Skip to content
v0.2.0MIT

Rigorous agent workflows for AMD GPU kernel development (aiter, FlyDSL, Triton, HIP/MFMA on gfx942 and gfx950) and Jira-ticket-driven work. A fork of pstack.

principle-explain-the-number

Apply before you trust, report, or act on a number you measured: a speedup, a regression, a throughput, a latency, or an eval result. Find what limits it, state a GPU kernel number as a percent of peak with the peak's source, and rule out that it measured something other than the work you think.

Read SKILL.md at the source

Pinned to revision bd140d538861, so it is the text this page describes rather than whatever the author pushed since.

Files

Every link opens the file at its source, pinned to the revision this page describes.