mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-03 05:01:25 +00:00
dd7af118d7
* Revert "Revert "feature:tf32:add initial conv3d fwd kernel support (#2763)" (#2848)"
This reverts commit 03b59f8c76.
* fix compile error on gf12x
* only run tf32 example on gfx942
* only build tf32 instance on gfx942
* ckProfiler:only support tf32 in gfx942
* delete unuseful messages
32 lines
1.5 KiB
CMake
32 lines
1.5 KiB
CMake
add_example_executable(example_convnd_fwd_xdl_fp32 convnd_fwd_xdl_fp32.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_fp16 convnd_fwd_xdl_fp16.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_bf16 convnd_fwd_xdl_bf16.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_int8 convnd_fwd_xdl_int8.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_fp8 convnd_fwd_xdl_fp8.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_bf8 convnd_fwd_xdl_bf8.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_fp16_comp_fp8 convnd_fwd_xdl_fp16_comp_fp8.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_fp8_bf8 convnd_fwd_xdl_fp8_bf8.cpp)
|
|
add_example_executable(example_convnd_fwd_xdl_bf8_fp8 convnd_fwd_xdl_bf8_fp8.cpp)
|
|
add_example_executable(example_convnd_fwd_dl_fp16 convnd_fwd_dl_fp16.cpp)
|
|
add_example_executable(example_convnd_fwd_dl_fp32 convnd_fwd_dl_fp32.cpp)
|
|
add_example_executable(example_convnd_fwd_dl_int8 convnd_fwd_dl_int8.cpp)
|
|
|
|
# only build fp64 example for the following targets
|
|
list(APPEND gpu_list gfx90a gfx942 gfx950)
|
|
set(target 0)
|
|
foreach(gpu IN LISTS GPU_TARGETS)
|
|
if(gpu IN_LIST gpu_list AND target EQUAL 0)
|
|
add_example_executable(example_convnd_fwd_xdl_fp64 convnd_fwd_xdl_fp64.cpp)
|
|
set(target 1)
|
|
endif()
|
|
endforeach()
|
|
|
|
list(APPEND gpu_list_tf32 gfx942)
|
|
set(target 0)
|
|
foreach(gpu IN LISTS GPU_TARGETS)
|
|
if(gpu IN_LIST gpu_list_tf32 AND target EQUAL 0)
|
|
add_example_executable(example_convnd_fwd_xdl_fp32_tf32 convnd_fwd_xdl_fp32_tf32.cpp)
|
|
set(target 1)
|
|
endif()
|
|
endforeach()
|