Skip to content

ci: run the suite and the examples on the Prescott kernel - #94

Merged
SMI-Lab-Inha merged 1 commit into
devfrom
ci/roundoff-kernel-job
Nov 7, 2026
Merged

SMI-Lab-Inha merged 1 commit into
devfrom
ci/roundoff-kernel-job

Conversation

@SMI-Lab-Inha

Copy link
Copy Markdown
Owner

Summary

A Linux job, round-off kernel (conda gfortran, ubuntu-latest, Prescott), runs the full CTest on the Prescott OpenBLAS kernel, then runs every shipped example deck and fails on a fallback note where a deck should converge cleanly. A test that passes on one kernel and fails on another is a bug.

  • tests/check_example_fallbacks.py lists offending decks and works on Linux and Windows; example_fallback_scan tests the scan itself.
  • The notes it looks for are the ones the solver really prints, including "residual floor above" and "resolving sub-step did not converge".
  • The job reads OpenBLAS's own Core: line and fails if the requested kernel did not take effect. The conda-forge package the CI uses is a DYNAMIC_ARCH build that honours OPENBLAS_CORETYPE.
  • It is not a required check.

Checks

  • All 54 decks are clean on the default kernel and on Prescott, every exit 0.
  • CTest 360 of 360 on both kernels on Windows.
  • This PR is the job's first run on Linux.

The CHANGELOG entry lands with the release batch.

A result that depends on the BLAS kernel passes on the runners' own AVX2
and AVX-512 kernels. Add a Linux Release job, separate from the required
ones, that runs the full CTest with OPENBLAS_CORETYPE=Prescott and then
every shipped example deck through tests/check_example_fallbacks.py.

The scan fails on a non-zero exit, a timeout and a fallback note, and lists
the offending decks. It matches the wording the solver prints for an
equilibrium accepted above the requested tolerance ("residual floor above"),
which does not contain "looser tolerance".

OPENBLAS_CORETYPE is ignored by an OpenBLAS without dynamic architecture
support and an unknown name falls back to detection, so the job first has
OpenBLAS name its kernel (OPENBLAS_VERBOSE=2) and stops on any other kernel
or none. example_fallback_scan proves the scan fails where it should.

Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
@SMI-Lab-Inha

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 7, 2026 •

Copy link
Copy Markdown
Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-07T13:39:47.650086Z b2a397a Manual request
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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@read-the-docs-community

Copy link
Copy Markdown

Documentation build overview

📚 CableDyn | 🛠️ Build #35719281 | 📁 Comparing b2a397a against latest (3d04008)

  🔍 Preview build  

2 files changed
± changelog.html
± development.html

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. 🎉

Reviewed commit: b2a397a12a

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".

@SMI-Lab-Inha
SMI-Lab-Inha merged commit df5606c into dev Nov 7, 2026
22 checks passed
@SMI-Lab-Inha
SMI-Lab-Inha deleted the ci/roundoff-kernel-job branch November 7, 2026 15:52
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