_log
+                                                                      false, // time_kernel
+                                                                      param,
+                                                                      2);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_bwd_weight_impl<2,
+                                                          ck::tensor_layout::convolution::NHWC,
+                                                          ck::tensor_layout::convolution::KYXC,
+                                                          ck::tensor_layout::convolution::NHWK,
+                                                          ck::bhalf_t,
+                                                          ck::bhalf_t,
+                                                          ck::bhalf_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param,
+                                                                       2);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 3d
+TEST_F(TestConvndBwdWeight, Conv3dBwdWeight)
+{
+    conv_params.clear();
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {1, 1, 1}, {7, 7, 7}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+    conv_params.push_back(
+        {3, 1, 32, 128, 256, {3, 3, 3}, {14, 14, 3}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}});
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {1, 1, 1}, {3, 3, 3}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_bwd_weight_impl<3,
+                                                          ck::tensor_layout::convolution::NDHWC,
+                                                          ck::tensor_layout::convolution::KZYXC,
+                                                          ck::tensor_layout::convolution::NDHWK,
+                                                          float,
+                                                          float,
+                                                          float>(true,  // do_verification
+                                                                 1,     // init_method
+                                                                 false, // do_log
+                                                                 false, // time_kernel
+                                                                 param,
+                                                                 2);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_bwd_weight_impl<3,
+                                                          ck::tensor_layout::convolution::NDHWC,
+                                                          ck::tensor_layout::convolution::KZYXC,
+                                                          ck::tensor_layout::convolution::NDHWK,
+                                                          ck::half_t,
+                                                          ck::half_t,
+                                                          ck::half_t>(true,  // do_verification
+                                                                      1,     // init_method
+                                                                      false, // do_log
+                                                                      false, // time_kernel
+                                                                      param,
+                                                                      2);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_bwd_weight_impl<3,
+                                                          ck::tensor_layout::convolution::NDHWC,
+                                                          ck::tensor_layout::convolution::KZYXC,
+                                                          ck::tensor_layout::convolution::NDHWK,
+                                                          ck::bhalf_t,
+                                                          ck::bhalf_t,
+                                                          ck::bhalf_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param,
+                                                                       2);
+
+        EXPECT_TRUE(pass);
     }
 }
diff --git a/test/convnd_fwd/CMakeLists.txt b/test/convnd_fwd/CMakeLists.txt
index 444ec6c8aa..97e170d851 100644
--- a/test/convnd_fwd/CMakeLists.txt
+++ b/test/convnd_fwd/CMakeLists.txt
@@ -1,13 +1,2 @@
-add_custom_target(test_convnd_fwd)
-
-add_gtest_executable(test_conv1d_fwd conv1d_fwd.cpp)
-target_link_libraries(test_conv1d_fwd PRIVATE host_tensor device_conv1d_fwd_instance conv_util)
-add_dependencies(test_convnd_fwd test_conv1d_fwd)
-
-add_gtest_executable(test_conv2d_fwd conv2d_fwd.cpp)
-target_link_libraries(test_conv2d_fwd PRIVATE host_tensor device_conv2d_fwd_instance device_convnd_2d_fwd_instance conv_util)
-add_dependencies(test_convnd_fwd test_conv2d_fwd)
-
-add_gtest_executable(test_conv3d_fwd conv3d_fwd.cpp)
-target_link_libraries(test_conv3d_fwd PRIVATE host_tensor device_conv3d_fwd_instance conv_util)
-add_dependencies(test_convnd_fwd test_conv3d_fwd)
+add_gtest_executable(test_convnd_fwd convnd_fwd.cpp)
+target_link_libraries(test_convnd_fwd PRIVATE utility device_conv2d_fwd_instance)
diff --git a/test/convnd_fwd/conv1d_fwd.cpp b/test/convnd_fwd/conv1d_fwd.cpp
deleted file mode 100644
index 4d2473f020..0000000000
--- a/test/convnd_fwd/conv1d_fwd.cpp
+++ /dev/null
@@ -1,192 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <tuple>
-#include <vector>
-#include <gtest/gtest.h>
-
-#include "ck/utility/data_type.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "test/convnd_fwd/conv_util.hpp"
-
-namespace {
-
-class Conv1dFwdNWCInstances : public ::testing::Test
-{
-    public:
-    template <typename T>
-    bool test_conv1d_nwc_instances(const std::vector<test::conv::DeviceConvFwdNoOpPtr>& conv_ptrs,
-                                   const ck::utils::conv::ConvParams& params)
-    {
-        using namespace std::placeholders;
-        using namespace ck::utils;
-        namespace ctl = ck::tensor_layout::convolution;
-
-        conv::ConvFwdOpInstance<T,
-                                T,
-                                T,
-                                ctl::NWC,
-                                ctl::KXC,
-                                ctl::NWK,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                FillUniformDistributionIntegerValue<T>,
-                                FillUniformDistributionIntegerValue<T>>
-            conv_instance(params,
-                          true,
-                          FillUniformDistributionIntegerValue<T>{},
-                          FillUniformDistributionIntegerValue<T>{});
-        auto reference_conv_fwd_fun =
-            std::bind(conv::run_reference_convolution_forward<1, T, T, T>, params, _1, _2, _3);
-        OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-        run_engine.SetAtol(atol_);
-        run_engine.SetRtol(rtol_);
-        return run_engine.Test(conv_ptrs);
-    }
-
-    template <typename T>
-    bool test_default()
-    {
-        return test_conv1d_nwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<1>(), params_default_);
-    }
-
-    template <typename T>
-    bool test_filter1x1_stride1_pad0()
-    {
-        return test_conv1d_nwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<1>(),
-            params_filter1x1_stride1_pad0_);
-    }
-
-    template <typename T>
-    bool test_filter1x1_pad0()
-    {
-        return test_conv1d_nwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<1>(),
-            params_filter1x1_pad0_);
-    }
-
-    static inline ck::utils::conv::ConvParams params_default_{
-        1, 4, 256, 64, {3}, {71}, {2}, {2}, {2}, {2}};
-    static inline ck::utils::conv::ConvParams params_filter1x1_stride1_pad0_{
-        1, 4, 256, 64, {1}, {28}, {1}, {1}, {0}, {0}};
-    static inline ck::utils::conv::ConvParams params_filter1x1_pad0_{
-        1, 4, 256, 64, {1}, {28}, {2}, {1}, {0}, {0}};
-
-    private:
-    double atol_{1e-5};
-    double rtol_{1e-4};
-};
-
-} // anonymous namespace
-
-TEST(Conv1DFwdNWC, IntegerValues)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-    namespace ctl = ck::tensor_layout::convolution;
-    using T       = float;
-
-    ck::utils::conv::ConvParams params{1, 4, 256, 64, {3}, {36}, {1}, {2}, {2}, {2}};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<1, T, T, T, T>(conv_ptrs);
-    conv::ConvFwdOpInstance<T,
-                            T,
-                            T,
-                            ctl::NWC,
-                            ctl::KXC,
-                            ctl::NWK,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            FillUniformDistributionIntegerValue<T>,
-                            FillUniformDistributionIntegerValue<T>>
-        conv_instance(params,
-                      true,
-                      FillUniformDistributionIntegerValue<T>{},
-                      FillUniformDistributionIntegerValue<T>{});
-
-    auto reference_conv_fwd_fun =
-        std::bind(conv::run_reference_convolution_forward<1, T, T, T>, params, _1, _2, _3);
-    OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-    run_engine.SetAtol(1e-5);
-    run_engine.SetRtol(1e-4);
-    EXPECT_TRUE(run_engine.Test(conv_ptrs));
-}
-
-TEST(Conv1DFwdNWC, FloatingPointValues)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-    namespace ctl = ck::tensor_layout::convolution;
-    using T       = ck::half_t;
-
-    ck::utils::conv::ConvParams params{1, 4, 256, 64, {3}, {36}, {1}, {2}, {2}, {2}};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<1, T, T, T, float>(conv_ptrs);
-    conv::ConvFwdOpInstance<T,
-                            T,
-                            T,
-                            ctl::NWC,
-                            ctl::KXC,
-                            ctl::NWK,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            FillUniformDistribution<T>,
-                            FillUniformDistribution<T>>
-        conv_instance(params, true, FillUniformDistribution<T>{}, FillUniformDistribution<T>{});
-
-    auto reference_conv_fwd_fun =
-        std::bind(conv::run_reference_convolution_forward<1, T, T, T>, params, _1, _2, _3);
-    OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-    run_engine.SetAtol(0.1);
-    run_engine.SetRtol(1e-2);
-    EXPECT_TRUE(run_engine.Test(conv_ptrs));
-}
-
-TEST_F(Conv1dFwdNWCInstances, BF16_default) { EXPECT_TRUE(this->test_default<ck::bhalf_t>()); }
-TEST_F(Conv1dFwdNWCInstances, BF16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::bhalf_t>());
-}
-TEST_F(Conv1dFwdNWCInstances, BF16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::bhalf_t>());
-}
-
-TEST_F(Conv1dFwdNWCInstances, F16_default) { EXPECT_TRUE(this->test_default<ck::half_t>()); }
-TEST_F(Conv1dFwdNWCInstances, F16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::half_t>());
-}
-TEST_F(Conv1dFwdNWCInstances, F16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::half_t>());
-}
-
-TEST_F(Conv1dFwdNWCInstances, F32_default) { EXPECT_TRUE(this->test_default<float>()); }
-TEST_F(Conv1dFwdNWCInstances, F32_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<float>());
-}
-TEST_F(Conv1dFwdNWCInstances, F32_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<float>());
-}
-
-TEST_F(Conv1dFwdNWCInstances, I8_default) { EXPECT_TRUE(this->test_default<int8_t>()); }
-TEST_F(Conv1dFwdNWCInstances, I8_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<int8_t>());
-}
-TEST_F(Conv1dFwdNWCInstances, I8_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<int8_t>());
-}
diff --git a/test/convnd_fwd/conv2d_fwd.cpp b/test/convnd_fwd/conv2d_fwd.cpp
deleted file mode 100644
index f45805782c..0000000000
--- a/test/convnd_fwd/conv2d_fwd.cpp
+++ /dev/null
@@ -1,266 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <tuple>
-#include <vector>
-#include <gtest/gtest.h>
-
-#include "ck/utility/data_type.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/utility/conv_util.hpp"
-#include "test/convnd_fwd/conv_util.hpp"
-
-namespace {
-
-class Conv2dFwdNHWCInstances : public ::testing::Test
-{
-    public:
-    template <typename T>
-    bool test_conv2d_nhwc_instances(const std::vector<test::conv::DeviceConvFwdNoOpPtr>& conv_ptrs,
-                                    const ck::utils::conv::ConvParams& params)
-    {
-        using namespace std::placeholders;
-        using namespace ck::utils;
-
-        conv::ConvFwdOpInstance<T,
-                                T,
-                                T,
-                                ck::tensor_layout::convolution::NHWC,
-                                ck::tensor_layout::convolution::KYXC,
-                                ck::tensor_layout::convolution::NHWK,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                FillUniformDistributionIntegerValue<T>,
-                                FillUniformDistributionIntegerValue<T>>
-            conv_instance(params,
-                          true,
-                          FillUniformDistributionIntegerValue<T>{},
-                          FillUniformDistributionIntegerValue<T>{});
-        auto reference_conv_fwd_fun =
-            std::bind(conv::run_reference_convolution_forward<2, T, T, T>, params, _1, _2, _3);
-        OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-        run_engine.SetAtol(atol_);
-        run_engine.SetRtol(rtol_);
-        return run_engine.Test(conv_ptrs);
-    }
-
-    template <typename T>
-    bool test_default(bool use_convnd = false)
-    {
-        if(use_convnd)
-        {
-            return test_conv2d_nhwc_instances<T>(
-                test::conv::ConvolutionNDFwdInstances<T, T, T>::Get(2), params_default_);
-        }
-        else
-        {
-            return test_conv2d_nhwc_instances<T>(
-                ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<2>(),
-                params_default_);
-        }
-    }
-
-    template <typename T>
-    bool test_filter1x1_stride1_pad0(bool use_convnd = false)
-    {
-        if(use_convnd)
-        {
-            return test_conv2d_nhwc_instances<T>(
-                test::conv::ConvolutionNDFwdInstances<T, T, T>::Get(2),
-                params_filter1x1_stride1_pad0_);
-        }
-        else
-        {
-            return test_conv2d_nhwc_instances<T>(
-                ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<2>(),
-                params_filter1x1_stride1_pad0_);
-        }
-    }
-
-    template <typename T>
-    bool test_filter1x1_pad0(bool use_convnd = false)
-    {
-        if(use_convnd)
-        {
-            return test_conv2d_nhwc_instances<T>(
-                test::conv::ConvolutionNDFwdInstances<T, T, T>::Get(2), params_filter1x1_pad0_);
-        }
-        else
-        {
-            return test_conv2d_nhwc_instances<T>(
-                ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<2>(),
-                params_filter1x1_pad0_);
-        }
-    }
-
-    template <typename T>
-    bool test_oddC()
-    {
-        return test_conv2d_nhwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<2>(), params_oddC_);
-    }
-
-    static inline ck::utils::conv::ConvParams params_default_{
-        2, 4, 256, 64, {3, 3}, {36, 36}, {2, 2}, {2, 2}, {2, 2}, {2, 2}};
-    static inline ck::utils::conv::ConvParams params_filter1x1_stride1_pad0_{
-        2, 4, 256, 64, {1, 1}, {28, 28}, {1, 1}, {1, 1}, {0, 0}, {0, 0}};
-    static inline ck::utils::conv::ConvParams params_filter1x1_pad0_{
-        2, 4, 256, 64, {1, 1}, {28, 28}, {2, 2}, {1, 1}, {0, 0}, {0, 0}};
-    static inline ck::utils::conv::ConvParams params_oddC_{
-        2, 4, 256, 3, {3, 3}, {28, 28}, {1, 1}, {1, 1}, {0, 0}, {0, 0}};
-
-    private:
-    double atol_{1e-5};
-    double rtol_{1e-4};
-};
-
-} // anonymous namespace
-
-TEST(Conv2DFwdNHWC, IntegerValues)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-    using T = float;
-
-    ck::utils::conv::ConvParams params{
-        2, 4, 256, 64, {3, 3}, {36, 36}, {1, 1}, {2, 2}, {2, 2}, {2, 2}};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<2, T, T, T, T>(conv_ptrs);
-    conv::ConvFwdOpInstance<T,
-                            T,
-                            T,
-                            ck::tensor_layout::convolution::NHWC,
-                            ck::tensor_layout::convolution::KYXC,
-                            ck::tensor_layout::convolution::NHWK,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            FillUniformDistributionIntegerValue<T>,
-                            FillUniformDistributionIntegerValue<T>>
-        conv_instance(params,
-                      true,
-                      FillUniformDistributionIntegerValue<T>{},
-                      FillUniformDistributionIntegerValue<T>{});
-
-    auto reference_conv_fwd_fun =
-        std::bind(conv::run_reference_convolution_forward<2, T, T, T>, params, _1, _2, _3);
-    OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-    run_engine.SetAtol(1e-5);
-    run_engine.SetRtol(1e-4);
-    EXPECT_TRUE(run_engine.Test(conv_ptrs));
-}
-
-TEST(Conv2DFwdNHWC, FloatingPointValues)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-    using T = ck::half_t;
-
-    ck::utils::conv::ConvParams params{
-        2, 4, 256, 64, {3, 3}, {36, 36}, {2, 2}, {2, 2}, {2, 2}, {2, 2}};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<2, T, T, T, float>(conv_ptrs);
-    conv::ConvFwdOpInstance<T,
