Kernel Bottleneck Tuning

Workflow for finding and fixing instruction-throughput bottlenecks in warp-specialized CuTeDSL kernels (FA4-style). Use when a kernel underperforms its roofline and you need to decide between reordering, eliminating, or re-pipelining work.

hao-ai-lab Updated 4.8k repo stars

File contents

hao-ai-lab/flash-attention-fp4 commit 4e2e34a336

Frequently asked questions

npx get-skill add hao-ai-lab/kernel-bottleneck-tuning