Switch VecOfArrays for RecursiveArrayTools's VectorOfArray

[benegee]

Continuing #2491

[github-actions]

Review checklist

This checklist is meant to assist creators of PRs (to let them know what reviewers will typically look for) and reviewers (to guide them in a structured review process). Items do not need to be checked explicitly for a PR to be eligible for merging.

Purpose and scope

• The PR has a single goal that is clear from the PR title and/or description.
• All code changes represent a single set of modifications that logically belong together.
• No more than 500 lines of code are changed or there is no obvious way to split the PR into multiple PRs.

Code quality

• The code can be understood easily.
• Newly introduced names for variables etc. are self-descriptive and consistent with existing naming conventions.
• There are no redundancies that can be removed by simple modularization/refactoring.
• There are no leftover debug statements or commented code sections.
• The code adheres to our conventions and style guide, and to the Julia guidelines.

Documentation

• New functions and types are documented with a docstring or top-level comment.
• Relevant publications are referenced in docstrings (see example for formatting).
• Inline comments are used to document longer or unusual code sections.
• Comments describe intent ("why?") and not just functionality ("what?").
• If the PR introduces a significant change or new feature, it is documented in NEWS.md with its PR number.

Testing

• The PR passes all tests.
• New or modified lines of code are covered by tests.
• New or modified tests run in less then 10 seconds.

Performance

• There are no type instabilities or memory allocations in performance-critical parts.
• If the PR intent is to improve performance, before/after time measurements are posted in the PR.

Verification

• The correctness of the code was verified using appropriate tests.
• If new equations/methods are added, a convergence test has been run and the results
  are posted in the PR.

Created with ❤️ by the Trixi.jl community.

[benegee]

Still seems to work. CI failures are due to codecov issues.

[sloede]

I'm not sure I understand - you seem to have replaced some VecOfArray with VectorOfArray, but there is no additional using and nothing is removed. How is this possible 😉

[JoshuaLampert]

I guess this file can now be removed, right @benegee? We already import VectorOfArray from RecursiveArrayTools.jl, see

Trixi.jl/src/Trixi.jl

Line 81 in bc7858a

using RecursiveArrayTools: VectorOfArray

[benegee]

Yes, exactly. I failed to delete the file while merging.

[sloede]

As far as I can tell, the main difference is that now we have to wrap the outer vector of vectors in a VectorOfArray and then access the inner vector/array with val.u[i] instead of val[i], is that correct?

Have you verified (for at least one 3D run maybe) that there are no differences in performance and memory allocations (especially with multithreading)?

[DanielDoehring]

This looks a bit odd :D But I guess this is what we get from using VectorOfArray. The .u might be a bit confusing, though

[sloede]

Would it make sense to hide it? I.e., store the VectorOfArray in cache.__u_threaded and then cache.u_threaded = cache.__u_threaded.u?

[DanielDoehring]

I don't know, it is a bit unfortunate that both for us and VectorOfArray u plays such a central role

[benegee]

As far as I can tell, the main difference is that now we have to wrap the outer vector of vectors in a VectorOfArray and then access the inner vector/array with val.u[i] instead of val[i], is that correct?

Yes!

This is what happened:

• In Use Adapt.jl to change storage and element type #2212 VecOfArrays was introduced, which goes back to Lars' GPU prototype where it was required because a Vector{Vector} is not Adapt.jlable
• RecursiveArrayTools was not Adapt.jl compatible either at that point
• But this has changed in the meantime, so we can now use RecursiveArrayTools and remove VecOfArrays again.
• At that time, conventions in RecursiveArrayTools changed. That is where the .u comes from, see https://docs.sciml.ai/RecursiveArrayTools/stable/breaking_changes_v4/

Would it make sense to hide it? I.e., store the VectorOfArray in cache.__u_threaded and then cache.u_threaded = cache.__u_threaded.u?

Well, that could work!

Have you verified (for at least one 3D run maybe) that there are no differences in performance and memory allocations (especially with multithreading)?

No, I haven't. We would need to compare the old Vector{Vector} implementation with the new one.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.14%. Comparing base (295379b) to head (0f46cc1).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2952      +/-   ##
==========================================
+ Coverage   97.13%   97.14%   +0.01%     
==========================================
  Files         623      622       -1     
  Lines       48385    48367      -18     
==========================================
- Hits        46996    46985      -11     
+ Misses       1389     1382       -7     

Flag	Coverage Δ
unittests	97.14% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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