Skip to content

[https://nvbugs/6198760][fix] Refresh FMHA cubins to fix SageAttention when KV-sequence is not a multiple of 128 - #17648

Merged
xrq-phys merged 3 commits into
NVIDIA:mainfrom
xrq-phys:ruqingx/fix/kernel-refresh
Aug 18, 2026
Merged

[https://nvbugs/6198760][fix] Refresh FMHA cubins to fix SageAttention when KV-sequence is not a multiple of 128#17648
xrq-phys merged 3 commits into
NVIDIA:mainfrom
xrq-phys:ruqingx/fix/kernel-refresh

Adapt SF handling for fixed SageAttn kernels: fix SeqLen%128!=0, fix …

724b944
Select commit
Loading
Failed to load commit list.
DCO / DCO succeeded Aug 18, 2026 in 0s

DCO

All commits are signed off!

View more details on DCO