-                            T,
-                            T,
-                            ck::tensor_layout::convolution::NHWC,
-                            ck::tensor_layout::convolution::KYXC,
-                            ck::tensor_layout::convolution::NHWK,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            FillUniformDistribution<T>,
-                            FillUniformDistribution<T>>
-        conv_instance(params, true, FillUniformDistribution<T>{}, FillUniformDistribution<T>{});
-
-    auto reference_conv_fwd_fun =
-        std::bind(conv::run_reference_convolution_forward<2, T, T, T>, params, _1, _2, _3);
-    OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-    run_engine.SetAtol(2e-4);
-    run_engine.SetRtol(1e-3);
-    EXPECT_TRUE(run_engine.Test(conv_ptrs));
-}
-
-TEST_F(Conv2dFwdNHWCInstances, BF16_default) { EXPECT_TRUE(this->test_default<ck::bhalf_t>()); }
-TEST_F(Conv2dFwdNHWCInstances, BF16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::bhalf_t>());
-}
-TEST_F(Conv2dFwdNHWCInstances, BF16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::bhalf_t>());
-}
-TEST_F(Conv2dFwdNHWCInstances, F16_default) { EXPECT_TRUE(this->test_default<ck::half_t>()); }
-TEST_F(Conv2dFwdNHWCInstances, F16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::half_t>());
-}
-TEST_F(Conv2dFwdNHWCInstances, F16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::half_t>());
-}
-TEST_F(Conv2dFwdNHWCInstances, F16_oddC) { EXPECT_TRUE(this->test_oddC<ck::half_t>()); }
-TEST_F(Conv2dFwdNHWCInstances, F32_default) { EXPECT_TRUE(this->test_default<float>()); }
-TEST_F(Conv2dFwdNHWCInstances, F32_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<float>());
-}
-TEST_F(Conv2dFwdNHWCInstances, F32_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<float>());
-}
-TEST_F(Conv2dFwdNHWCInstances, I8_default) { EXPECT_TRUE(this->test_default<int8_t>()); }
-TEST_F(Conv2dFwdNHWCInstances, I8_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<int8_t>());
-}
-TEST_F(Conv2dFwdNHWCInstances, I8_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<int8_t>());
-}
-
-TEST_F(Conv2dFwdNHWCInstances, ND_BF16_default)
-{
-    EXPECT_TRUE(this->test_default<ck::bhalf_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_BF16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::bhalf_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_BF16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::bhalf_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_F16_default)
-{
-    EXPECT_TRUE(this->test_default<ck::half_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_F16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::half_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_F16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::half_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_F32_default) { EXPECT_TRUE(this->test_default<float>(true)); }
-TEST_F(Conv2dFwdNHWCInstances, ND_F32_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<float>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_F32_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<float>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_I8_default) { EXPECT_TRUE(this->test_default<int8_t>(true)); }
-TEST_F(Conv2dFwdNHWCInstances, ND_I8_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<int8_t>(true));
-}
-TEST_F(Conv2dFwdNHWCInstances, ND_I8_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<int8_t>(true));
-}
diff --git a/test/convnd_fwd/conv3d_fwd.cpp b/test/convnd_fwd/conv3d_fwd.cpp
deleted file mode 100644
index 0cc2b2416e..0000000000
--- a/test/convnd_fwd/conv3d_fwd.cpp
+++ /dev/null
@@ -1,317 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <iostream>
-#include <stdexcept>
-#include <tuple>
-#include <vector>
-#include <gtest/gtest.h>
-
-#include "ck/utility/data_type.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-
-#include "ck/library/utility/conv_util.hpp"
-
-#include "test/convnd_fwd/conv_util.hpp"
-
-namespace {
-
-class Conv3dFwdNDHWCInstances : public ::testing::Test
-{
-    public:
-    template <typename T>
-    bool test_conv3d_nwc_instances(const std::vector<test::conv::DeviceConvFwdNoOpPtr>& conv_ptrs,
-                                   const ck::utils::conv::ConvParams& params)
-    {
-        using namespace std::placeholders;
-        using namespace ck::utils;
-        namespace ctl = ck::tensor_layout::convolution;
-
-        conv::ConvFwdOpInstance<T,
-                                T,
-                                T,
-                                ctl::NDHWC,
-                                ctl::KZYXC,
-                                ctl::NDHWK,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                ck::tensor_operation::element_wise::PassThrough,
-                                FillUniformDistributionIntegerValue<T>,
-                                FillUniformDistributionIntegerValue<T>>
-            conv_instance(params,
-                          true,
-                          FillUniformDistributionIntegerValue<T>{},
-                          FillUniformDistributionIntegerValue<T>{});
-        auto reference_conv_fwd_fun =
-            std::bind(conv::run_reference_convolution_forward<3, T, T, T>, params, _1, _2, _3);
-        OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-        run_engine.SetAtol(atol_);
-        run_engine.SetRtol(rtol_);
-        return run_engine.Test(conv_ptrs);
-    }
-
-    template <typename T>
-    bool test_default()
-    {
-        return test_conv3d_nwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<3>(), params_default_);
-    }
-
-    template <typename T>
-    bool test_filter1x1_stride1_pad0()
-    {
-        return test_conv3d_nwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<3>(),
-            params_filter1x1_stride1_pad0_);
-    }
-
-    template <typename T>
-    bool test_filter1x1_pad0()
-    {
-        return test_conv3d_nwc_instances<T>(
-            ck::utils::conv::ConvolutionFwdInstances<T, T, T>::template Get<3>(),
-            params_filter1x1_pad0_);
-    }
-
-    static inline ck::utils::conv::ConvParams params_default_{
-        3, 4, 256, 64, {3, 3, 3}, {28, 28, 28}, {2, 2, 2}, {2, 2, 2}, {2, 2, 2}, {2, 2, 2}};
-    static inline ck::utils::conv::ConvParams params_filter1x1_stride1_pad0_{
-        3, 4, 256, 64, {1, 1, 1}, {28, 28, 28}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}};
-    static inline ck::utils::conv::ConvParams params_filter1x1_pad0_{
-        3, 4, 256, 64, {1, 1, 1}, {28, 28, 28}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}};
-
-    private:
-    double atol_{1e-5};
-    double rtol_{1e-4};
-};
-
-} // anonymous namespace
-
-TEST(Conv3DFwdNDHWC, IntegerValues)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-    namespace ctl = ck::tensor_layout::convolution;
-    using T       = float;
-
-    ck::utils::conv::ConvParams params{
-        3, 4, 256, 64, {3, 3, 3}, {18, 18, 18}, {1, 1, 1}, {2, 2, 2}, {2, 2, 2}, {2, 2, 2}};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<3, T, T, T, T>(conv_ptrs);
-    conv::ConvFwdOpInstance<T,
-                            T,
-                            T,
-                            ctl::NDHWC,
-                            ctl::KZYXC,
-                            ctl::NDHWK,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            FillUniformDistributionIntegerValue<T>,
-                            FillUniformDistributionIntegerValue<T>>
-        conv_instance(params,
-                      true,
-                      FillUniformDistributionIntegerValue<T>{},
-                      FillUniformDistributionIntegerValue<T>{});
-
-    auto reference_conv_fwd_fun =
-        std::bind(conv::run_reference_convolution_forward<3, T, T, T>, params, _1, _2, _3);
-    OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-    run_engine.SetAtol(1e-5);
-    run_engine.SetRtol(1e-3);
-    EXPECT_TRUE(run_engine.Test(conv_ptrs));
-}
-
-TEST(Conv3DFwdNDHWC, FloatingPointValues)
-{
-    using namespace std::placeholders;
-    using namespace ck::utils;
-    namespace ctl = ck::tensor_layout::convolution;
-    using T       = ck::half_t;
-
-    ck::utils::conv::ConvParams params{
-        3, 4, 256, 64, {3, 3, 3}, {18, 18, 18}, {1, 1, 1}, {2, 2, 2}, {2, 2, 2}, {2, 2, 2}};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<3, T, T, T, float>(conv_ptrs);
-    conv::ConvFwdOpInstance<T,
-                            T,
-                            T,
-                            ctl::NDHWC,
-                            ctl::KZYXC,
-                            ctl::NDHWK,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            ck::tensor_operation::element_wise::PassThrough,
-                            FillUniformDistribution<T>,
-                            FillUniformDistribution<T>>
-        conv_instance(params, true, FillUniformDistribution<T>{}, FillUniformDistribution<T>{});
-
-    auto reference_conv_fwd_fun =
-        std::bind(conv::run_reference_convolution_forward<3, T, T, T>, params, _1, _2, _3);
-    OpInstanceRunEngine<T, T, T> run_engine(conv_instance, reference_conv_fwd_fun);
-    run_engine.SetAtol(1e-3);
-    run_engine.SetRtol(1e-3);
-    EXPECT_TRUE(run_engine.Test(conv_ptrs));
-}
-
-TEST(Conv3DFwdNDHWC, InputOver2GB)
-{
-    using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-    using namespace ck::utils;
-    using T = float;
-
-    // >2GB Input
-    conv::ConvParams params;
-    params.num_dim_spatial_        = 3;
-    params.N_                      = 2;
-    params.K_                      = 16;
-    params.C_                      = 32;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3, 3, 3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{32, 1000, 1000};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1, 1, 1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1, 1, 1};
-    params.input_left_pads_        = std::vector<ck::index_t>{1, 1, 1};
-    params.input_right_pads_       = std::vector<ck::index_t>{1, 1, 1};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<3, T, T, T, T>(conv_ptrs);
-    auto arg = conv_ptrs.back()->MakeArgumentPointer(nullptr,
-                                                     nullptr,
-                                                     nullptr,
-                                                     params.N_,
-                                                     params.K_,
-                                                     params.C_,
-                                                     params.input_spatial_lengths_,
-                                                     params.filter_spatial_lengths_,
-                                                     params.GetOutputSpatialLengths(),
-                                                     params.conv_filter_strides_,
-                                                     params.conv_filter_dilations_,
-                                                     params.input_left_pads_,
-                                                     params.input_right_pads_,
-                                                     PassThrough{},
-                                                     PassThrough{},
-                                                     PassThrough{});
-    EXPECT_FALSE(conv_ptrs.back()->IsSupportedArgument(arg.get()));
-}
-
-TEST(Conv3DFwdNDHWC, FiltersOver2GB)
-{
-    using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-    using namespace ck::utils;
-    using T = float;
-
-    // >2GB Filters
-    conv::ConvParams params;
-    params.num_dim_spatial_        = 3;
-    params.N_                      = 2;
-    params.K_                      = 16;
-    params.C_                      = 32;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{4, 1000, 1000};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{16, 16, 16};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1, 1, 1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1, 1, 1};
-    params.input_left_pads_        = std::vector<ck::index_t>{1, 1, 1};
-    params.input_right_pads_       = std::vector<ck::index_t>{1, 1, 1};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<3, T, T, T, T>(conv_ptrs);
-    auto arg = conv_ptrs.back()->MakeArgumentPointer(nullptr,
-                                                     nullptr,
-                                                     nullptr,
-                                                     params.N_,
-                                                     params.K_,
-                                                     params.C_,
-                                                     params.input_spatial_lengths_,
-                                                     params.filter_spatial_lengths_,
-                                                     params.GetOutputSpatialLengths(),
-                                                     params.conv_filter_strides_,
-                                                     params.conv_filter_dilations_,
-                                                     params.input_left_pads_,
-                                                     params.input_right_pads_,
-                                                     PassThrough{},
-                                                     PassThrough{},
-                                                     PassThrough{});
-    EXPECT_FALSE(conv_ptrs.back()->IsSupportedArgument(arg.get()));
-}
-
-TEST(Conv3DFwdNDHWC, OutputOver2GB)
-{
-    using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-    using namespace ck::utils;
-    using T = float;
-
-    // >2GB Output
-    conv::ConvParams params;
-    params.num_dim_spatial_        = 3;
-    params.N_                      = 2;
-    params.K_                      = 16;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{1, 1, 1};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{1000, 1000, 30};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1, 1, 1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1, 1, 1};
-    params.input_left_pads_        = std::vector<ck::index_t>{2, 2, 2};
-    params.input_right_pads_       = std::vector<ck::index_t>{2, 2, 2};
-
-    std::vector<test::conv::DeviceConvFwdNoOpPtr> conv_ptrs;
-    test::conv::get_test_convolution_fwd_instance<3, T, T, T, T>(conv_ptrs);
-    auto arg = conv_ptrs.back()->MakeArgumentPointer(nullptr,
-                                                     nullptr,
-                                                     nullptr,
-                                                     params.N_,
-                                                     params.K_,
-                                                     params.C_,
-                                                     params.input_spatial_lengths_,
-                                                     params.filter_spatial_lengths_,
-                                                     params.GetOutputSpatialLengths(),
-                                                     params.conv_filter_strides_,
-                                                     params.conv_filter_dilations_,
-                                                     params.input_left_pads_,
-                                                     params.input_right_pads_,
-                                                     PassThrough{},
-                                                     PassThrough{},
-                                                     PassThrough{});
-    EXPECT_FALSE(conv_ptrs.back()->IsSupportedArgument(arg.get()));
-}
-
-TEST_F(Conv3dFwdNDHWCInstances, BF16_default) { EXPECT_TRUE(this->test_default<ck::bhalf_t>()); }
-TEST_F(Conv3dFwdNDHWCInstances, BF16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::bhalf_t>());
-}
-TEST_F(Conv3dFwdNDHWCInstances, BF16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::bhalf_t>());
-}
-
-TEST_F(Conv3dFwdNDHWCInstances, F16_default) { EXPECT_TRUE(this->test_default<ck::half_t>()); }
-TEST_F(Conv3dFwdNDHWCInstances, F16_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<ck::half_t>());
-}
-TEST_F(Conv3dFwdNDHWCInstances, F16_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<ck::half_t>());
-}
-
-TEST_F(Conv3dFwdNDHWCInstances, F32_default) { EXPECT_TRUE(this->test_default<float>()); }
-TEST_F(Conv3dFwdNDHWCInstances, F32_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<float>());
-}
-TEST_F(Conv3dFwdNDHWCInstances, F32_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<float>());
-}
-
-TEST_F(Conv3dFwdNDHWCInstances, I8_default) { EXPECT_TRUE(this->test_default<int8_t>()); }
-TEST_F(Conv3dFwdNDHWCInstances, I8_filter1x1_stride1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_stride1_pad0<int8_t>());
-}
-TEST_F(Conv3dFwdNDHWCInstances, I8_filter1x1_pad0)
-{
-    EXPECT_TRUE(this->test_filter1x1_pad0<int8_t>());
-}
diff --git a/test/convnd_fwd/conv_util.hpp b/test/convnd_fwd/conv_util.hpp
deleted file mode 100644
index c698bbd05c..0000000000
--- a/test/convnd_fwd/conv_util.hpp
+++ /dev/null
@@ -1,174 +0,0 @@
-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#pragma once
-
-#include <tuple>
-
-#include "ck/ck.hpp"
-#include "ck/utility/sequence.hpp"
-#include "ck/utility/data_type.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-
-using DeviceConvFwdNoOpPtr = DeviceConvFwdPtr<element_wise::PassThrough,
-                                              element_wise::PassThrough,
