From 9629aa9fa2e552435c274725299a24f300931cd1 Mon Sep 17 00:00:00 2001 From: Loom Auditor Date: Sun, 11 Oct 2026 03:08:51 +0000 Subject: [PATCH] docs(benchmarks): mark the six electron bench digests stale after #241/#339/#342 (#366) Co-Authored-By: Claude Sonnet 5.5 Loom-Story: 2AMLogic/lindhard#366 Loom-Trace-Id: ca7ed8c8abe83ea584817efdd4925df8 Loom-Build: 0.19.1039 e83d19dc0a9b217339ea2f14040c50ad6f46700d clean --- docs/benchmarks.md | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/docs/benchmarks.md b/docs/benchmarks.md index 9c7a9283..631bb703 100644 --- a/docs/benchmarks.md +++ b/docs/benchmarks.md @@ -418,6 +418,18 @@ Not measured yet: no `bench-electron-*.json` summary has a Nebula timing (see th **Caveats, as measured.** +* **Stale: the six summaries predate the band-bottom changes (status of + 2026-10-11, #366).** All six `bench-electron-*.json` digests were made on + 2026-10-08 at `0942eb2`, before the band-bottom inelastic tables (#241), the + above-Fermi-level rate refinement (#339) and the loss-window correction + (#342) reached `main`. Those changes alter the inelastic tables, so the + table-build times and the work per history, and hence the electrons/s, of + the rows above no longer describe the current engine. They were not + regenerated because the cstool clone (`CSTOOL_SRC`) that + `bench_electron.py` needs for the matched inputs was not available on the + host that prepared #366; no result has been invented. The determinism + statement is also from that build. The "Still to measure" run below + replaces all six together from one current build. * **Not a quiet machine.** The lindhard rows were taken on 2026-10-08 at `0942eb2` (rustc 1.97.1, release profile with `lto = "fat"`, `codegen-units = 1`, matched inputs from cstool `0c739eb`). The host is a shared