Skip to content

Prefer FlashAttention MLA as default over FlashMLA #27363

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
LucasWilkinson merged 1 commit into from
Nov 11, 2025
Merged

Prefer FlashAttention MLA as default over FlashMLA #27363

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions vllm/platforms/cuda.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,15 +55,15 @@ def _get_backend_priorities(
return [
AttentionBackendEnum.CUTLASS_MLA,
AttentionBackendEnum.FLASHINFER_MLA,
AttentionBackendEnum.FLASHMLA,
AttentionBackendEnum.FLASH_ATTN_MLA,
AttentionBackendEnum.FLASHMLA,
Comment on lines -58 to +59
Copy link
Member

@mgoin mgoin Nov 11, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just with respect to Blackwell, do we actually know how flashmla and fa mla work on there? I'm curious if fa mla is even a valid option

Copy link
Contributor Author

@MatthewBonanni MatthewBonanni Nov 11, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I actually marked FA MLA as supporting only CC 9.x, so it'll get skipped on Blackwell regardless of its spot in the priority list. It could probably just be removed from the list, but I'll test it out anyway to check. FlashMLA supports Blackwell but I haven't benchmarked it

AttentionBackendEnum.TRITON_MLA,
AttentionBackendEnum.FLASHMLA_SPARSE,
]
else:
return [
AttentionBackendEnum.FLASHMLA,
AttentionBackendEnum.FLASH_ATTN_MLA,
AttentionBackendEnum.FLASHMLA,
AttentionBackendEnum.FLASHINFER_MLA,
AttentionBackendEnum.TRITON_MLA,
AttentionBackendEnum.FLASHMLA_SPARSE,
Expand Down
Loading