Repository navigation
docs(perf): whole-app before/after for #1475 (PM3) - #1487
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e45a09a544
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| | same, reopen | 1,589 (179 ms) | 8 (12 ms) | | ||
| | `threads:changed` from a rename, per push | 2,680,470 B; 3.5 ms send | 1,341 B; 0.02 ms send; 0.15 ms encoder diff | | ||
|
|
||
| Every `threads:changed` in these windows (run start, run end, rename) went out as a row patch. A full list only goes out on thread create, fork or reorder. |
There was a problem hiding this comment.
On the first threads:changed event after each main-process launch, createThreadListEncoder() has no previous list and therefore sends { seq, threads: next }, so an ordinary first run-status update can still transmit the full ≈2.68 MB list. The row-patch result is thus steady-state behavior, not every status-change push as this sentence and the headline imply; document or measure the initial full push separately.
Useful? React with 👍 / 👎.
End-to-end measurement of #1478–#1486. It compares
a1c696aa, main just before the first fix, with8b4bcc71, the same tree asnightly-202610060729-8b4bcc71. Both were measured in one session on the same frozen copy of today's data, using the PM1 and PM2 scenarios.Headlines
threads:summariescold: 1,669 ms → 8 ms.threads:changed: 2.68 MB → 1.3 KB.files:resolveon a big-thread open: 614 → 4 IPCs.The report also lists what did not move: the big-thread switch is only −8%, soak RSS still climbs, and the main-heap LRU caps by count, not bytes.
Leak check: clean across 33 launches. The only ref diffs were commits by a live agent thread in another repo, each attributed by reflog.
Only the report commit is cherry-picked here. The worker's branch also carried two unrelated commits from a stale local HEAD.
🤖 Generated with Claude Code