AMD GPU CSCS CI

[luraess]

Whats the purpose of this PR?

• Bug fix
• New feature
• Documentation update
• Other, please explain

Describe it in more detail below:
This PR adds support for AMD GPU CI pipeline on MI300 GPU from the Beverin runner at CSCS.

Checklist

• The PR title is descriptive and starts with the appropriate tag: [BUGFIX], [ADDITION], [DOC], etc.
• New tests (either assessing the correct behaviour of new internal functions or the correctness of a miniapps or benchmarks) were added, or old tests were updated
• Affected miniapps have also been updated
• The new feature was added in a way that does not break public API
• New documentation related to the new feature was added
• The new code follows the contributor guidelines, in particular the Runic Style

[luraess]

Note that this is WIP. I did so far only check we can access the runner and fire up AMDGPU on it. I would need to figure out a few more steps to enable the CI and replicate the GH200 setup. Also we need a patch bump in JP for JuliaGeodynamics/JustPIC.jl#284

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[luraess]

I'll need to finalise this one before merging as for now I just test that AMDGPU installs and display version info.

[aelligp]

@luraess any progress on this PR?

[luraess]

I was about to ask the same^^ Well, my plan was to look into it indeed. Could you see if there are still some AMDGPU blocker in JR. IIRC fixes in JuliaGPU/AMDGPU.jl#1033 could make it pass now. Maybe one could give it on shot (cc @albert-de-montserrat )

[aelligp]

@luraess I just re-enabled AMDGPU buildkite CI in #516 , lets see if it passes already and if not what is failing

[luraess]

Thanks. Actually, I could try to come up with the CSCS CI yml asap so we could further test there to not bloat the Buildkite queue which struggles quickly for AMDGPU given we have only 2 runners.

[aelligp]

One thing I'm not sure on CSCS/Beverin is how to deal with the IGG MPI config in the yml. that would need to be tested

[luraess]

Some test fails, and the MPI related ones as well. Could one skip MPI tests for now?