Why we moved to a 32‑bit floating‑point fuel model
Moving the M1 fuel model from fixed‑point to 32‑bit floating point removed a whole class of scaling bugs — for about 4 µs…
Experiments, bench notes and design considerations from the FUP workshop.
Moving the M1 fuel model from fixed‑point to 32‑bit floating point removed a whole class of scaling bugs — for about 4 µs…
Validating the wide‑band knock detection pipeline on the M1 prototype against a reference engine — latency, false positives, and one surprising noise…