-                                              element_wise::PassThrough>;
-namespace instance {
-
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(std::vector<DeviceConvFwdNoOpPtr>&);
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck
-
-namespace test {
-namespace conv {
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
-using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
-using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
-
-using DeviceConvFwdNoOpPtr =
-    ck::tensor_operation::device::DeviceConvFwdPtr<InElementOp, WeiElementOp, OutElementOp>;
-
-static constexpr auto ConvFwdDefault =
-    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
-
-template <ck::index_t SpatialDims,
-          typename InDataType,
-          typename WeiDataType,
-          typename OutDataType,
-          typename AccDataType>
-using DeviceConvNDFwdInstance = ck::tensor_operation::device::
-    DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
-        // clang-format off
-        InDataType,         // 
-        WeiDataType,        //
-        OutDataType,        //
-        AccDataType,        // Accumulator data type.
-        InElementOp,        // Input Elementwise Operation
-        WeiElementOp,       // Weights Elementwise Operation
-        OutElementOp,       // Output Elementwise Operation
-        ConvFwdDefault,     // ConvForwardSpecialization
-        SpatialDims,        // SptialDims
-        256,                // BlockSize
-        128,                // MPerBlock
-        256,                // NPerBlock
-        4,                  // K0PerBlock
-        8,                  // K1
-        32,                 // MPerXdl
-        32,                 // NPerXdl
-        2,                  // MXdlPerWave
-        4,                  // NXdlPerWave
-        S<4, 64, 1>,        // ABlockTransferThreadClusterLengths_K0_M_K1
-        S<1, 0, 2>,         // ABlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // ABlockTransferSrcAccessOrder
-        2,                  // ABlockTransferSrcVectorDim
-        8,                  // ABlockTransferSrcScalarPerVector
-        8,                  // ABlockTransferDstScalarPerVector_K1
-        true,               // ABlockLdsAddExtraM
-        S<4, 64, 1>,        // BBlockTransferThreadClusterLengths_K0_N_K1
-        S<1, 0, 2>,         // BBlockTransferThreadClusterArrangeOrder
-        S<1, 0, 2>,         // BBlockTransferSrcAccessOrder
-        2,                  // BBlockTransferSrcVectorDim
-        8,                  // BBlockTransferSrcScalarPerVector
-        8,                  // BBlockTransferDstScalarPerVector_K1
-        true,               // BBlockLdsAddExtraN
-        7,                  // CThreadTransferSrcDstVectorDim
-        1>;                // CThreadTransferDstScalarPerVector
-// clang-format on
-
-template <ck::index_t NDim,
-          typename InDataType,
-          typename WeiDataType,
-          typename OutDataType,
-          typename AccDataType>
-void get_test_convolution_fwd_instance(std::vector<DeviceConvFwdNoOpPtr>& instances)
-{
-    using ConvInstanceT =
-        DeviceConvNDFwdInstance<NDim, InDataType, WeiDataType, OutDataType, AccDataType>;
-    instances.emplace_back(std::make_unique<ConvInstanceT>());
-}
-
-// TODO (aosewski)
-// Temporary solution to get all DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
-// instances. When switched over to DeviceConvNDFwdXdl for 2D remove ConvolutionNDFwdInstances
-// structures.
-template <typename InDataType, typename WeiDataType, typename OutDataType>
-struct ConvolutionNDFwdInstances;
-
-template <>
-struct ConvolutionNDFwdInstances<float, float, float>
-{
-    static std::vector<DeviceConvFwdNoOpPtr> Get(std::size_t num_dim_spatial)
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if(num_dim_spatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f32_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <>
-struct ConvolutionNDFwdInstances<ck::half_t, ck::half_t, ck::half_t>
-{
-    static std::vector<DeviceConvFwdNoOpPtr> Get(std::size_t num_dim_spatial)
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if(num_dim_spatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_f16_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <>
-struct ConvolutionNDFwdInstances<ck::bhalf_t, ck::bhalf_t, ck::bhalf_t>
-{
-    static std::vector<DeviceConvFwdNoOpPtr> Get(std::size_t num_dim_spatial)
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if(num_dim_spatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_bf16_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-template <>
-struct ConvolutionNDFwdInstances<int8_t, int8_t, int8_t>
-{
-    static std::vector<DeviceConvFwdNoOpPtr> Get(std::size_t num_dim_spatial)
-    {
-        std::vector<DeviceConvFwdNoOpPtr> conv_ptrs;
-        if(num_dim_spatial == 2)
-        {
-            ck::tensor_operation::device::instance::
-                add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(conv_ptrs);
-        }
-        return conv_ptrs;
-    }
-};
-
-} // namespace conv
-} // namespace test
diff --git a/test/convnd_fwd/convnd_fwd.cpp b/test/convnd_fwd/convnd_fwd.cpp
new file mode 100644
index 0000000000..5d4aae2951
--- /dev/null
+++ b/test/convnd_fwd/convnd_fwd.cpp
@@ -0,0 +1,241 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <initializer_list>
+#include <vector>
+#include <gtest/gtest.h>
+
+#include "profiler/include/profile_conv_fwd_impl.hpp"
+
+class TestConvndFwd : public ::testing::Test
+{
+    protected:
+    std::vector<ck::utils::conv::ConvParam> conv_params;
+};
+
+// 1d
+TEST_F(TestConvndFwd, Conv1dFwd)
+{
+    conv_params.clear();
+    conv_params.push_back({1, 1, 128, 128, 256, {1}, {14}, {2}, {1}, {0}, {0}});
+    conv_params.push_back({1, 1, 128, 128, 256, {3}, {28}, {1}, {1}, {1}, {1}});
+    conv_params.push_back({1, 1, 128, 128, 256, {1}, {3}, {1}, {1}, {0}, {0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_fwd_impl<1,
+                                                   ck::tensor_layout::convolution::NWC,
+                                                   ck::tensor_layout::convolution::KXC,
+                                                   ck::tensor_layout::convolution::NWK,
+                                                   float,
+                                                   float,
+                                                   float>(true,  // do_verification
+                                                          1,     // init_method
+                                                          false, // do_log
+                                                          false, // time_kernel
+                                                          param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_fwd_impl<1,
+                                                   ck::tensor_layout::convolution::NWC,
+                                                   ck::tensor_layout::convolution::KXC,
+                                                   ck::tensor_layout::convolution::NWK,
+                                                   ck::half_t,
+                                                   ck::half_t,
+                                                   ck::half_t>(true,  // do_verification
+                                                               1,     // init_method
+                                                               false, // do_log
+                                                               false, // time_kernel
+                                                               param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_fwd_impl<1,
+                                                   ck::tensor_layout::convolution::NWC,
+                                                   ck::tensor_layout::convolution::KXC,
+                                                   ck::tensor_layout::convolution::NWK,
+                                                   ck::bhalf_t,
+                                                   ck::bhalf_t,
+                                                   ck::bhalf_t>(true,  // do_verification
+                                                                1,     // init_method
+                                                                false, // do_log
+                                                                false, // time_kernel
+                                                                param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_conv_fwd_impl<1,
+                                                   ck::tensor_layout::convolution::NWC,
+                                                   ck::tensor_layout::convolution::KXC,
+                                                   ck::tensor_layout::convolution::NWK,
+                                                   int8_t,
+                                                   int8_t,
+                                                   int8_t>(true,  // do_verification
+                                                           1,     // init_method
+                                                           false, // do_log
+                                                           false, // time_kernel
+                                                           param);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 2d
+TEST_F(TestConvndFwd, Conv2dFwd)
+{
+    conv_params.clear();
+    conv_params.push_back({2, 1, 128, 128, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
+    conv_params.push_back({2, 1, 128, 128, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
+    conv_params.push_back({2, 1, 128, 128, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_fwd_impl<2,
+                                                   ck::tensor_layout::convolution::NHWC,
+                                                   ck::tensor_layout::convolution::KYXC,
+                                                   ck::tensor_layout::convolution::NHWK,
+                                                   float,
+                                                   float,
+                                                   float>(true,  // do_verification
+                                                          1,     // init_method
+                                                          false, // do_log
+                                                          false, // time_kernel
+                                                          param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_fwd_impl<2,
+                                                   ck::tensor_layout::convolution::NHWC,
+                                                   ck::tensor_layout::convolution::KYXC,
+                                                   ck::tensor_layout::convolution::NHWK,
+                                                   ck::half_t,
+                                                   ck::half_t,
+                                                   ck::half_t>(true,  // do_verification
+                                                               1,     // init_method
+                                                               false, // do_log
+                                                               false, // time_kernel
+                                                               param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_fwd_impl<2,
+                                                   ck::tensor_layout::convolution::NHWC,
+                                                   ck::tensor_layout::convolution::KYXC,
+                                                   ck::tensor_layout::convolution::NHWK,
+                                                   ck::bhalf_t,
+                                                   ck::bhalf_t,
+                                                   ck::bhalf_t>(true,  // do_verification
+                                                                1,     // init_method
+                                                                false, // do_log
+                                                                false, // time_kernel
+                                                                param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_conv_fwd_impl<2,
+                                                   ck::tensor_layout::convolution::NHWC,
+                                                   ck::tensor_layout::convolution::KYXC,
+                                                   ck::tensor_layout::convolution::NHWK,
+                                                   int8_t,
+                                                   int8_t,
+                                                   int8_t>(true,  // do_verification
+                                                           1,     // init_method
+                                                           false, // do_log
+                                                           false, // time_kernel
+                                                           param);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 3d
+TEST_F(TestConvndFwd, Conv3dFwd)
+{
+    conv_params.clear();
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {1, 1, 1}, {7, 7, 7}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {3, 3, 3}, {14, 14, 3}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}});
+    conv_params.push_back(
+        {3, 1, 128, 128, 256, {1, 1, 1}, {3, 3, 3}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_conv_fwd_impl<3,
+                                                   ck::tensor_layout::convolution::NDHWC,
+                                                   ck::tensor_layout::convolution::KZYXC,
+                                                   ck::tensor_layout::convolution::NDHWK,
+                                                   float,
+                                                   float,
+                                                   float>(true,  // do_verification
+                                                          1,     // init_method
+                                                          false, // do_log
+                                                          false, // time_kernel
+                                                          param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_conv_fwd_impl<3,
+                                                   ck::tensor_layout::convolution::NDHWC,
+                                                   ck::tensor_layout::convolution::KZYXC,
+                                                   ck::tensor_layout::convolution::NDHWK,
+                                                   ck::half_t,
+                                                   ck::half_t,
+                                                   ck::half_t>(true,  // do_verification
+                                                               1,     // init_method
+                                                               false, // do_log
+                                                               false, // time_kernel
+                                                               param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_conv_fwd_impl<3,
+                                                   ck::tensor_layout::convolution::NDHWC,
+                                                   ck::tensor_layout::convolution::KZYXC,
+                                                   ck::tensor_layout::convolution::NDHWK,
+                                                   ck::bhalf_t,
+                                                   ck::bhalf_t,
+                                                   ck::bhalf_t>(true,  // do_verification
+                                                                1,     // init_method
+                                                                false, // do_log
+                                                                false, // time_kernel
+                                                                param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_conv_fwd_impl<3,
+                                                   ck::tensor_layout::convolution::NDHWC,
+                                                   ck::tensor_layout::convolution::KZYXC,
+                                                   ck::tensor_layout::convolution::NDHWK,
+                                                   int8_t,
+                                                   int8_t,
+                                                   int8_t>(true,  // do_verification
+                                                           1,     // init_method
+                                                           false, // do_log
+                                                           false, // time_kernel
+                                                           param);
+
+        EXPECT_TRUE(pass);
+    }
+}
diff --git a/test/gemm/CMakeLists.txt b/test/gemm/CMakeLists.txt
index 83b3c1e2e3..8069dac157 100644
--- a/test/gemm/CMakeLists.txt
+++ b/test/gemm/CMakeLists.txt
@@ -1,15 +1,15 @@
 add_test_executable(test_gemm_fp32 gemm_fp32.cpp)
-target_link_libraries(test_gemm_fp32 PRIVATE host_tensor)
+target_link_libraries(test_gemm_fp32 PRIVATE utility)
 target_link_libraries(test_gemm_fp32 PRIVATE device_gemm_instance)
 
