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, 12:15 UTC
[Build/CI] Upgrade CUTLASS to 3.9.1 (#17602)
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
Tyler Michael Smith committed
May 3, 2025 at 01:25 UTC
c8386fa61d97657235497cf65b4d6c7c48be9a8a
1 file changed
+3
-4
CMakeLists.txt
+3
-4
@@ -249,9 +249,8 @@ set(VLLM_EXT_SRC
249
if(VLLM_GPU_LANG STREQUAL "CUDA")
250
SET(CUTLASS_ENABLE_HEADERS_ONLY ON CACHE BOOL "Enable only the header library")
251
252
- # Set CUTLASS_REVISION manually -- its revision detection doesn't work in this case.
253
- # Please keep this in sync with FetchContent_Declare line below.
254
- set(CUTLASS_REVISION "v3.9.0" CACHE STRING "CUTLASS revision to use")
252
+ # Set CUTLASS_REVISION. Used for FetchContent. Also fixes some bogus messages when building.
253
+ set(CUTLASS_REVISION "v3.9.1" CACHE STRING "CUTLASS revision to use")
254
255
# Use the specified CUTLASS source directory for compilation if VLLM_CUTLASS_SRC_DIR is provided
256
if (DEFINED ENV{VLLM_CUTLASS_SRC_DIR})
@@ -269,7 +268,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
268
cutlass
269
GIT_REPOSITORY https://github.com/nvidia/cutlass.git
270
# Please keep this in sync with CUTLASS_REVISION line above.
272
- GIT_TAG v3.9.0
271
+ GIT_TAG ${CUTLASS_REVISION}
272
GIT_PROGRESS TRUE
273
274
# Speed up CUTLASS download by retrieving only the specified GIT_TAG instead of the history.