ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-16 19:09:59 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
123f9b50777093414fe661a41e239892736eddfb
composable_kernel/include/ck/tensor_operation/gpu/device/impl
History
Sami Remes 22f5a997fb Add other layouts for FP8 block scaled gemm (#2665) 
* Start adding other layouts for gemm_ab_scale

* Add some instances

* Create tensor descriptors for A/B scales depending on A/B layout

* Fix formatting

* Revert some comments

* Revert commented instances in CMakeLists.txt

* Add some more instances for col-row gemm

* enable more row,row instances

* Use occupancy=1 for col,row layout to avoid spills

[ROCm/composable_kernel commit: 26d3300930]
2025-08-18 01:46:10 -07:00
..
codegen_device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_avgpool2d_bwd_nhwc_nhwc.hpp
…
device_avgpool3d_bwd_ndhwc_ndhwc.hpp
…
device_batched_contraction_multiple_d_wmma_cshuffle.hpp
…
device_batched_contraction_multiple_d_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_e_permute_xdl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_gemm_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_multi_d_xdl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_multiple_d_dl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_multiple_d_gemm_multiple_d_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_multiple_d_xdl_cshuffle_v3.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00
device_batched_gemm_reduce_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_softmax_gemm_permute_wmma_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_softmax_gemm_permute_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_softmax_gemm_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_wmma_cshuffle_v3.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_xdl_fpAintB_b_scale.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batched_gemm_xdl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_batchnorm_backward_impl.hpp
…
device_batchnorm_forward_impl_obsolete.hpp
…
device_batchnorm_forward_impl.hpp
…
device_cgemm_4gemm_xdl_cshuffle.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_column_to_image_impl.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_contraction_multiple_abd_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_contraction_multiple_d_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_contraction_utils.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
…
device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_conv3d_fwd_naive_ndhwc_kzyxc_ndhwk.hpp
…
device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_convnd_bwd_data_nwc_kxc_nwk_dl.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_elementwise_dynamic_vector_dims_impl.hpp
…
device_elementwise_normalization_impl.hpp
…
device_elementwise_scale_impl.hpp
…
device_fpAintB_gemm_wmma.hpp
…
device_gemm_bias_add_reduce_xdl_cshuffle.hpp
…
device_gemm_dl.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_gemm_dpp.hpp
…
device_gemm_multiple_abd_xdl_cshuffle.hpp
…
device_gemm_multiple_d_dl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_gemm_multiple_d_layernorm_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_gemm_multiple_d_multiple_r_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_gemm_multiple_d_wmma_cshuffle.hpp
…
device_gemm_multiple_d_xdl_cshuffle_lds_direct_load.hpp
…
device_gemm_multiple_d_xdl_cshuffle_v3_ab_scale.hpp
Add other layouts for FP8 block scaled gemm (#2665)
2025-08-18 01:46:10 -07:00
device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp
…
device_gemm_multiple_d_xdl_cshuffle_v3_blockscale_bpreshuffle.hpp
Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)
2025-06-12 09:25:59 +08:00
device_gemm_multiple_d_xdl_cshuffle_v3.hpp
…
device_gemm_multiple_d_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_gemm_reduce_xdl_cshuffle.hpp
…
device_gemm_wmma_cshuffle_v3_b_scale.hpp
Support b_scale: (#2350)
2025-07-24 18:49:58 -07:00
device_gemm_wmma_cshuffle_v3_common.hpp
Support b_scale: (#2350)
2025-07-24 18:49:58 -07:00
device_gemm_wmma_cshuffle_v3.hpp
Support b_scale: (#2350)
2025-07-24 18:49:58 -07:00
device_gemm_wmma.hpp
…
device_gemm_xdl_cshuffle_lds_direct_load.hpp
…
device_gemm_xdl_cshuffle_streamk_v3.hpp
Stream-K Reduction option as Runtime parameter and Compilation Error Fix (SK- Reduction) (#2145)
2025-06-11 10:59:44 -07:00
device_gemm_xdl_cshuffle_v2.hpp
…
device_gemm_xdl_cshuffle_v3_b_preshuffle.hpp
CK pk_i4_t test failures fix (SWDEV-518629) (#2075)
2025-04-14 16:58:57 +08:00
device_gemm_xdl_cshuffle_v3_b_scale.hpp
CK pk_i4_t test failures fix (SWDEV-518629) (#2075)
2025-04-14 16:58:57 +08:00
device_gemm_xdl_cshuffle_v3_mx.hpp
MX GEMM - FP6 Example (#2419)
2025-07-07 10:33:26 -06:00
device_gemm_xdl_cshuffle_v3.hpp
CK pk_i4_t test failures fix (SWDEV-518629) (#2075)
2025-04-14 16:58:57 +08:00
