Skip to content

fix: cap M_BTMM tile cost to actual seq/kv dimensions in flash_attention

d15d1f6
Select commit
Loading
Failed to load commit list.
Sign in for the full log view
Merged

feat: add analytic performance models (GEMM latency, ASM profiler, roofline) #3

fix: cap M_BTMM tile cost to actual seq/kv dimensions in flash_attention
d15d1f6
Select commit
Loading
Failed to load commit list.

Annotations

1 error and 1 warning
lint
failed Apr 1, 2026 in 35s