Technical debt changes in radix_sort.hpp

[oleksandr-pavlyk]

1. Add missing #include <array> needed since std::array is used several times in the code.

2. Made peer_helper class specializations members const, as well as marking peer_contribution method const.

   This allows to make peer_prefix_hlp class instance const as well.

3. Added a comment to why subgroup_ballot algorithm for peer helper is only applicable for sub-groups with sizes 32 and narrower.

---

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?
• If this PR is a work in progress, are you opening the PR as a draft?

[github-actions]

Deleted rendered PR docs from intelpython.github.com/dpctl, latest should be updated shortly. 🤞

[coveralls]

coverage: 87.671%. remained the same
when pulling 19a7e75 on radix-sort-technical-debt
into 631d4c3 on master.

[github-actions]

Array API standard conformance tests for dpctl=0.19.0dev0=py310hdf72452_287 ran successfully.
Passed: 895
Failed: 1
Skipped: 118

[ndgrigorian]

The atomic_fetch_or peer_prefix_algo seems to be entirely unused in the code.

Is it dead code, or should it be kept around?

[oleksandr-pavlyk]

The atomic_fetch_or peer_prefix_algo seems to be entirely unused in the code.

Is it dead code, or should it be kept around?

At present, it is indeed unused. It is useful to keep around for debugging purposes. I did check, locally, that the code compiles and works correctly if atomic_fetch_or is used instead of subgroup_ballot algorithm.

The other useful feature of atomic_fetch_or algorithm is that work-group size in reorder phase kernels need not be constrained to the size of the sub-group, although radix_sort_reorder_submit always sets it to be.

[ndgrigorian]

In that case LGTM!