Retrigger v7 outputs for 5 cancelled folders (batch 2)#1559
Closed
ChrisRackauckas-Claude wants to merge 1 commit into
Closed
Retrigger v7 outputs for 5 cancelled folders (batch 2)#1559ChrisRackauckas-Claude wants to merge 1 commit into
ChrisRackauckas-Claude wants to merge 1 commit into
Conversation
Continuation of the v7-stack-output retrigger campaign begun in PR SciML#1558. These five folders also had their Manifest.toml regenerated under the v7 stack in PR SciML#1551 but were cancelled by the 24h queue-wait timeout in run 25374549824 without producing artifacts: - AutomaticDifferentiation - AutomaticDifferentiationSparse - NBodySimulator - MultiLanguage - NonlinearProblem This PR is a no-op apart from a comment line in each Project.toml — its sole purpose is to mark these folders as "changed" so detect-changed- benchmarks queues them for the matrix. With SciML#1556 (publish-gate relax) already merged, the successful folders' outputs will publish even if one or two in this batch fail. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
This was referenced May 15, 2026
Contributor
Author
|
Superseded — closing this 'trigger comment' PR in favor of clean per-folder PRs with real content changes (compat bumps, manifest regens, or actual benchmark fixes) instead of the hack |
ChrisRackauckas-Claude
pushed a commit
to ChrisRackauckas-Claude/SciMLBenchmarks.jl
that referenced
this pull request
May 22, 2026
amdci3-1 corrupts its workspace under rapid dispatch — when 2+ jobs arrive within seconds, all but one die with errors like: ##[error]Missing file at path: ..._runner_file_commands/set_output_... ##[error]Can't find action.yml under .../_actions/actions/checkout/v6 ##[error]The operation was canceled. Single jobs on the runner run cleanly; the bug is purely concurrency-on-the-same-runner. Reproduced 3 times now: 1. Original 25374549824 v7 run: 3 jobs cancelled-mid-checkout in seconds 2. 2026-05-16 07:58 UTC burst: 6 jobs from PRs SciML#1559/SciML#1560/SciML#1561 dispatched to amdci3-1, all but one corrupted (StiffODE — which had been cleanly running for 16h — also got cancelled in the cascade) 3. 2026-05-16 08:31 UTC burst: all 5 single-folder PRs SciML#1563-SciML#1567 dispatched simultaneously, all 5 corrupted within seconds Splitting into per-folder PRs (per Chris's direction) doesn't help when multiple PRs are opened together — cross-PR dispatch hits the same race. Fix: job-level concurrency with a constant group + cancel-in-progress:false makes every matrix entry across every Benchmarks workflow run queue behind any other in-flight matrix entry. detect-changes / manual-matrix / publish jobs are unaffected (they're separate GHA jobs). Adds max-parallel: 1 to the matrix as defense-in-depth — within a single run, matrix entries already serialize via the concurrency group, but max-parallel makes it explicit. Throughput cost: even with two healthy runners (amdci1-1 + amdci3-1), only one benchmark job runs at a time. Acceptable given the alternative is a ~60% job-loss rate. Can be relaxed once the runner bug is fixed — e.g. parameterize the concurrency group by matrix.runner to allow one job per runner. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
ChrisRackauckas-Claude
pushed a commit
to ChrisRackauckas-Claude/SciMLBenchmarks.jl
that referenced
this pull request
May 22, 2026
The NBodySimulator/liquid_argon.jmd weave job ran for ~47h on amdci1-1 under the v7 stack before the runner workspace was reaped mid-run (observed in PR SciML#1559, run 25912449328, job 76160856440). Chunk 5 (`run_benchmark!(symplectic, 10 τs, t=40)`) alone accounted for the bulk of the runtime. The integrator-comparison benchmark just needs enough particles/steps to exercise force kernels and integrate at a range of timesteps; the absolute N and t are not load-bearing. Reduce the workload in both .jmd files so the suite finishes in a sensible window: - liquid_argon.jmd: N=350 → 128, t=40.0 → 10.0 - liquid_argon_long.jmd: N=350 → 128, t=200.0 → 50.0 Net work reduction is roughly (350/128)² × (40/10) ≈ 30× for the short suite and ≈ 30× for the long suite. The hardcoded `c_symplectic` / `c_adaptive` cost ratios remain valid since they are per-step ratios normalised to VelocityVerlet, essentially N-independent. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
ChrisRackauckas-Claude
pushed a commit
to ChrisRackauckas-Claude/SciMLBenchmarks.jl
that referenced
this pull request
May 31, 2026
The NBodySimulator/liquid_argon.jmd weave job ran for ~47h on amdci1-1 under the v7 stack before the runner workspace was reaped mid-run (observed in PR SciML#1559, run 25912449328, job 76160856440). Chunk 5 (`run_benchmark!(symplectic, 10 τs, t=40)`) alone accounted for the bulk of the runtime. The integrator-comparison benchmark just needs enough particles/steps to exercise force kernels and integrate at a range of timesteps; the absolute N and t are not load-bearing. Reduce the workload in both .jmd files so the suite finishes in a sensible window: - liquid_argon.jmd: N=350 → 128, t=40.0 → 10.0 - liquid_argon_long.jmd: N=350 → 128, t=200.0 → 50.0 Net work reduction is roughly (350/128)² × (40/10) ≈ 30× for the short suite and ≈ 30× for the long suite. The hardcoded `c_symplectic` / `c_adaptive` cost ratios remain valid since they are per-step ratios normalised to VelocityVerlet, essentially N-independent. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Re-queues five more benchmark folders whose Manifests resolved cleanly under the v7 stack (#1551) but were cancelled by the 24h queue-wait timeout in run 25374549824:
benchmarks/AutomaticDifferentiationbenchmarks/AutomaticDifferentiationSparsebenchmarks/NBodySimulatorbenchmarks/MultiLanguagebenchmarks/NonlinearProblemBuilds on #1558 (batch 1). With #1556 already in master, partial successes will auto-publish.
Mechanism: one-line comment appended to each Project.toml so
detect-changed-benchmarks.shqueues them.Note for review
Please ignore until reviewed by @ChrisRackauckas.