From cb3935a8fc1968cdd2d7fa2dfd963900419820a0 Mon Sep 17 00:00:00 2001 From: Lucas Wilkinson Date: Thu, 2 Apr 2026 13:02:37 -0400 Subject: [PATCH] [FA4] Update flash-attention to latest upstream FA4 (#38690) Signed-off-by: Lucas Wilkinson --- cmake/external_projects/vllm_flash_attn.cmake | 2 +- requirements/cuda.txt | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/cmake/external_projects/vllm_flash_attn.cmake b/cmake/external_projects/vllm_flash_attn.cmake index 443d41d5a21..9414f5af372 100644 --- a/cmake/external_projects/vllm_flash_attn.cmake +++ b/cmake/external_projects/vllm_flash_attn.cmake @@ -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 diff --git a/requirements/cuda.txt b/requirements/cuda.txt index 6d7f9693f75..cfee494b5a6 100644 --- a/requirements/cuda.txt +++ b/requirements/cuda.txt @@ -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