A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 02:15 UTC
[Bugfix][Build][Non-CUDA] Only referencing CMAKE_CUDA_COMPILER_VERSION on CUDA where it is defined (#20738)
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
Gregory Shtrasberg committed
Jul 10, 2025 at 05:58 UTC
dc221ad72d1bd64d1ab3bf916354b1608c6a61dc
1 file changed
+4
-2
CMakeLists.txt
+4
-2
@@ -174,8 +174,10 @@ endif()
174
#
175
# Set nvcc fatbin compression.
176
#
177
-if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND VLLM_GPU_LANG STREQUAL "CUDA")
178
- list(APPEND VLLM_GPU_FLAGS "-Xfatbin" "-compress-all" "-compress-mode=size")
177
+if(VLLM_GPU_LANG STREQUAL "CUDA")
178
+ if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8)
179
+ list(APPEND VLLM_GPU_FLAGS "-Xfatbin" "-compress-all" "-compress-mode=size")
180
+ endif()
181
endif()
182
183