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 6, 14:30 UTC
pick up tuned prefill configs for FP8 FA3 (#36265)
Signed-off-by: Jonas M. Kübler <44084297+jmkuebler@users.noreply.github.com> Signed-off-by: Jonas Kuebler <kuebj@amazon.com>
Jonas M. Kübler committed
Mar 17, 2026 at 15:00 UTC
77d2a5f17b38941f969cec3c91bceb45e2ba10cf
1 file changed
+1
-1
cmake/external_projects/vllm_flash_attn.cmake
+1
-1
@@ -39,7 +39,7 @@ else()
39
FetchContent_Declare(
40
vllm-flash-attn
41
GIT_REPOSITORY https://github.com/vllm-project/flash-attention.git
42
- GIT_TAG 1488682bb545f7d020e958a33116b1419d1cfc83
42
+ GIT_TAG 29210221863736a08f71a866459e368ad1ac4a95
43
GIT_PROGRESS TRUE
44
# Don't share the vllm-flash-attn build between build types
45
BINARY_DIR ${CMAKE_BINARY_DIR}/vllm-flash-attn