xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-08 18:09:08 +08:00
Code Issues Packages Projects Releases Wiki Activity

undo

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
This commit is contained in:
Lucas Wilkinson 2025-05-15 05:47:08 +00:00
parent 4e00778a60
commit 7ebe64f94b
2 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
CMakeLists.txt
View File

@ -483,6 +483,10 @@ set(VLLM_MOE_EXT_SRC
"csrc/moe/moe_align_sum_kernels.cu" "csrc/moe/moe_align_sum_kernels.cu"
"csrc/moe/topk_softmax_kernels.cu") "csrc/moe/topk_softmax_kernels.cu")
if(VLLM_GPU_LANG STREQUAL "CUDA")
list(APPEND VLLM_MOE_EXT_SRC "csrc/moe/moe_wna16.cu")
endif()
# Apply gencode flags to base MOE extension sources # Apply gencode flags to base MOE extension sources
set_gencode_flags_for_srcs( set_gencode_flags_for_srcs(
SRCS "${VLLM_MOE_EXT_SRC}" SRCS "${VLLM_MOE_EXT_SRC}"
@ -495,7 +499,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
ARCHS "8.0;9.0+PTX" ARCHS "8.0;9.0+PTX"
GEN_SCRIPT "${CMAKE_CURRENT_SOURCE_DIR}/csrc/moe/marlin_moe_wna16/generate_kernels.py" GEN_SCRIPT "${CMAKE_CURRENT_SOURCE_DIR}/csrc/moe/marlin_moe_wna16/generate_kernels.py"
GEN_GLOB "csrc/moe/marlin_moe_wna16/*.cu" GEN_GLOB "csrc/moe/marlin_moe_wna16/*.cu"
SRCS "csrc/moe/moe_wna16.cu"
OUT_SRCS_VAR VLLM_MOE_EXT_SRC OUT_SRCS_VAR VLLM_MOE_EXT_SRC
) )
endif() endif()

4
cmake/utils.cmake
View File

@ -140,8 +140,8 @@ macro(optional_cuda_sources)
if(NOT OCS_ARCHS) if(NOT OCS_ARCHS)
message(FATAL_ERROR "optional_cuda_sources ${OCS_NAME}: ARCHS is required") message(FATAL_ERROR "optional_cuda_sources ${OCS_NAME}: ARCHS is required")
endif() endif()
if(NOT OCS_SRCS) if(NOT OCS_SRCS AND NOT OCS_GEN_SCRIPT)
message(FATAL_ERROR "optional_cuda_sources ${OCS_NAME}: SRCS is required") message(FATAL_ERROR "optional_cuda_sources ${OCS_NAME}: either SRCS or GEN_SCRIPT must be provided")
endif() endif()
if(NOT OCS_MIN_VERSION) if(NOT OCS_MIN_VERSION)
set(OCS_MIN_VERSION "0.0") set(OCS_MIN_VERSION "0.0")