Commit 0cd7fed

authored

[autoWS][FA] Add vectorization and fadd2_reduce (#540)

Summary: mark N_CTX constexpr

1 parent ff11de3 commit 0cd7fedCopy full SHA for 0cd7fed

2 files changed

+210

-42

lines changed

tritonbench/kernels
- blackwell_triton_fused_attention.py
- blackwell_triton_fused_attention_dp.py

2 files changed

+210

-42

lines changed

Comments

(0)