yadaish c0ee71d735 Dev/a8w4 and a8w8splitk (#3447) ...

* Ck moe bs splitk pr (#3440)

* splitk kick-off. Compilation fail

* splitk hack pass

* fix scale offset calc.

* clang-format for a8w8_moe_blk_gemm1 splitk change

* fix testcase error

---------

Co-authored-by: oscar <huaiguxu@amd.com>
Co-authored-by: huaiguxu <145733371+huaiguxu@users.noreply.github.com>

* Zan/moe a8w4 (#3441)

* update

* update

* update ck moe a8w4

* update

* update

* update

* compile pass

* update

* update

* python3 op_tests/test_moe_2stage.py -t 16 -e 1 -k 1 -dim 256,256 ready

* support new a8w4 kernel

* update

* update ck_tile

* re format

* update

* update

* fix conflict

* fix build

* update ck_tile moe

* fix clang format

* fix the problem

* fix accruacy issue

* fix

---------

Co-authored-by: oscar <huaiguxu@amd.com>
Co-authored-by: huaiguxu <145733371+huaiguxu@users.noreply.github.com>
Co-authored-by: Zzz9990 <zanzhang@amd.com>
Co-authored-by: felix <felix.li@amd.com>

2025-12-19 09:26:52 +08:00

..

blockwise_gemm_dl_v2r3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_dlops_v2r2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_dlops_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_dpp.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_mx_pipeline_xdlops_base.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_wmma_selector.hpp

Wmma support for gemm_multiply_multiply_wp (#3278)

2025-12-03 07:38:23 -08:00

blockwise_gemm_pipeline_wmmaops_base.hpp

Wmma support for gemm_ab_scale (#3314)

2025-12-11 09:06:20 +01:00

blockwise_gemm_pipeline_wmmaops_v1.hpp

Wmma support for gemm_ab_scale (#3314)

2025-12-11 09:06:20 +01:00

blockwise_gemm_pipeline_wmmaops_v3.hpp

Wmma support for gemm_ab_scale (#3314)

2025-12-11 09:06:20 +01:00

blockwise_gemm_pipeline_wmmaops.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_ab_scale_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_dequant_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_dequant_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_gufusion_dequant_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_gufusion_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_gufusion_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_gufusion_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_v2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_preshuffle_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_b_scale_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_base.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_blockscale_b_preshuffle_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_blockscale_b_preshuffle_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_blockscale_b_preshuffle_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_gufusion_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_gufusion_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_v1.hpp

Dev/a8w4 and a8w8splitk (#3447)

2025-12-19 09:26:52 +08:00

blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_bpreshuffle_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_gufusion_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_nbs_gufusion_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_nbs_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_nbs_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_nbs_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_moe_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_mx_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_selector.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v1_ab_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v1_b_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v1_mx.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v2_ab_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v2_b_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v3_ab_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v3_b_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v3_mx_bpreshuffle.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v3_mx.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v4_b_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v4.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops_v5.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_pipeline_xdlops.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_smfmac_xdlops.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_wmma.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_xdlops_skip_b_lds.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_gemm_xdlops.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_softmax.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_tensor_slice_transfer_v5r1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

blockwise_welford.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

reduction_functions_blockwise.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_direct_load.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_gather_direct_load.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_global.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v4r1_dequant.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v4r1_gather.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v4r1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v4r2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v6r1.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v6r1r2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v6r2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v6r3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v7.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v7r2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v7r3_scatter.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

thread_group_tensor_slice_transfer_v7r3.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00