We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent 8ce664f commit ecc8815Copy full SHA for ecc8815
1 file changed
src/turbomind/kernels/gemm/CMakeLists.txt
@@ -16,7 +16,6 @@ set(GEMM2_KERNELS_SM80
16
kernel/sm80_16816_16.cu
17
)
18
set(GEMM2_KERNELS_SM90
19
- tma.cu
20
kernel/sm90_16816_4.cu
21
kernel/sm90_16816_8.cu
22
kernel/sm90_16816_16.cu
@@ -87,7 +86,7 @@ set_property(TARGET gemm2 PROPERTY CUDA_RESOLVE_DEVICE_SYMBOLS ON)
87
86
88
if(GEMM2_ARCH_90_ENABLED)
89
# SM90 kernels only compile for 90/90a; avoid building them for sm_100.
90
- add_library(gemm2_sm90 STATIC ${GEMM2_KERNELS_SM90})
+ add_library(gemm2_sm90 STATIC ${GEMM2_KERNELS_SM90} tma.cu )
91
set_target_properties(gemm2_sm90 PROPERTIES
92
CUDA_ARCHITECTURES "${_sm90_archs}"
93
POSITION_INDEPENDENT_CODE ON
0 commit comments