Skip to content

Linear Solver Tests: Loosen convergence tolerance for single precision - #5319

Merged
WeiqunZhang merged 2 commits into
AMReX-Codes:developmentfrom
WeiqunZhang:curlcurl_tol
Apr 12, 2026
Merged

Linear Solver Tests: Loosen convergence tolerance for single precision#5319
WeiqunZhang merged 2 commits into
AMReX-Codes:developmentfrom
WeiqunZhang:curlcurl_tol

Conversation

@WeiqunZhang

Copy link
Copy Markdown
Member

This should solve the HPSF GitLab CI failures.

The tests started to fail after #5305, not because that PR introduced a bug, but because that PR fixed a CMake bug that prevented the tests from running.

A similar CMake bug is also fixed in this PR.

This should solve the HPSF GitLab CI failures.

The tests started to fail after AMReX-Codes#5305, not because that PR introduced a bug,
but because that PR fixed a CMake bug that prevented the tests from running.

A similar CMake bug is also fixed in this PR.
@WeiqunZhang
WeiqunZhang requested review from asalmgren and ax3l April 12, 2026 03:37
@WeiqunZhang

Copy link
Copy Markdown
Member Author

/run-hpsf-gitlab-ci

@github-actions

Copy link
Copy Markdown

GitLab CI has started at https://gitlab.spack.io/amrex/amrex/-/pipelines/1521836.

@amrex-gitlab-ci-reporter

Copy link
Copy Markdown

GitLab CI 1521836 finished with status: failed. See details at https://gitlab.spack.io/amrex/amrex/-/pipelines/1521836.

@WeiqunZhang WeiqunZhang changed the title CurlCurl Test: Loosen convergence tolerance for single precision Linear Solver Tests: Loosen convergence tolerance for single precision Apr 12, 2026
@WeiqunZhang

Copy link
Copy Markdown
Member Author

/run-hpsf-gitlab-ci

@github-actions

Copy link
Copy Markdown

GitLab CI has started at https://gitlab.spack.io/amrex/amrex/-/pipelines/1521862.

@amrex-gitlab-ci-reporter

Copy link
Copy Markdown

GitLab CI 1521862 finished with status: success. See details at https://gitlab.spack.io/amrex/amrex/-/pipelines/1521862.

@WeiqunZhang
WeiqunZhang merged commit cd7b6af into AMReX-Codes:development Apr 12, 2026
74 checks passed
@WeiqunZhang
WeiqunZhang deleted the curlcurl_tol branch April 12, 2026 18:38
WeiqunZhang added a commit that referenced this pull request Sep 10, 2026
Build in single precision with EB, linear solvers, AmrLevel, particles
and FFT, then run ctest and clang-tidy. Single precision was only
exercised by the GitLab H100 GPU runs, so solver breakage kept surfacing
late: #4785, #5319 and #5672. The job found a float-conversion bug in
the 2D nodal EB kernel and a NodeTensorLap tolerance that cannot be met
in float. 263 -Wfloat-conversion lines remain under Tests/, so that
warning is not an error in this job yet; cleanup follows.
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.

2 participants