diff --git a/include/ck/tensor_operation/gpu/device/impl/device_contraction_multiple_d_xdl_cshuffle.hpp b/include/ck/tensor_operation/gpu/device/impl/device_contraction_multiple_d_xdl_cshuffle.hpp index 72c6d0b6f7..7a4c8bf267 100644 --- a/include/ck/tensor_operation/gpu/device/impl/device_contraction_multiple_d_xdl_cshuffle.hpp +++ b/include/ck/tensor_operation/gpu/device/impl/device_contraction_multiple_d_xdl_cshuffle.hpp @@ -586,6 +586,11 @@ struct DeviceContractionMultipleD_Xdl_CShuffle return false; } + if(ck::get_device_name() != "gfx90a" && std::is_same::value) + { + return false; + } + if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_m_k_, arg.b_grid_desc_n_k_, arg.ds_grid_desc_m_n_, diff --git a/script/cmake-ck-dev.sh b/script/cmake-ck-dev.sh index 3e530478b0..2e605ce8de 100755 --- a/script/cmake-ck-dev.sh +++ b/script/cmake-ck-dev.sh @@ -10,8 +10,8 @@ cmake -D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc \ -D CMAKE_CXX_FLAGS="-O3 -ftemplate-backtrace-limit=0 -gline-tables-only -save-temps=$PWD" \ -D CMAKE_BUILD_TYPE=Release \ --D BUILD_DEV=OFF \ --D GPU_TARGETS="gfx90a" \ +-D BUILD_DEV=ON \ +-D GPU_TARGETS="gfx908;gfx90a" \ -D CMAKE_VERBOSE_MAKEFILE:BOOL=ON \ -D USE_BITINT_EXTENSION_INT4=OFF \ ${MY_PROJECT_SOURCE}