Merge remote-tracking branch 'origin/cka8w8_devtimer' into update_cka8w8_uc

e92395d9 · coderfeli · 842d910e · 7efafa11 · e92395d9 · e92395d9
Commit e92395d9 authored Dec 27, 2024 by coderfeli
20 changed files
--- a/include/ck_tile/ops/gemm/kernel/batched_gemm_kernel.hpp
+++ b/include/ck_tile/ops/gemm/kernel/batched_gemm_kernel.hpp
--- a/include/ck_tile/ops/gemm/kernel/gemm_kernel.hpp
+++ b/include/ck_tile/ops/gemm/kernel/gemm_kernel.hpp
--- a/include/ck_tile/ops/gemm/warp/warp_gemm.hpp
+++ b/include/ck_tile/ops/gemm/warp/warp_gemm.hpp
--- a/include/ck_tile/ops/gemm/warp/warp_gemm_attribute_mfma.hpp
+++ b/include/ck_tile/ops/gemm/warp/warp_gemm_attribute_mfma.hpp
--- a/include/ck_tile/ops/gemm/warp/warp_gemm_attribute_mfma_impl.hpp
+++ b/include/ck_tile/ops/gemm/warp/warp_gemm_attribute_mfma_impl.hpp
--- a/include/ck_tile/ops/gemm/warp/warp_gemm_dispatcher.hpp
+++ b/include/ck_tile/ops/gemm/warp/warp_gemm_dispatcher.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm_universal_batched/device_batched_gemm_xdl_universal_bf16_bf16_bf16/device_batched_gemm_xdl_universal_bf16_bf16_bf16_mk_nk_mn.hpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_universal_batched/device_batched_gemm_xdl_universal_bf16_bf16_bf16/device_batched_gemm_xdl_universal_bf16_bf16_bf16_mk_nk_mn.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm_universal_batched/device_batched_gemm_xdl_universal_f8_f8_bf16/device_batched_gemm_xdl_universal_f8_f8_bf16_mk_nk_mn.hpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_universal_batched/device_batched_gemm_xdl_universal_f8_f8_bf16/device_batched_gemm_xdl_universal_f8_f8_bf16_mk_nk_mn.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_f16_f16_f16/device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn.hpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_f16_f16_f16/device_gemm_xdl_universal_streamk_f16_f16_f16_mk_kn_mn.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_f16_f16_f16/device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn.hpp
+++ b/library/src/tensor_operation_instance/gpu/gemm_universal_streamk/device_gemm_xdl_universal_streamk_f16_f16_f16/device_gemm_xdl_universal_streamk_f16_f16_f16_mk_nk_mn.hpp
--- a/profiler/include/profiler/profile_gemm_multiply_multiply_impl.hpp
+++ b/profiler/include/profiler/profile_gemm_multiply_multiply_impl.hpp
--- a/profiler/include/profiler/profile_gemm_universal_batched_impl.hpp
+++ b/profiler/include/profiler/profile_gemm_universal_batched_impl.hpp
--- a/profiler/include/profiler/profile_gemm_universal_impl.hpp
+++ b/profiler/include/profiler/profile_gemm_universal_impl.hpp
--- a/profiler/include/profiler/profile_grouped_gemm_impl.hpp
+++ b/profiler/include/profiler/profile_grouped_gemm_impl.hpp
--- a/profiler/src/profile_gemm_universal_batched.cpp
+++ b/profiler/src/profile_gemm_universal_batched.cpp
--- a/script/process_perf_data.py
+++ b/script/process_perf_data.py
--- a/test/ck_tile/batched_gemm/test_batched_gemm_util.hpp
+++ b/test/ck_tile/batched_gemm/test_batched_gemm_util.hpp
--- a/test/ck_tile/gemm/CMakeLists.txt
+++ b/test/ck_tile/gemm/CMakeLists.txt
 # Currently ck_tile is only built on gfx9
 if(GPU_TARGETS MATCHES "gfx9")
-    add_gtest_executable(test_ck_tile_gemm_mem_pipeline test_gemm_mem_pipeline.cpp)
+    add_gtest_executable(test_ck_tile_gemm_pipeline test_gemm_pipeline.cpp)
 endif()
--- a/test/ck_tile/gemm/test_gemm_mem_pipeline.cpp
+++ b/test/ck_tile/gemm/test_gemm_mem_pipeline.cpp
--- a/test/ck_tile/gemm/test_gemm_mem_pipeline_ut_cases.inc
+++ b/test/ck_tile/gemm/test_gemm_mem_pipeline_ut_cases.inc