ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

[ck] Enable missing op for gfx11 and gfx12 (#3187)

Browse Source
This commit is contained in:
linqunAMD
2025-11-11 02:58:20 +08:00
committed by GitHub
parent e593a14ae1
commit 7b6ba8d5c2
1 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
profiler/src/CMakeLists.txt
View File

@@ -40,6 +40,9 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_OPS profile_contraction_bilinear.cpp)
list(APPEND PROFILER_OPS profile_contraction_scale.cpp)
endif()
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx1[12]")
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
list(APPEND PROFILER_OPS profile_gemm_reduce.cpp)
list(APPEND PROFILER_OPS profile_batched_gemm_add_relu_gemm_add.cpp)
@@ -53,7 +56,7 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_OPS profile_grouped_gemm_tile_loop.cpp)
list(APPEND PROFILER_OPS profile_grouped_gemm_multiply_tile_loop.cpp)
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9[45]")
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9[45]" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
list(APPEND PROFILER_OPS profile_gemm_multiply_multiply_wp.cpp)
list(APPEND PROFILER_OPS profile_gemm_ab_scale.cpp)
list(APPEND PROFILER_OPS profile_gemm_blockscale_wp.cpp)
@@ -74,7 +77,6 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_OPS profile_conv_bwd_data.cpp)
list(APPEND PROFILER_OPS profile_conv_fwd.cpp)
list(APPEND PROFILER_OPS profile_grouped_conv_fwd_outelementop.cpp)
endif()
if((SUPPORTED_GPU_TARGETS MATCHES "gfx9" AND (DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)) OR