Skip to content

fix(CI-05A): separate oracle setup from worker timings - #755

Merged
ellainng merged 7 commits into
mainfrom
fix/CI-05A-worker-benchmark
Nov 2, 2026
Merged

ellainng merged 7 commits into
mainfrom
fix/CI-05A-worker-benchmark

Conversation

@ellainng

@ellainng ellainng commented Nov 2, 2026 •

Copy link
Copy Markdown
Member

A fresh reference cache would charge oracle generation only to the serial sample in the worker benchmark. Validate/warm the oracle once before all timed browser samples and record reference_setup separately, so each pool size measures equivalent browser work.

Validation: initially empty cache, complete 39-case class corpus at 1/2/4/8 real Chromium workers, identical ordered results/resource observations and a successful JSON report. All 39 oracle cases passed; the 1.030s cold setup is recorded separately. No runtime, CI matrix or deadline changes.

#753 is merged. Closes #754.

Verification: all 16 regression jobs pass in CI run 41425599611. The affected 39-case, four-pool real-browser benchmark passes again after parent reconciliation; its oracle setup is recorded separately and the focused two-file change is retained.

Integrated-main verification: eeb0348ebdc7fff24f6261a8face882cc8703538; main wheel/installed-CLI/source-startup smoke passes. Both local source-profile startup checks pass, and main is clean and synchronized. All five batch tasks are complete; no release or deployment accompanies these CI changes.

@ellainng
ellainng changed the base branch from perf/CI-07-measured-shards to main November 2, 2026 10:17
@ellainng
ellainng merged commit eeb0348 into main Nov 2, 2026
@ellainng
ellainng deleted the fix/CI-05A-worker-benchmark branch November 2, 2026 10:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

CI-05A: Exclude oracle setup from worker comparisons

1 participant