WIP: fix(dev): a timed run's noise names the queue's jobs that ran beside it #2187

Draft
dyst wants to merge 1 commit from dyst/Longshot:fix/queue-noise-beside into main
Member

Summary

A timed run's history keeps how busy the rest of the PC was, so a noisy measurement shows. Two of those readings, the CPU pressure and the load, come from Linux alone and are always 0 on Windows. And the queue's own other jobs running beside a timed run were never noted. A light job starts beside a timing one. In two dev.py ab series for #2185, another session's runtime builds ran beside three runs and prepared shaders on the GPU: those runs gave 27.9, 29.1 and 41.1 fps against 65 to 72, with the noise all zeros.

The noise now names each other queue job active while the timed run ran ("jobs_beside": ["rt-pr1937 (light)"]), on every system. The run's line then says the PC was busy while it measured.

Read first: Noise and _keep_apart in scripts/devtools/jobqueue.py.

Testing

Tested: not run yet.

  • Checks: tests/test_job_queue.py has a case for the record's new field. Not run yet.

Still to test:

  • Windows, any GPU: a timing job with a light job of another name started beside it, expecting dev.py queue history to name that job under jobs_beside, and the run's line to say the PC was busy (by hand)

Measurements

Nothing to measure: what the queue records, not how a run goes.

Risks

A timed run with a job beside it now ends with the "busy while it measured" line; nothing is refused or moved. Reverting is enough.

Checklist

Ready for review

  • One topic, on current main; the title is a Conventional Commit.
  • Tested on my own PC, in the game for a change the game runs, as Testing says. Not run yet.
  • mise run test passes (any failure that also fails on main is named). Not run yet.
  • Measured against main, or Measurements says why there is nothing to measure.
  • No game files, build output or hand-edited generated files; third_party/ untouched; nothing GPL-incompatible.

When it applies

  • AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.

Ready to merge

  • Everything in Still to test has run, its results in Tested.
  • Tier 1 hardware it can affect is tested (docs/dev/standards.md, "Hardware tiers").
  • Approved in review, every comment answered.
  • CI passes on the head; on current main, measured again or a comment says why the numbers still hold.

🤖 Generated with Claude Code

## Summary A timed run's history keeps how busy the rest of the PC was, so a noisy measurement shows. Two of those readings, the CPU pressure and the load, come from Linux alone and are always 0 on Windows. And the queue's own other jobs running beside a timed run were never noted. A light job starts beside a timing one. In two `dev.py ab` series for #2185, another session's runtime builds ran beside three runs and prepared shaders on the GPU: those runs gave 27.9, 29.1 and 41.1 fps against 65 to 72, with the noise all zeros. The noise now names each other queue job active while the timed run ran (`"jobs_beside": ["rt-pr1937 (light)"]`), on every system. The run's line then says the PC was busy while it measured. Read first: `Noise` and `_keep_apart` in `scripts/devtools/jobqueue.py`. ## Testing **Tested:** not run yet. - **Checks:** `tests/test_job_queue.py` has a case for the record's new field. Not run yet. **Still to test:** - Windows, any GPU: a timing job with a light job of another name started beside it, expecting `dev.py queue history` to name that job under `jobs_beside`, and the run's line to say the PC was busy (by hand) ## Measurements Nothing to measure: what the queue records, not how a run goes. ## Risks A timed run with a job beside it now ends with the "busy while it measured" line; nothing is refused or moved. Reverting is enough. ## Checklist ### Ready for review - [x] One topic, on current `main`; the title is a Conventional Commit. - [ ] Tested on my own PC, in the game for a change the game runs, as Testing says. Not run yet. - [ ] `mise run test` passes (any failure that also fails on `main` is named). Not run yet. - [x] Measured against `main`, or Measurements says why there is nothing to measure. - [x] No game files, build output or hand-edited generated files; `third_party/` untouched; nothing GPL-incompatible. ### When it applies - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. ### Ready to merge - [ ] Everything in Still to test has run, its results in Tested. - [ ] Tier 1 hardware it can affect is tested (docs/dev/standards.md, "Hardware tiers"). - [ ] Approved in review, every comment answered. - [ ] CI passes on the head; on current `main`, measured again or a comment says why the numbers still hold. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
fix(dev): a timed run's noise names the queue's jobs that ran beside it
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 10s
CI / Website build and checks (pull_request) Failing after 23s
Website / Website on Cloudflare (pull_request_target) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 26s
PR description / Description follows the template (pull_request_target) Successful in 36s
Third-party parts / As their notes say (pull_request) Successful in 13s
CI / Workflow lint (actionlint) (pull_request) Successful in 44s
a77e22141f
A timed run's history kept the PC's CPU pressure and load, both read from Linux alone, and the builds outside the queue.
On Windows the first two are always 0, and the queue's own jobs beside a timed run were never noted: a light job starts
beside a timing one, and two series of dev.py ab had runs spoiled by another session's runtime builds (their shader
preparation took the GPU; 27.9 and 29.1 fps against 65 to 68) with the noise all zeros. The noise now names each other job
active while the timed run ran ("name (mode)"), on every system, and the run's line says the PC was busy when there was one.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 10s
CI / Website build and checks (pull_request) Failing after 23s
Website / Website on Cloudflare (pull_request_target) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 26s
PR description / Description follows the template (pull_request_target) Successful in 36s
Third-party parts / As their notes say (pull_request) Successful in 13s
CI / Workflow lint (actionlint) (pull_request) Successful in 44s
This pull request is marked as a work in progress.
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u 'fix/queue-noise-beside:dyst-fix/queue-noise-beside'
git switch 'dyst-fix/queue-noise-beside'
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
MagicBeanDev/Longshot!2187
No description provided.