Skip to content

perf: turbo VEC flash attention — +9% decode on CUDA via autoresearch #44

perf: turbo VEC flash attention — +9% decode on CUDA via autoresearch

perf: turbo VEC flash attention — +9% decode on CUDA via autoresearch #44

Triggered via pull request April 9, 2026 17:07
Status Failure
Total duration 28m 42s
Artifacts

hip-quality-check.yml

on: pull_request
ubuntu-22-hip-quality-check
28m 38s
ubuntu-22-hip-quality-check
Fit to window
Zoom out
Zoom in

Annotations

1 error
ubuntu-22-hip-quality-check
Process completed with exit code 1.