Skip to content

benchmarks: bump aca-model pin to c31b5dd#347

Closed
hmgaudecker wants to merge 1 commit into
feat/canonical-float-dtypefrom
benchmarks/bump-aca-model-pin
Closed

benchmarks: bump aca-model pin to c31b5dd#347
hmgaudecker wants to merge 1 commit into
feat/canonical-float-dtypefrom
benchmarks/bump-aca-model-pin

Conversation

@hmgaudecker
Copy link
Copy Markdown
Member

Summary

  • Single-line bump of the aca-model rev pinned by the benchmarks pixi feature
  • Picks up the aca-model consumption_unequiv rename + the regenerated benchmark snapshot (workarounds removed)

Test plan

  • pixi run -e benchmarks-cuda12 asv run --quick smoke-tests the bench harness against the new aca-model

🤖 Generated with Claude Code

@read-the-docs-community
Copy link
Copy Markdown

read-the-docs-community Bot commented May 10, 2026

@hmgaudecker hmgaudecker force-pushed the benchmarks/bump-aca-model-pin branch from 69c0c4b to 35f8cfa Compare May 10, 2026 17:24
@github-actions
Copy link
Copy Markdown

github-actions Bot commented May 10, 2026

Benchmark comparison (main → HEAD)

Comparing a4eca9bf (main) → e18791f8 (HEAD)

Benchmark Statistic before after Ratio Alert
aca-baseline execution time 46.946 s 27.599 s 0.59
peak GPU mem 671 MB 741 MB 1.10
compilation time 427.20 s 291.94 s 0.68
peak CPU mem 8.39 GB 12.15 GB 1.45
Mahler-Yum execution time 4.756 s 4.691 s 0.99
peak GPU mem 522 MB 529 MB 1.01
compilation time 16.46 s 14.61 s 0.89
peak CPU mem 1.68 GB 1.69 GB 1.01
Precautionary Savings - Solve execution time 45.5 ms 48.8 ms 1.07
peak GPU mem 101 MB 101 MB 1.00
compilation time 2.52 s 2.64 s 1.05
peak CPU mem 1.12 GB 1.13 GB 1.01
Precautionary Savings - Simulate execution time 119.6 ms 123.8 ms 1.04
peak GPU mem 340 MB 344 MB 1.01
compilation time 5.99 s 4.89 s 0.82
peak CPU mem 1.29 GB 1.31 GB 1.02
Precautionary Savings - Solve & Simulate execution time 136.7 ms 159.8 ms 1.17
peak GPU mem 577 MB 578 MB 1.00
compilation time 7.78 s 7.18 s 0.92
peak CPU mem 1.28 GB 1.29 GB 1.00
Precautionary Savings - Solve & Simulate (irreg) execution time 280.2 ms 285.6 ms 1.02
peak GPU mem 2.19 GB 2.19 GB 1.00
compilation time 8.59 s 7.31 s 0.85
peak CPU mem 1.33 GB 1.34 GB 1.01

Picks up the consumption_equiv_floor / consumption_unequiv_floor DAG
split, the consumption_weights / coefficients_rra plural rename,
the regenerated benchmark snapshot, and the ty-clean test fixes.
No other pylcm changes.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@hmgaudecker hmgaudecker force-pushed the benchmarks/bump-aca-model-pin branch from 35f8cfa to e18791f Compare May 10, 2026 18:12
@hmgaudecker
Copy link
Copy Markdown
Member Author

Subsumed by feat/canonical-float-dtype 5b49bea — the aca-model pin bump now lives directly on #345 alongside the dtype-filter promotion. Closing this PR + deleting the branch.

hmgaudecker added a commit that referenced this pull request May 11, 2026
The squash-merge of #340 onto main carried a small int-cast loop
inside `broadcast_to_template` that duplicates work already done by
`cast_params_to_canonical_dtypes` (the float-side reshuffle separated
broadcast and cast into two passes). Drop it.

Bump the `benchmarks` feature's aca-model rev to 9ac2043 so this
branch carries the same pin PR #347 was opened for; #347 can close.

Lockfile updated to track the merged pylcm HEAD and the new
aca-model rev.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@hmgaudecker hmgaudecker deleted the benchmarks/bump-aca-model-pin branch May 11, 2026 05:51
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.

1 participant