 add_test_executable(test_gemm_fp16 gemm_fp16.cpp)
-target_link_libraries(test_gemm_fp16 PRIVATE host_tensor)
+target_link_libraries(test_gemm_fp16 PRIVATE utility)
 target_link_libraries(test_gemm_fp16 PRIVATE device_gemm_instance)
 
 add_test_executable(test_gemm_bf16 gemm_bf16.cpp)
-target_link_libraries(test_gemm_bf16 PRIVATE host_tensor)
+target_link_libraries(test_gemm_bf16 PRIVATE utility)
 target_link_libraries(test_gemm_bf16 PRIVATE device_gemm_instance)
 
 add_test_executable(test_gemm_int8 gemm_int8.cpp)
-target_link_libraries(test_gemm_int8 PRIVATE host_tensor)
+target_link_libraries(test_gemm_int8 PRIVATE utility)
 target_link_libraries(test_gemm_int8 PRIVATE device_gemm_instance)
diff --git a/test/gemm/gemm_bf16.cpp b/test/gemm/gemm_bf16.cpp
index d7ecc892dc..6130ec9bc2 100644
--- a/test/gemm/gemm_bf16.cpp
+++ b/test/gemm/gemm_bf16.cpp
@@ -17,9 +17,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 #include "test/gemm/gemm_util.hpp"
diff --git a/test/gemm/gemm_fp16.cpp b/test/gemm/gemm_fp16.cpp
index ea9864abeb..05e696cad3 100644
--- a/test/gemm/gemm_fp16.cpp
+++ b/test/gemm/gemm_fp16.cpp
@@ -17,9 +17,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 #include "test/gemm/gemm_util.hpp"
diff --git a/test/gemm/gemm_fp32.cpp b/test/gemm/gemm_fp32.cpp
index b66addd712..3e141d7b30 100644
--- a/test/gemm/gemm_fp32.cpp
+++ b/test/gemm/gemm_fp32.cpp
@@ -17,9 +17,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 #include "test/gemm/gemm_util.hpp"
diff --git a/test/gemm/gemm_fp64.cpp b/test/gemm/gemm_fp64.cpp
index e0b9cab370..96dc459a3a 100644
--- a/test/gemm/gemm_fp64.cpp
+++ b/test/gemm/gemm_fp64.cpp
@@ -17,9 +17,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 #include "test/gemm/gemm_util.hpp"
diff --git a/test/gemm/gemm_int8.cpp b/test/gemm/gemm_int8.cpp
index 972f407975..c7d79782a1 100644
--- a/test/gemm/gemm_int8.cpp
+++ b/test/gemm/gemm_int8.cpp
@@ -17,9 +17,9 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 #include "test/gemm/gemm_util.hpp"
diff --git a/test/gemm/gemm_util.hpp b/test/gemm/gemm_util.hpp
index 4528c4aaef..2df605be10 100644
--- a/test/gemm/gemm_util.hpp
+++ b/test/gemm/gemm_util.hpp
@@ -6,9 +6,9 @@
 #include "ck/ck.hpp"
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
 namespace ck {
@@ -71,9 +71,9 @@ bool RunDeviceGEMM(DeviceGemmPtr_& gemmPtr,
                    BElementwiseOperation b_element_op,
                    CElementwiseOperation c_element_op)
 {
-    DeviceMem a_m_k_device_buf(sizeof(ADataType) * A.mDesc.GetElementSpace());
-    DeviceMem b_k_n_device_buf(sizeof(BDataType) * B.mDesc.GetElementSpace());
-    DeviceMem c_m_n_device_buf(sizeof(CDataType) * C.mDesc.GetElementSpace());
+    DeviceMem a_m_k_device_buf(sizeof(ADataType) * A.mDesc.GetElementSpaceSize());
+    DeviceMem b_k_n_device_buf(sizeof(BDataType) * B.mDesc.GetElementSpaceSize());
+    DeviceMem c_m_n_device_buf(sizeof(CDataType) * C.mDesc.GetElementSpaceSize());
 
