Skip to content

Remove dpad==dvpad limit in CK FA bwd codegen #677

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
valarLip merged 2 commits into from
Jul 18, 2025
Merged

Remove dpad==dvpad limit in CK FA bwd codegen #677

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e650836
update 3rdparty
slippedJim Jul 18, 2025
a753b80
update 3rdprty
slippedJim Jul 18, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion 3rdparty/composable_kernel
Open in desktop
Submodule composable_kernel updated 83 files
+34 −8 Jenkinsfile
+0 −3 example/ck_tile/01_fmha/codegen/ops/fmha_bwd.py
+10 −21 example/ck_tile/01_fmha/codegen/ops/fmha_fwd.py
+1 −1 example/ck_tile/03_gemm/README.md
+4 −4 example/ck_tile/03_gemm/gemm_utils.hpp
+1 −1 example/ck_tile/03_gemm/gemm_weight_preshuffle.cpp
+10 −2 example/ck_tile/03_gemm/run_gemm_example.inc
+1 −1 example/ck_tile/18_flatmm/run_flatmm_example.inc
+4 −2 example/ck_tile/37_transpose/transpose_policy.hpp
+168 −56 include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v3_mx_bpreshuffle.hpp
+1 −0 include/ck_tile/core.hpp
+6 −4 include/ck_tile/core/arch/amd_buffer_addressing.hpp
+6 −4 include/ck_tile/core/arch/amd_buffer_addressing_builtins.hpp
+30 −28 include/ck_tile/core/arch/amd_transpose_load_encoding.hpp
+25 −40 include/ck_tile/core/tensor/buffer_view.hpp
+206 −122 include/ck_tile/core/tensor/load_tile_transpose.hpp
+156 −0 include/ck_tile/core/utility/debug.hpp
+2 −2 include/ck_tile/ops/flatmm/block/flatmm_32x512x128_1x4x1_16x16x32.hpp
+2 −2 include/ck_tile/ops/flatmm/block/flatmm_sn_32x128x512_1x4x1_16x16x32.hpp
+3 −3 include/ck_tile/ops/fmha.hpp
+3 −3 include/ck_tile/ops/gemm.hpp
+40 −6 include/ck_tile/ops/gemm/block/block_gemm_asmem_bsmem_creg_v1_default_policy.hpp
+119 −35 include/ck_tile/ops/gemm/block/block_universal_gemm_as_bs_cr.hpp
+50 −17 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_base.hpp
+13 −7 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_comp_v3.hpp
+76 −65 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_comp_v4.hpp
+15 −3 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_comp_v4_default_policy.hpp
+51 −28 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_mem.hpp
+2 −0 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_agmem_bgmem_creg_v1.hpp
+3 −0 include/ck_tile/ops/gemm/pipeline/gemm_pipeline_problem.hpp
+97 −47 include/ck_tile/ops/gemm/pipeline/gemm_universal_pipeline_ag_bg_cr_policy.hpp
+1 −1 include/ck_tile/ops/gemm/pipeline/wp_pipeline_agmem_bgmem_creg_v1.hpp
+61 −23 include/ck_tile/ops/gemm/warp/warp_gemm.hpp
+154 −125 include/ck_tile/ops/gemm/warp/warp_gemm_attribute_mfma.hpp
+44 −17 include/ck_tile/ops/gemm/warp/warp_gemm_dispatcher.hpp
+13 −4 ...sor_operation_instance/gpu/gemm_mx/device_gemm_mx_xdl_f4_f4_f16/device_gemm_mx_xdl_f4_f4_f16_mk_mfma_mn.hpp
+5 −2 script/cmake-ck-dev.sh
+5 −2 script/cmake-ck-release.sh
+3 −0 test/CMakeLists.txt
+2 −0 test/ck_tile/CMakeLists.txt
+33 −0 test/ck_tile/batched_transpose/CMakeLists.txt
+25 −0 test/ck_tile/batched_transpose/batched_transpose.hpp
+283 −0 test/ck_tile/batched_transpose/batched_transpose.inc
+113 −0 test/ck_tile/batched_transpose/batched_transpose_api.cpp
+10 −0 test/ck_tile/batched_transpose/batched_transpose_bf16.cpp
+10 −0 test/ck_tile/batched_transpose/batched_transpose_fp16.cpp
+10 −0 test/ck_tile/batched_transpose/batched_transpose_fp8.cpp
+2 −2 test/ck_tile/gemm/test_gemm_pipeline_util.hpp
+28 −0 test/ck_tile/smoothquant/CMakeLists.txt
+21 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n1024_instance.cpp
+12 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n1536_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n2048_instance.cpp
+11 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n256_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n3072_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n4096_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n4096_tp_instance.cpp
+12 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n512_instance.cpp
+11 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n64_n128_instance.cpp
+11 −0 test/ck_tile/smoothquant/instances/smoothquant_bf16_n768_instance.cpp
+21 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n1024_instance.cpp
+12 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n1536_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n2048_instance.cpp
+11 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n256_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n3072_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n4096_instance.cpp
+13 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n4096_tp_instance.cpp
+12 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n512_instance.cpp
+11 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n64_n128_instance.cpp
+11 −0 test/ck_tile/smoothquant/instances/smoothquant_fp16_n768_instance.cpp
+143 −0 test/ck_tile/smoothquant/instances/smoothquant_fwd_api.cpp
+61 −0 test/ck_tile/smoothquant/instances/smoothquant_instance_common.hpp
+114 −0 test/ck_tile/smoothquant/smoothquant.hpp
+274 −0 test/ck_tile/smoothquant/smoothquant.inc
+11 −0 test/ck_tile/smoothquant/smoothquant_bf16.cpp
+11 −0 test/ck_tile/smoothquant/smoothquant_fp16.cpp
+35 −21 tile_engine/ops/gemm/CMakeLists.txt
+15 −11 tile_engine/ops/gemm/README.md
+0 −15 tile_engine/ops/gemm/configs/benchmark.json
+82 −0 tile_engine/ops/gemm/configs/custom_ci_config.json
+0 −15 tile_engine/ops/gemm/configs/default_config.json
+0 −15 tile_engine/ops/gemm/configs/user_provided_config.json
+34 −16 tile_engine/ops/gemm/gemm_instance_builder.py
+29 −19 tile_engine/ops/gemm/json_config.py