mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-26 08:00:13 +00:00
7c9100b53f
* Added bwd data v3r1: breaking down compute into a series of load balanced GEMM, and launch in a single kernel
* Added bwd data v4r1: like v3r1, but launch GEMMs in multiple kernels
* Tweaked v1r1 and v1r2 (atomic) on AMD GPU
[ROCm/composable_kernel commit: c5da0377fb]
33 lines
1.0 KiB
CMake
33 lines
1.0 KiB
CMake
set(TENSOR_SOURCE
|
|
src/tensor.cpp;
|
|
src/device.cpp;
|
|
)
|
|
|
|
add_library(host SHARED ${TENSOR_SOURCE})
|
|
target_compile_features(host PUBLIC)
|
|
set_target_properties(host PROPERTIES POSITION_INDEPENDENT_CODE ON)
|
|
|
|
if(DEVICE_BACKEND STREQUAL "NVIDIA")
|
|
target_link_libraries(host nvToolsExt cudart)
|
|
endif()
|
|
|
|
install(TARGETS host LIBRARY DESTINATION lib)
|
|
|
|
|
|
if(DEVICE_BACKEND STREQUAL "AMD")
|
|
set(CONV_SOURCE src/conv_driver.cpp)
|
|
set(COL2IM_SOURCE src/col2im_driver.cpp)
|
|
set(CONV_BWD_DATA_SOURCE src/conv_bwd_data_driver.cpp)
|
|
elseif(DEVICE_BACKEND STREQUAL "NVIDIA")
|
|
set(CONV_SOURCE src/conv_driver.cu)
|
|
set(COL2IM_SOURCE src/col2im_driver.cu)
|
|
set(CONV_BWD_DATA_SOURCE src/conv_bwd_data_driver.cu)
|
|
endif()
|
|
|
|
add_executable(conv_driver ${CONV_SOURCE})
|
|
add_executable(col2im_driver ${COL2IM_SOURCE})
|
|
add_executable(conv_bwd_data_driver ${CONV_BWD_DATA_SOURCE})
|
|
target_link_libraries(conv_driver PRIVATE host)
|
|
target_link_libraries(col2im_driver PRIVATE host)
|
|
target_link_libraries(conv_bwd_data_driver PRIVATE host)
|