     auto invoker_ptr = gemmPtr->MakeInvokerPointer();
     auto argument_ptr =
diff --git a/test/gemm_reduce/CMakeLists.txt b/test/gemm_reduce/CMakeLists.txt
index 74b787ac27..349f892c19 100644
--- a/test/gemm_reduce/CMakeLists.txt
+++ b/test/gemm_reduce/CMakeLists.txt
@@ -1,3 +1,3 @@
 add_test_executable(test_gemm_reduce_fp16 gemm_reduce_fp16.cpp)
-target_link_libraries(test_gemm_reduce_fp16 PRIVATE host_tensor)
+target_link_libraries(test_gemm_reduce_fp16 PRIVATE utility)
 target_link_libraries(test_gemm_reduce_fp16 PRIVATE device_gemm_reduce_instance)
diff --git a/test/gemm_split_k/CMakeLists.txt b/test/gemm_split_k/CMakeLists.txt
index ab1d016c9d..793091e53c 100644
--- a/test/gemm_split_k/CMakeLists.txt
+++ b/test/gemm_split_k/CMakeLists.txt
@@ -1,3 +1,3 @@
 add_test_executable(test_gemm_split_k gemm_split_k.cpp)
-target_link_libraries(test_gemm_split_k PRIVATE host_tensor)
+target_link_libraries(test_gemm_split_k PRIVATE utility)
 target_link_libraries(test_gemm_split_k PRIVATE device_gemm_splitk_instance)
diff --git a/test/gemm_split_k/gemm_split_k.cpp b/test/gemm_split_k/gemm_split_k.cpp
index fa06d76e36..e03cd4fa19 100644
--- a/test/gemm_split_k/gemm_split_k.cpp
+++ b/test/gemm_split_k/gemm_split_k.cpp
@@ -14,12 +14,12 @@
 #include "ck/library/tensor_operation_instance/gpu/gemm_splitk.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_gemm.hpp"
 
-#include "ck/library/host_tensor/host_gemm.hpp"
+#include "ck/library/utility/host_gemm.hpp"
 
 enum struct GemmMatrixLayout
 {
@@ -127,9 +127,9 @@ int test_gemm(const gemmArgs& args)
                        ck::tensor_operation::element_wise::PassThrough{},
                        ck::tensor_operation::element_wise::PassThrough{});
 
-    DeviceMem a_device_buf(sizeof(float) * a_m_k.mDesc.GetElementSpace());
-    DeviceMem b_device_buf(sizeof(float) * b_k_n.mDesc.GetElementSpace());
-    DeviceMem c_device_buf(sizeof(float) * c_m_n_device_result.mDesc.GetElementSpace());
+    DeviceMem a_device_buf(sizeof(float) * a_m_k.mDesc.GetElementSpaceSize());
+    DeviceMem b_device_buf(sizeof(float) * b_k_n.mDesc.GetElementSpaceSize());
+    DeviceMem c_device_buf(sizeof(float) * c_m_n_device_result.mDesc.GetElementSpaceSize());
 
     a_device_buf.ToDevice(a_m_k.mData.data());
     b_device_buf.ToDevice(b_k_n.mData.data());
diff --git a/test/grouped_convnd_fwd/CMakeLists.txt b/test/grouped_convnd_fwd/CMakeLists.txt
new file mode 100644
index 0000000000..38da884734
--- /dev/null
+++ b/test/grouped_convnd_fwd/CMakeLists.txt
@@ -0,0 +1,3 @@
+add_gtest_executable(test_grouped_convnd_fwd grouped_convnd_fwd.cpp)
+target_link_libraries(test_grouped_convnd_fwd PRIVATE utility device_grouped_conv1d_fwd_instance device_grouped_conv2d_fwd_instance device_grouped_conv3d_fwd_instance)
+
diff --git a/test/grouped_convnd_fwd/grouped_convnd_fwd.cpp b/test/grouped_convnd_fwd/grouped_convnd_fwd.cpp
new file mode 100644
index 0000000000..fbd6e9972f
--- /dev/null
+++ b/test/grouped_convnd_fwd/grouped_convnd_fwd.cpp
@@ -0,0 +1,270 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <initializer_list>
+#include <vector>
+#include <gtest/gtest.h>
+
+#include "profiler/include/profile_grouped_conv_fwd_impl.hpp"
+
+class TestGroupedConvNdFwd : public ::testing::Test
+{
+    protected:
+    std::vector<ck::utils::conv::ConvParam> conv_params;
+};
+
+// 1d GNWC/GKXC/GNWK
+TEST_F(TestGroupedConvNdFwd, GroupedConv1dFwdGNWC)
+{
+    conv_params.clear();
+    conv_params.push_back({1, 2, 128, 128, 256, {1}, {14}, {2}, {1}, {0}, {0}});
+    conv_params.push_back({1, 2, 128, 128, 256, {3}, {28}, {1}, {1}, {1}, {1}});
+    conv_params.push_back({1, 2, 128, 128, 256, {1}, {3}, {1}, {1}, {0}, {0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<1,
+                                                           ck::tensor_layout::convolution::GNWC,
+                                                           ck::tensor_layout::convolution::GKXC,
+                                                           ck::tensor_layout::convolution::GNWK,
+                                                           float,
+                                                           float,
+                                                           float>(true,  // do_verification
+                                                                  1,     // init_method
+                                                                  false, // do_log
+                                                                  false, // time_kernel
+                                                                  param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<1,
+                                                           ck::tensor_layout::convolution::GNWC,
+                                                           ck::tensor_layout::convolution::GKXC,
+                                                           ck::tensor_layout::convolution::GNWK,
+                                                           ck::half_t,
+                                                           ck::half_t,
+                                                           ck::half_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<1,
+                                                           ck::tensor_layout::convolution::GNWC,
+                                                           ck::tensor_layout::convolution::GKXC,
+                                                           ck::tensor_layout::convolution::GNWK,
+                                                           ck::bhalf_t,
+                                                           ck::bhalf_t,
+                                                           ck::bhalf_t>(true,  // do_verification
+                                                                        1,     // init_method
+                                                                        false, // do_log
+                                                                        false, // time_kernel
+                                                                        param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<1,
+                                                           ck::tensor_layout::convolution::GNWC,
+                                                           ck::tensor_layout::convolution::GKXC,
+                                                           ck::tensor_layout::convolution::GNWK,
+                                                           int8_t,
+                                                           int8_t,
+                                                           int8_t>(true,  // do_verification
+                                                                   1,     // init_method
+                                                                   false, // do_log
+                                                                   false, // time_kernel
+                                                                   param);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 2d GNHWC/GKYXC/GNHWK
+TEST_F(TestGroupedConvNdFwd, GroupedConv2dFwdGNHWC)
+{
+    conv_params.clear();
+    conv_params.push_back({2, 2, 128, 128, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
+    conv_params.push_back({2, 2, 128, 128, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
+    conv_params.push_back({2, 2, 128, 128, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<2,
+                                                           ck::tensor_layout::convolution::GNHWC,
+                                                           ck::tensor_layout::convolution::GKYXC,
+                                                           ck::tensor_layout::convolution::GNHWK,
+                                                           float,
+                                                           float,
+                                                           float>(true,  // do_verification
+                                                                  1,     // init_method
+                                                                  false, // do_log
+                                                                  false, // time_kernel
+                                                                  param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<2,
+                                                           ck::tensor_layout::convolution::GNHWC,
+                                                           ck::tensor_layout::convolution::GKYXC,
+                                                           ck::tensor_layout::convolution::GNHWK,
+                                                           ck::half_t,
+                                                           ck::half_t,
+                                                           ck::half_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<2,
+                                                           ck::tensor_layout::convolution::GNHWC,
+                                                           ck::tensor_layout::convolution::GKYXC,
+                                                           ck::tensor_layout::convolution::GNHWK,
+                                                           ck::bhalf_t,
+                                                           ck::bhalf_t,
+                                                           ck::bhalf_t>(true,  // do_verification
+                                                                        1,     // init_method
+                                                                        false, // do_log
+                                                                        false, // time_kernel
+                                                                        param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<2,
+                                                           ck::tensor_layout::convolution::GNHWC,
+                                                           ck::tensor_layout::convolution::GKYXC,
+                                                           ck::tensor_layout::convolution::GNHWK,
+                                                           int8_t,
+                                                           int8_t,
+                                                           int8_t>(true,  // do_verification
+                                                                   1,     // init_method
+                                                                   false, // do_log
+                                                                   false, // time_kernel
+                                                                   param);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 3d GNDHWC/GKZYXC/GNDHWK
+TEST_F(TestGroupedConvNdFwd, GroupedConv3dFwdGNDHWC)
+{
+    conv_params.clear();
+    conv_params.push_back(
+        {3, 2, 128, 128, 256, {1, 1, 1}, {7, 7, 7}, {2, 2, 2}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+    conv_params.push_back(
+        {3, 2, 128, 128, 256, {3, 3, 3}, {14, 14, 3}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}, {1, 1, 1}});
+    conv_params.push_back(
+        {3, 2, 128, 128, 256, {1, 1, 1}, {3, 3, 3}, {1, 1, 1}, {1, 1, 1}, {0, 0, 0}, {0, 0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp32
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<3,
+                                                           ck::tensor_layout::convolution::GNDHWC,
+                                                           ck::tensor_layout::convolution::GKZYXC,
+                                                           ck::tensor_layout::convolution::GNDHWK,
+                                                           float,
+                                                           float,
+                                                           float>(true,  // do_verification
+                                                                  1,     // init_method
+                                                                  false, // do_log
+                                                                  false, // time_kernel
+                                                                  param);
+
+        EXPECT_TRUE(pass);
+
+        // fp16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<3,
+                                                           ck::tensor_layout::convolution::GNDHWC,
+                                                           ck::tensor_layout::convolution::GKZYXC,
+                                                           ck::tensor_layout::convolution::GNDHWK,
+                                                           ck::half_t,
+                                                           ck::half_t,
+                                                           ck::half_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param);
+
+        EXPECT_TRUE(pass);
+
+        // bf16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<3,
+                                                           ck::tensor_layout::convolution::GNDHWC,
+                                                           ck::tensor_layout::convolution::GKZYXC,
+                                                           ck::tensor_layout::convolution::GNDHWK,
+                                                           ck::bhalf_t,
+                                                           ck::bhalf_t,
+                                                           ck::bhalf_t>(true,  // do_verification
+                                                                        1,     // init_method
+                                                                        false, // do_log
+                                                                        false, // time_kernel
+                                                                        param);
+
+        EXPECT_TRUE(pass);
+
+        // int8
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<3,
+                                                           ck::tensor_layout::convolution::GNDHWC,
+                                                           ck::tensor_layout::convolution::GKZYXC,
+                                                           ck::tensor_layout::convolution::GNDHWK,
+                                                           int8_t,
+                                                           int8_t,
+                                                           int8_t>(true,  // do_verification
+                                                                   1,     // init_method
+                                                                   false, // do_log
+                                                                   false, // time_kernel
+                                                                   param);
+
+        EXPECT_TRUE(pass);
+    }
+}
+
+// 2d NHWGC/KYXGC/NHWGK
+TEST_F(TestGroupedConvNdFwd, GroupedConv2dFwdNHWGC)
+{
+    conv_params.clear();
+    conv_params.push_back({2, 2, 128, 128, 256, {1, 1}, {7, 7}, {2, 2}, {1, 1}, {0, 0}, {0, 0}});
+    conv_params.push_back({2, 2, 128, 128, 256, {3, 3}, {14, 14}, {1, 1}, {1, 1}, {1, 1}, {1, 1}});
+    conv_params.push_back({2, 2, 128, 128, 256, {1, 1}, {3, 3}, {1, 1}, {1, 1}, {0, 0}, {0, 0}});
+
+    for(auto& param : conv_params)
+    {
+        bool pass;
+
+        // fp16
+        pass = ck::profiler::profile_grouped_conv_fwd_impl<2,
+                                                           ck::tensor_layout::convolution::NHWGC,
+                                                           ck::tensor_layout::convolution::KYXGC,
+                                                           ck::tensor_layout::convolution::NHWGK,
+                                                           ck::half_t,
+                                                           ck::half_t,
+                                                           ck::half_t>(true,  // do_verification
+                                                                       1,     // init_method
+                                                                       false, // do_log
+                                                                       false, // time_kernel
+                                                                       param);
+
+        EXPECT_TRUE(pass);
+    }
+}
diff --git a/test/grouped_gemm/CMakeLists.txt b/test/grouped_gemm/CMakeLists.txt
index f04ee77062..31a78733d3 100644
--- a/test/grouped_gemm/CMakeLists.txt
+++ b/test/grouped_gemm/CMakeLists.txt
@@ -1,3 +1,3 @@
 add_test_executable(test_grouped_gemm_fp16 grouped_gemm_fp16.cpp)
-target_link_libraries(test_grouped_gemm_fp16 PRIVATE host_tensor)
+target_link_libraries(test_grouped_gemm_fp16 PRIVATE utility)
 target_link_libraries(test_grouped_gemm_fp16 PRIVATE device_grouped_gemm_instance)
diff --git a/test/layernorm/CMakeLists.txt b/test/layernorm/CMakeLists.txt
index 5021edf653..ad681583d1 100644
--- a/test/layernorm/CMakeLists.txt
+++ b/test/layernorm/CMakeLists.txt
@@ -2,7 +2,9 @@ add_custom_target(test_layernorm)
 
