Skip to content

[Cute,Fwd,Sm100] hdim 192 smem fix #2235

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
tridao merged 1 commit into from
Feb 5, 2026
+7 −1
Merged

[Cute,Fwd,Sm100] hdim 192 smem fix #2235

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
8 changes: 7 additions & 1 deletion flash_attn/cute/flash_fwd_sm100.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -234,7 +234,13 @@ def _setup_attributes(self):
- Configures pipeline stages for softmax, correction, and epilogue operations
"""

self.kv_stage = 4 if self.q_dtype.width == 8 or self.q_stage == 1 else 3
self.kv_stage = (
4
if (self.q_dtype.width == 8 or self.q_stage == 1)
and self.head_dim_padded <= 128
and self.head_dim_v_padded <= 128
else 3
)
self.acc_stage = 1
# For hdim 192,128, we don't have enough smem to store all 3 stages of KV:
# 128 x 192 x 2 bytes x 3 stages = 144KB, and we need 96KB for Q.
Expand Down
Loading