You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Michael Fischer edited this page Jun 24, 2026
·
2 revisions
Recording Investigation Notes
When an experiment, capture, or debugging session on the NPU produces a non-obvious
conclusion, record it here so the next person (or agent) doesn't re-derive it.
Where it goes
A new register meaning, tile layout, limit, or encoding detail → add it to
regcmd ISA Reference (the durable reference).
A milestone, result, or a path that did/didn't work → append a dated entry to the
Reverse-Engineering Roadmap.
A ruled-out interpretation → add it to the "Ruled out" section of the ISA Reference with why it
failed, so it isn't re-walked.
An operational how-to (build, add an SoC) → that belongs in the repo (README.md,
docs/), not the wiki.
How to write it
State the conclusion first, then the evidence. Quote concrete numbers (shapes, register
values, mismatch counts, tok/s) — the value is in the specifics.
Note the SoC and conditions (the validated board is RK3588; numbers can be SoC-specific).
If a finding fixed a bug, name the regression case that now guards it.
Prefer capture→diff→decode evidence over speculation; mark unverified hypotheses as such.