 add_gtest_executable(test_layernorm_fp32 test_layernorm_fp32.cpp)
 add_gtest_executable(test_layernorm_fp16 test_layernorm_fp16.cpp)
-target_link_libraries(test_layernorm_fp32 PRIVATE host_tensor)
-target_link_libraries(test_layernorm_fp16 PRIVATE host_tensor)
+
+target_link_libraries(test_layernorm_fp32 PRIVATE utility)
+target_link_libraries(test_layernorm_fp16 PRIVATE utility)
+
 add_dependencies(test_layernorm test_layernorm_fp32)
-add_dependencies(test_layernorm test_layernorm_fp16)
\ No newline at end of file
+add_dependencies(test_layernorm test_layernorm_fp16)
diff --git a/test/layernorm/test_layernorm_util.hpp b/test/layernorm/test_layernorm_util.hpp
index 167c2ec9ca..37374839c5 100644
--- a/test/layernorm/test_layernorm_util.hpp
+++ b/test/layernorm/test_layernorm_util.hpp
@@ -12,8 +12,8 @@
 #include "ck/tensor_operation/gpu/device/device_layernorm.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/device_memory.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp"
 
 namespace ck {
@@ -102,10 +102,10 @@ class TestLayernorm : public ::testing::Test
         gamma.GenerateTensorValue(GeneratorTensor_3<GammaDataType>{0.0, 1.0});
         beta.GenerateTensorValue(GeneratorTensor_3<BetaDataType>{0.0, 1.0});
 
-        DeviceMem x_dev(sizeof(XDataType) * x.mDesc.GetElementSpace());
-        DeviceMem gamma_dev(sizeof(GammaDataType) * gamma.mDesc.GetElementSpace());
-        DeviceMem beta_dev(sizeof(BetaDataType) * beta.mDesc.GetElementSpace());
-        DeviceMem y_dev(sizeof(YDataType) * y.mDesc.GetElementSpace());
+        DeviceMem x_dev(sizeof(XDataType) * x.mDesc.GetElementSpaceSize());
+        DeviceMem gamma_dev(sizeof(GammaDataType) * gamma.mDesc.GetElementSpaceSize());
+        DeviceMem beta_dev(sizeof(BetaDataType) * beta.mDesc.GetElementSpaceSize());
+        DeviceMem y_dev(sizeof(YDataType) * y.mDesc.GetElementSpaceSize());
 
         x_dev.ToDevice(x.mData.data());
         gamma_dev.ToDevice(gamma.mData.data());
diff --git a/test/magic_number_division/CMakeLists.txt b/test/magic_number_division/CMakeLists.txt
index c7d3f45cd4..e7fc6ee5df 100644
--- a/test/magic_number_division/CMakeLists.txt
+++ b/test/magic_number_division/CMakeLists.txt
@@ -1,2 +1,2 @@
 add_test_executable(test_magic_number_division magic_number_division.cpp)
-target_link_libraries(test_magic_number_division PRIVATE host_tensor)
+target_link_libraries(test_magic_number_division PRIVATE utility)
diff --git a/test/magic_number_division/magic_number_division.cpp b/test/magic_number_division/magic_number_division.cpp
index 7981141608..680fddf193 100644
--- a/test/magic_number_division/magic_number_division.cpp
+++ b/test/magic_number_division/magic_number_division.cpp
@@ -9,9 +9,9 @@
 #include "ck/ck.hpp"
 #include "ck/utility/magic_division.hpp"
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/host_tensor_generator.hpp"
+#include "ck/library/utility/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/host_tensor_generator.hpp"
 
 __global__ void gpu_magic_number_division(uint32_t magic_multiplier,
                                           uint32_t magic_shift,
diff --git a/test/reduce/CMakeLists.txt b/test/reduce/CMakeLists.txt
index 4e11b049a8..fb436165ea 100644
--- a/test/reduce/CMakeLists.txt
+++ b/test/reduce/CMakeLists.txt
@@ -1,7 +1,7 @@
 add_test_executable(test_reduce_no_index reduce_no_index.cpp)
 add_test_executable(test_reduce_with_index reduce_with_index.cpp)
-target_link_libraries(test_reduce_no_index PRIVATE host_tensor)
+target_link_libraries(test_reduce_no_index PRIVATE utility)
 target_link_libraries(test_reduce_no_index PRIVATE device_reduce_instance)
-target_link_libraries(test_reduce_with_index PRIVATE host_tensor)
+target_link_libraries(test_reduce_with_index PRIVATE utility)
 target_link_libraries(test_reduce_with_index PRIVATE device_reduce_instance)
 
diff --git a/test/reduce/reduce_no_index.cpp b/test/reduce/reduce_no_index.cpp
index 843a6b110a..475ebfd080 100644
--- a/test/reduce/reduce_no_index.cpp
+++ b/test/reduce/reduce_no_index.cpp
@@ -3,7 +3,7 @@
 
 #include <getopt.h>
 
-#include "ck/library/host_tensor/host_common_util.hpp"
+#include "ck/library/utility/host_common_util.hpp"
 #include "profiler/include/profile_reduce_impl.hpp"
 
 using namespace ck;
diff --git a/test/reduce/reduce_with_index.cpp b/test/reduce/reduce_with_index.cpp
index 64f16b8085..c319dca69c 100644
--- a/test/reduce/reduce_with_index.cpp
+++ b/test/reduce/reduce_with_index.cpp
@@ -3,7 +3,7 @@
 
 #include <getopt.h>
 
-#include "ck/library/host_tensor/host_common_util.hpp"
+#include "ck/library/utility/host_common_util.hpp"
 #include "profiler/include/profile_reduce_impl.hpp"
 
 using namespace ck;
diff --git a/test/reference_conv_fwd/CMakeLists.txt b/test/reference_conv_fwd/CMakeLists.txt
index 04b720b169..b40b9a1ed0 100644
--- a/test/reference_conv_fwd/CMakeLists.txt
+++ b/test/reference_conv_fwd/CMakeLists.txt
@@ -1,2 +1,2 @@
 add_gtest_executable(test_reference_conv_fwd reference_conv_fwd.cpp)
-target_link_libraries(test_reference_conv_fwd PRIVATE host_tensor conv_util)
+target_link_libraries(test_reference_conv_fwd PRIVATE utility)
diff --git a/test/reference_conv_fwd/reference_conv_fwd.cpp b/test/reference_conv_fwd/reference_conv_fwd.cpp
index 2b5591675f..82a8dbbd06 100644
--- a/test/reference_conv_fwd/reference_conv_fwd.cpp
+++ b/test/reference_conv_fwd/reference_conv_fwd.cpp
@@ -13,74 +13,64 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/utility/conv_util.hpp"
 #include "ck/library/utility/fill.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/convolution_parameter.hpp"
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp"
 
 namespace {
+
 using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
 using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
 using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
 
