biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[FA4] Update flash-attention to latest upstream FA4 (#38690)

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
This commit is contained in:
Lucas Wilkinson
2026-04-02 13:02:37 -04:00
committed by GitHub
parent 82a006beeb
commit cb3935a8fc
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
cmake/external_projects/vllm_flash_attn.cmake
View File

@@ -39,7 +39,7 @@ else()
FetchContent_Declare(
vllm-flash-attn
GIT_REPOSITORY https://github.com/vllm-project/flash-attention.git
GIT_TAG 29210221863736a08f71a866459e368ad1ac4a95
GIT_TAG c0ec424fd8a546d0cbbf4bf050bbcfe837c55afb
GIT_PROGRESS TRUE
# Don't share the vllm-flash-attn build between build types
BINARY_DIR ${CMAKE_BINARY_DIR}/vllm-flash-attn

4
requirements/cuda.txt
View File

@@ -16,5 +16,5 @@ flashinfer-cubin==0.6.7
nvidia-cudnn-frontend>=1.13.0,<1.19.0
# QuACK and Cutlass DSL for FA4 (cute-DSL implementation)
nvidia-cutlass-dsl>=4.4.0.dev1
quack-kernels>=0.2.7
nvidia-cutlass-dsl>=4.4.2
quack-kernels>=0.3.3