Skip to content

fallback to fa2 (instead of fa3) for unsupported configuration (bf16 Q, Fp8 KV) #2536

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
yzh119 merged 2 commits into from
Feb 12, 2026
Merged

fallback to fa2 (instead of fa3) for unsupported configuration (bf16 Q, Fp8 KV) #2536

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0607bc8
fallback to fa2 (instead of fa3) for unsupported bf16 Q fp8 kv
saltyminty Feb 10, 2026
fa95004
make gemini happy
saltyminty Feb 10, 2026
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
6 changes: 6 additions & 0 deletions flashinfer/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -408,6 +408,12 @@ def is_fa3_backend_supported(
return False
if use_fp16_qk_reductions:
return False
# FA3 FP8 KV cache currently requires FP8 query.
if dtype_kv in {torch.float8_e4m3fn, torch.float8_e5m2} and dtype_q not in {
torch.float8_e4m3fn,
torch.float8_e5m2,
}:
return False
return True


Expand Down
Loading