-template <ck::index_t NDim,
+template <ck::index_t NDimSpatial,
           typename InDataType    = float,
           typename WeiDataType   = float,
           typename OutDataType   = float,
-          typename InLayout      = ck::tensor_layout::convolution::NHWC,
-          typename WeiLayout     = ck::tensor_layout::convolution::KYXC,
-          typename OutLayout     = ck::tensor_layout::convolution::NHWK,
+          typename InLayout      = ck::tensor_layout::convolution::GNHWC,
+          typename WeiLayout     = ck::tensor_layout::convolution::GKYXC,
+          typename OutLayout     = ck::tensor_layout::convolution::GNHWK,
           typename FillInputOp   = ck::utils::FillMonotonicSeq<InDataType>,
           typename FillWeightsOp = ck::utils::FillConstant<WeiDataType>>
 Tensor<OutDataType>
-run_reference_convolution_forward(const ck::utils::conv::ConvParams& params,
+run_reference_convolution_forward(const ck::utils::conv::ConvParam& conv_param,
                                   const FillInputOp& fill_input_op     = FillInputOp{},
                                   const FillWeightsOp& fill_weights_op = FillWeightsOp{0.5f})
 {
-    std::vector<std::size_t> input_dims{static_cast<std::size_t>(params.N_),
-                                        static_cast<std::size_t>(params.C_)};
-    input_dims.insert(std::end(input_dims),
-                      std::begin(params.input_spatial_lengths_),
-                      std::end(params.input_spatial_lengths_));
+    const auto in_g_n_c_wis_desc =
+        ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);
 
-    std::vector<std::size_t> filter_dims{static_cast<std::size_t>(params.K_),
-                                         static_cast<std::size_t>(params.C_)};
-    filter_dims.insert(std::end(filter_dims),
-                       std::begin(params.filter_spatial_lengths_),
-                       std::end(params.filter_spatial_lengths_));
+    const auto wei_g_k_c_xs_desc =
+        ck::utils::conv::make_weight_host_tensor_descriptor_g_k_c_xs_packed<WeiLayout>(conv_param);
 
-    const std::vector<ck::index_t>& output_spatial_lengths = params.GetOutputSpatialLengths();
-    std::vector<std::size_t> output_dims{static_cast<std::size_t>(params.N_),
-                                         static_cast<std::size_t>(params.K_)};
-    output_dims.insert(std::end(output_dims),
-                       std::begin(output_spatial_lengths),
-                       std::end(output_spatial_lengths));
+    const auto out_g_n_k_wos_desc =
+        ck::utils::conv::make_output_host_tensor_descriptor_g_n_k_wos_packed<OutLayout>(conv_param);
 
-    Tensor<InDataType> input(ck::utils::conv::get_host_tensor_descriptor(input_dims, InLayout{}));
-    Tensor<WeiDataType> weights(
-        ck::utils::conv::get_host_tensor_descriptor(filter_dims, WeiLayout{}));
-    Tensor<OutDataType> host_output(
-        ck::utils::conv::get_host_tensor_descriptor(output_dims, OutLayout{}));
+    Tensor<InDataType> input(in_g_n_c_wis_desc);
+    Tensor<WeiDataType> weights(wei_g_k_c_xs_desc);
+    Tensor<OutDataType> host_output(out_g_n_k_wos_desc);
 
     fill_input_op(input.begin(), input.end());
     fill_weights_op(weights.begin(), weights.end());
     std::fill(host_output.begin(), host_output.end(), OutDataType(0.f));
 
-    auto ref_conv     = ck::tensor_operation::host::ReferenceConvFwd<InDataType,
+    auto ref_conv     = ck::tensor_operation::host::ReferenceConvFwd<NDimSpatial,
+                                                                 InDataType,
                                                                  WeiDataType,
                                                                  OutDataType,
                                                                  InElementOp,
                                                                  WeiElementOp,
-                                                                 OutElementOp,
-                                                                 NDim>();
+                                                                 OutElementOp>();
     auto ref_invoker  = ref_conv.MakeInvoker();
     auto ref_argument = ref_conv.MakeArgument(input,
                                               weights,
                                               host_output,
-                                              params.conv_filter_strides_,
-                                              params.conv_filter_dilations_,
-                                              params.input_left_pads_,
-                                              params.input_right_pads_,
+                                              conv_param.conv_filter_strides_,
+                                              conv_param.conv_filter_dilations_,
+                                              conv_param.input_left_pads_,
+                                              conv_param.input_right_pads_,
                                               InElementOp{},
                                               WeiElementOp{},
                                               OutElementOp{});
@@ -91,21 +81,29 @@ run_reference_convolution_forward(const ck::utils::conv::ConvParams& params,
 
 } // anonymous namespace
 
-TEST(ReferenceConvolutionFWD, Conv2DNHWC)
+// Eeference convolution assume dimensions of tensor descriptors are in GNCDHW/GKCZYX/GNKDHW order,
+// regardless of physical tensor layouts in  memory.
+// Some tests below assume dimensions of tensor descriptors can be in other order, and therefore
+// are disabled
+// TODO: add more tests, which comply with assumption about dimension order of reference convolution
+// and add tests for more physical layout
+#if 0
+TEST(ReferenceConvolutionFWD, Conv2DGNHWC)
 {
-    ck::utils::conv::ConvParams params;
-    params.N_                      = 1;
-    params.K_                      = 1;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3, 3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{6, 6};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1, 1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1, 1};
-    params.input_left_pads_        = std::vector<ck::index_t>{0, 0};
-    params.input_right_pads_       = std::vector<ck::index_t>{0, 0};
+    ck::utils::conv::ConvParam conv_param(2,
+                                          1,
+                                          1,
+                                          1,
+                                          2,
+                                          std::vector<ck::index_t>{3, 3},
+                                          std::vector<ck::index_t>{6, 6},
+                                          std::vector<ck::index_t>{1, 1},
+                                          std::vector<ck::index_t>{1, 1},
+                                          std::vector<ck::index_t>{0, 0},
+                                          std::vector<ck::index_t>{0, 0});
 
-    auto out_tensor = run_reference_convolution_forward<2>(params);
-    std::vector<std::size_t> ref_dims{1, 1, 4, 4};
+    auto out_tensor = run_reference_convolution_forward<2>(conv_param);
+    std::vector<std::size_t> ref_dims{1, 1, 4, 4, 1};
     std::vector<float> ref_data{130.5,
                                 148.5,
                                 166.5,
@@ -127,21 +125,22 @@ TEST(ReferenceConvolutionFWD, Conv2DNHWC)
     EXPECT_TRUE(ck::utils::check_err(out_tensor.mData, ref_data, "Error: incorrect results!"));
 }
 
-TEST(ReferenceConvolutionFWD, Conv2DNHWCStridesDilationsPadding)
+TEST(ReferenceConvolutionFWD, Conv2DGNHWCStridesDilationsPadding)
 {
-    ck::utils::conv::ConvParams params;
-    params.N_                      = 1;
-    params.K_                      = 2;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3, 3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{12, 12};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{2, 2};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{2, 2};
-    params.input_left_pads_        = std::vector<ck::index_t>{1, 1};
-    params.input_right_pads_       = std::vector<ck::index_t>{1, 1};
+    ck::utils::conv::ConvParam conv_param(2,
+                                          1,
+                                          1,
+                                          2,
+                                          2,
+                                          std::vector<ck::index_t>{3, 3},
+                                          std::vector<ck::index_t>{12, 12},
+                                          std::vector<ck::index_t>{2, 2},
+                                          std::vector<ck::index_t>{2, 2},
+                                          std::vector<ck::index_t>{1, 1},
+                                          std::vector<ck::index_t>{1, 1});
 
-    auto out_tensor                   = run_reference_convolution_forward<2>(params);
-    std::vector<std::size_t> ref_dims = std::vector<std::size_t>{1, 2, 5, 5};
+    auto out_tensor                   = run_reference_convolution_forward<2>(conv_param);
+    std::vector<std::size_t> ref_dims = std::vector<std::size_t>{1, 5, 5, 2};
     std::vector<float> ref_data{
         210.,  210.,  327.,   327.,   351.,   351.,   375.,   375.,   399.,   399.,
         459.,  459.,  706.5,  706.5,  742.5,  742.5,  778.5,  778.5,  814.5,  814.5,
@@ -153,88 +152,88 @@ TEST(ReferenceConvolutionFWD, Conv2DNHWCStridesDilationsPadding)
     EXPECT_TRUE(ck::utils::check_err(out_tensor.mData, ref_data, "Error: incorrect results!"));
 }
 
-TEST(ReferenceConvolutionFWD, Conv1DNWC)
+TEST(ReferenceConvolutionFWD, Conv1DGNWC)
 {
-    ck::utils::conv::ConvParams params;
-    params.num_dim_spatial_        = 1;
-    params.N_                      = 1;
-    params.K_                      = 1;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{6};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1};
-    params.input_left_pads_        = std::vector<ck::index_t>{0};
-    params.input_right_pads_       = std::vector<ck::index_t>{0};
+    ck::utils::conv::ConvParam conv_param(1,
+                                          1,
+                                          1,
+                                          1,
+                                          2,
+                                          std::vector<ck::index_t>{3},
+                                          std::vector<ck::index_t>{6},
+                                          std::vector<ck::index_t>{1},
+                                          std::vector<ck::index_t>{1},
+                                          std::vector<ck::index_t>{0},
+                                          std::vector<ck::index_t>{0});
 
     auto out_tensor =
         run_reference_convolution_forward<1,
                                           float,
                                           float,
                                           float,
-                                          ck::tensor_layout::convolution::NWC,
-                                          ck::tensor_layout::convolution::KXC,
-                                          ck::tensor_layout::convolution::NWK>(params);
-    std::vector<std::size_t> ref_dims{1, 1, 4};
+                                          ck::tensor_layout::convolution::GNWC,
+                                          ck::tensor_layout::convolution::GKXC,
+                                          ck::tensor_layout::convolution::GNWK>(conv_param);
+    std::vector<std::size_t> ref_dims{1, 1, 4, 1};
     std::vector<float> ref_data{7.5, 13.5, 19.5, 25.5};
     EXPECT_TRUE(ck::utils::check_err(
         out_tensor.mDesc.GetLengths(), ref_dims, "Error: wrong output tensor dimensions!"));
     EXPECT_TRUE(ck::utils::check_err(out_tensor.mData, ref_data, "Error: incorrect results!"));
 }
 