device_gemm_xdl_cshuffle_v3r1.hpp
…
device_gemm_xdl_cshuffle.hpp
…
device_gemm_xdl_layernorm_cshuffle.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_gemm_xdl_skip_b_lds.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
device_gemm_xdl_splitk_c_shuffle_lds_direct_load.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
device_gemm_xdl_splitk_c_shuffle.hpp
…
device_gemm_xdl_streamk.hpp
…
device_gemm_xdl_waveletmodel_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_gemm_xdl.hpp
…
device_grouped_contraction_multiple_d_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_conv_bwd_data_multiple_d_wmma_cshuffle.hpp
Move SetZero functions inside the kernels for Grouped Conv (#2255)
2025-06-11 23:41:03 +02:00
device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_conv_bwd_weight_dl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_conv_bwd_weight_explicit_xdl.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00
device_grouped_conv_bwd_weight_multiple_d_xdl_cshuffle.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00
device_grouped_conv_bwd_weight_two_stage_xdl_cshuffle.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00
device_grouped_conv_bwd_weight_wmma_cshuffle.hpp
Fix grid size calc for bwd wei (#2226)
2025-05-26 16:51:09 +02:00
device_grouped_conv_bwd_weight_xdl_cshuffle_v3.hpp
Add padding to 1x1Stride1Pad0 conv specialization (grouped conv bwd weight) (#2675)
2025-08-14 00:21:09 +02:00
device_grouped_conv_bwd_weight_xdl_cshuffle.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00
device_grouped_conv_fwd_dl_multiple_d_nhwc_kyxc_nhwk.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_conv_fwd_dl_nhwc_kyxc_nhwk.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_conv_fwd_multiple_abd_xdl_cshuffle_v3.hpp
Enable multiple D for grouped conv fwd large tensors (#2572)
2025-07-28 22:39:07 +02:00
device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp
Enable multiple D for grouped conv fwd large tensors (#2572)
2025-07-28 22:39:07 +02:00
device_grouped_conv_fwd_multiple_d_multiple_r_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_conv_fwd_multiple_d_multiple_r.hpp
…
device_grouped_conv_fwd_multiple_d_wmma_cshuffle.hpp
…
device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp
…
device_grouped_conv_fwd_multiple_d_xdl_large_tensor_cshuffle.hpp
Enable multiple D for grouped conv fwd large tensors (#2572)
2025-07-28 22:39:07 +02:00
device_grouped_conv_utils.hpp
…
device_grouped_gemm_multi_abd_xdl_fixed_nk.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_gemm_multiple_d_dl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_gemm_multiple_d_splitk_xdl_cshuffle_two_stage.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_grouped_gemm_multiple_d_xdl_cshuffle_tile_loop.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_gemm_softmax_gemm_permute_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_gemm_xdl_fixed_nk.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_gemm_xdl_splitk_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_gemm_xdl.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_grouped_query_attention_forward_wmma.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_image_to_column_impl.hpp
…
device_max_pool_bwd_impl.hpp
…
device_moe_gemm_blockscale.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_moe_gemm.hpp
Handle moe_fp8 no-mainloop cases. Supprese no-mainloop check (#2438)
2025-07-16 15:44:34 +08:00
device_moe_mx_gemm_bns.hpp
Add MoE & FP8 Blockscale WP Kernels for GFX950 (#2297)
2025-06-12 09:25:59 +08:00
device_moe_mx_gemm_bpreshuffle.hpp
upgrade from clang-format-12 to clang-format-18 (#2568)
2025-07-28 11:34:07 -07:00
device_moe_mx_gemm.hpp
[CK] Mxfp4 moe blockscale buf2lds version support (#2455)
2025-07-06 15:42:00 +08:00
device_multi_query_attention_forward_wmma.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
device_multiple_reduce_multiblock.hpp
…
device_multiple_reduce_threadwise.hpp
…
device_normalization_bwd_data_impl.hpp
…
device_normalization_bwd_gamma_beta_impl.hpp
…
device_normalization_fwd_impl.hpp
…
device_normalization_fwd_splitk_impl.hpp
…
device_permute_impl.hpp
…
device_pool2d_fwd_nhwc_nhwc.hpp
…
device_pool3d_fwd_ndhwc_ndhwc.hpp
…
device_put_element_impl.hpp
…
device_reduce_common.hpp
…
device_reduce_multiblock.hpp
…
device_reduce_threadwise_multi_d.hpp
…
device_reduce_threadwise.hpp
…
device_softmax_impl.hpp
…
device_sparse_embeddings_forward_layernorm.hpp
update the switch condition for buffer built-ins (#2602)
2025-08-01 14:30:07 -07:00
device_splitk_contraction_multiple_d_xdl_cshuffle.hpp
Remove !defined(__HIP_DEVICE_COMPILE__) in CK kernel (#2564)
2025-07-28 13:01:07 -07:00
split_k_arg.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00
split_k_utils.hpp
Automatic deduction of split-K value for grouped convolution (#2491)
2025-07-31 12:08:45 +02:00