-TEST(ReferenceConvolutionFWD, Conv1DNWCStridesDilationsPadding)
+TEST(ReferenceConvolutionFWD, Conv1DGNWCStridesDilationsPadding)
 {
-    ck::utils::conv::ConvParams params;
-    params.num_dim_spatial_        = 1;
-    params.N_                      = 1;
-    params.K_                      = 2;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{12};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{2};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{2};
-    params.input_left_pads_        = std::vector<ck::index_t>{1};
-    params.input_right_pads_       = std::vector<ck::index_t>{1};
+    ck::utils::conv::ConvParam conv_param(1,
+                                          1,
+                                          1,
+                                          2,
+                                          2,
+                                          std::vector<ck::index_t>{3},
+                                          std::vector<ck::index_t>{12},
+                                          std::vector<ck::index_t>{2},
+                                          std::vector<ck::index_t>{2},
+                                          std::vector<ck::index_t>{1},
+                                          std::vector<ck::index_t>{1});
 
     auto out_tensor =
         run_reference_convolution_forward<1,
                                           float,
                                           float,
                                           float,
-                                          ck::tensor_layout::convolution::NWC,
-                                          ck::tensor_layout::convolution::KXC,
-                                          ck::tensor_layout::convolution::NWK>(params);
-    std::vector<std::size_t> ref_dims{1, 2, 5};
+                                          ck::tensor_layout::convolution::GNWC,
+                                          ck::tensor_layout::convolution::GKXC,
+                                          ck::tensor_layout::convolution::GNWK>(conv_param);
+    std::vector<std::size_t> ref_dims{1, 1, 5, 2};
     std::vector<float> ref_data{9., 9., 19.5, 19.5, 31.5, 31.5, 43.5, 43.5, 55.5, 55.5};
     EXPECT_TRUE(ck::utils::check_err(
         out_tensor.mDesc.GetLengths(), ref_dims, "Error: wrong output tensor dimensions!"));
     EXPECT_TRUE(ck::utils::check_err(out_tensor.mData, ref_data, "Error: incorrect results!"));
 }
 
-TEST(ReferenceConvolutionFWD, Conv1DNWCSameOutputSize)
+TEST(ReferenceConvolutionFWD, Conv1DGNWCSameOutputSize)
 {
-    ck::utils::conv::ConvParams params;
-    params.num_dim_spatial_        = 1;
-    params.N_                      = 2;
-    params.K_                      = 16;
-    params.C_                      = 4;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{16};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1};
-    params.input_left_pads_        = std::vector<ck::index_t>{1};
-    params.input_right_pads_       = std::vector<ck::index_t>{1};
+    ck::utils::conv::ConvParam conv_param(1,
+                                          1,
+                                          2,
+                                          16,
+                                          4,
+                                          std::vector<ck::index_t>{3},
+                                          std::vector<ck::index_t>{16},
+                                          std::vector<ck::index_t>{1},
+                                          std::vector<ck::index_t>{1},
+                                          std::vector<ck::index_t>{1},
+                                          std::vector<ck::index_t>{1});
 
     auto out_tensor2 = run_reference_convolution_forward<1,
                                                          float,
                                                          float,
                                                          float,
-                                                         ck::tensor_layout::convolution::NWC,
-                                                         ck::tensor_layout::convolution::KXC,
-                                                         ck::tensor_layout::convolution::NWK>(
-        params, ck::utils::FillMonotonicSeq<float>{0.f, 0.1f});
+                                                         ck::tensor_layout::convolution::GNWC,
+                                                         ck::tensor_layout::convolution::GKXC,
+                                                         ck::tensor_layout::convolution::GNWK>(
+        conv_param, ck::utils::FillMonotonicSeq<float>{0.f, 0.1f});
 
-    std::vector<std::size_t> ref_dims{2, 16, 16};
+    std::vector<std::size_t> ref_dims{1, 2, 16, 16};
     std::vector<float> ref_data{
         1.4,       1.4,       1.4,       1.4,       1.4,       1.4,       1.4,       1.4,
         1.4,       1.4,       1.4,       1.4,       1.4,       1.4,       1.4,       1.4,
@@ -304,30 +303,31 @@ TEST(ReferenceConvolutionFWD, Conv1DNWCSameOutputSize)
         out_tensor2.mDesc.GetLengths(), ref_dims, "Error: wrong output tensor dimensions!"));
     EXPECT_TRUE(ck::utils::check_err(out_tensor2.mData, ref_data, "Error: incorrect results!"));
 }
+#endif
 
-TEST(ReferenceConvolutionFWD, Conv3DNCDHW)
+TEST(ReferenceConvolutionFWD, Conv3DGNCDHW)
 {
-    ck::utils::conv::ConvParams params;
-    params.num_dim_spatial_        = 3;
-    params.N_                      = 1;
-    params.K_                      = 1;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3, 3, 3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{6, 6, 6};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{1, 1, 1};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1, 1, 1};
-    params.input_left_pads_        = std::vector<ck::index_t>{0, 0, 0};
-    params.input_right_pads_       = std::vector<ck::index_t>{0, 0, 0};
+    ck::utils::conv::ConvParam conv_param(3,
+                                          1,
+                                          1,
+                                          1,
+                                          2,
+                                          std::vector<ck::index_t>{3, 3, 3},
+                                          std::vector<ck::index_t>{6, 6, 6},
+                                          std::vector<ck::index_t>{1, 1, 1},
+                                          std::vector<ck::index_t>{1, 1, 1},
+                                          std::vector<ck::index_t>{0, 0, 0},
+                                          std::vector<ck::index_t>{0, 0, 0});
 
     auto out_tensor = run_reference_convolution_forward<3,
                                                         float,
                                                         float,
                                                         float,
-                                                        ck::tensor_layout::convolution::NCDHW,
-                                                        ck::tensor_layout::convolution::KCZYX,
-                                                        ck::tensor_layout::convolution::NKDHW>(
-        params, ck::utils::FillMonotonicSeq<float>{0.f, 0.1f});
-    std::vector<std::size_t> ref_dims{1, 1, 4, 4, 4};
+                                                        ck::tensor_layout::convolution::GNCDHW,
+                                                        ck::tensor_layout::convolution::GKCZYX,
+                                                        ck::tensor_layout::convolution::GNKDHW>(
+        conv_param, ck::utils::FillMonotonicSeq<float>{0.f, 0.1f});
+    std::vector<std::size_t> ref_dims{1, 1, 1, 4, 4, 4};
     std::vector<float> ref_data{
         407.7,     410.40002, 413.09998, 415.80002, 423.90002, 426.6,     429.30002, 432.,
         440.1,     442.80002, 445.5,     448.2,     456.30002, 459.,      461.7,     464.40002,
@@ -344,29 +344,29 @@ TEST(ReferenceConvolutionFWD, Conv3DNCDHW)
         ck::utils::check_err(out_tensor.mData, ref_data, "Error [case 1]: incorrect results!"));
 }
 
-TEST(ReferenceConvolutionFWD, Conv3DNCDHWStridesDilations)
+TEST(ReferenceConvolutionFWD, Conv3DGNCDHWStridesDilations)
 {
-    ck::utils::conv::ConvParams params;
-    params.num_dim_spatial_        = 3;
-    params.N_                      = 1;
-    params.K_                      = 2;
-    params.C_                      = 2;
-    params.filter_spatial_lengths_ = std::vector<ck::index_t>{3, 3, 3};
-    params.input_spatial_lengths_  = std::vector<ck::index_t>{12, 12, 12};
-    params.conv_filter_strides_    = std::vector<ck::index_t>{3, 3, 3};
-    params.conv_filter_dilations_  = std::vector<ck::index_t>{1, 1, 1};
-    params.input_left_pads_        = std::vector<ck::index_t>{0, 0, 0};
-    params.input_right_pads_       = std::vector<ck::index_t>{0, 0, 0};
+    ck::utils::conv::ConvParam conv_param(3,
+                                          1,
+                                          1,
+                                          2,
+                                          2,
+                                          std::vector<ck::index_t>{3, 3, 3},
+                                          std::vector<ck::index_t>{12, 12, 12},
+                                          std::vector<ck::index_t>{3, 3, 3},
+                                          std::vector<ck::index_t>{1, 1, 1},
+                                          std::vector<ck::index_t>{0, 0, 0},
+                                          std::vector<ck::index_t>{0, 0, 0});
 
     auto out_tensor = run_reference_convolution_forward<3,
                                                         float,
                                                         float,
                                                         float,
-                                                        ck::tensor_layout::convolution::NCDHW,
-                                                        ck::tensor_layout::convolution::KCZYX,
-                                                        ck::tensor_layout::convolution::NKDHW>(
-        params, ck::utils::FillMonotonicSeq<float>{0.f, 0.1f});
-    std::vector<std::size_t> ref_dims{1, 2, 4, 4, 4};
+                                                        ck::tensor_layout::convolution::GNCDHW,
+                                                        ck::tensor_layout::convolution::GKCZYX,
+                                                        ck::tensor_layout::convolution::GNKDHW>(
+        conv_param, ck::utils::FillMonotonicSeq<float>{0.f, 0.1f});
+    std::vector<std::size_t> ref_dims{1, 1, 2, 4, 4, 4};
     std::vector<float> ref_data{
         2756.7002, 2764.7998, 2772.9001, 2781.,     2853.9001, 2862.,     2870.1,    2878.2002,
         2951.1,    2959.2002, 2967.2998, 2975.4001, 3048.2998, 3056.4001, 3064.5,    3072.6,
diff --git a/test/softmax/CMakeLists.txt b/test/softmax/CMakeLists.txt
index da80e372ea..a7013eece1 100644
--- a/test/softmax/CMakeLists.txt
+++ b/test/softmax/CMakeLists.txt
@@ -3,9 +3,9 @@ add_custom_target(test_softmax)
 add_gtest_executable(test_softmax_fp32 test_softmax_fp32.cpp)
 add_gtest_executable(test_softmax_fp16 test_softmax_fp16.cpp)
 add_gtest_executable(test_softmax_int8 test_softmax_int8.cpp)
-target_link_libraries(test_softmax_fp32 PRIVATE host_tensor)
-target_link_libraries(test_softmax_fp16 PRIVATE host_tensor)
-target_link_libraries(test_softmax_int8 PRIVATE host_tensor)
+target_link_libraries(test_softmax_fp32 PRIVATE utility)
+target_link_libraries(test_softmax_fp16 PRIVATE utility)
+target_link_libraries(test_softmax_int8 PRIVATE utility)
 add_dependencies(test_softmax test_softmax_fp32)
 add_dependencies(test_softmax test_softmax_fp16)
-add_dependencies(test_softmax test_softmax_int8)
\ No newline at end of file
+add_dependencies(test_softmax test_softmax_int8)
diff --git a/test/softmax/test_softmax_util.hpp b/test/softmax/test_softmax_util.hpp
index 2ca3b47abc..97a641e8e9 100644
--- a/test/softmax/test_softmax_util.hpp
+++ b/test/softmax/test_softmax_util.hpp
@@ -12,8 +12,8 @@
 #include "ck/tensor_operation/gpu/device/device_softmax.hpp"
 
 #include "ck/library/utility/check_err.hpp"
-#include "ck/library/host_tensor/host_tensor.hpp"
-#include "ck/library/host_tensor/device_memory.hpp"
+#include "ck/library/utility/host_tensor.hpp"
+#include "ck/library/utility/device_memory.hpp"
 #include "ck/library/reference_tensor_operation/cpu/reference_softmax.hpp"
 
 namespace ck {
@@ -80,8 +80,8 @@ class TestSoftmax : public ::testing::Test
 
         Tensor<OutDataType> out_ref(out);
 
-        DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpace());
-        DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpace());
+        DeviceMem in_dev(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());
+        DeviceMem out_dev(sizeof(OutDataType) * out.mDesc.GetElementSpaceSize());
         in_dev.ToDevice(in.mData.data());
         out_dev.ToDevice(out.mData.data());