Add parallel distributed algorithms for segmented copy

[arpittkhandelwal]

Description

This PR extends HPX's distributed algorithms library by implementing parallel, distributed-aware segmented_copy algorithm for hpx::partitioned_vector.

Previously, applying these algorithms to a partitioned vector would fall back to sequential or non-segmented executions, which resulted in suboptimal performance and cross-node data transfer bottlenecks. This implementation directly maps the execution to the underlying segments (localities) using asynchronous dataflow execution and localized dispatching.

Key Changes:

• Implementation: Added the cross-locality dispatch logic in libs/full/segmented_algorithms/include/hpx/parallel/segmented_algorithms/copy.hpp.
• CPO Registration: Registered the segmented tag_invoke overloads for both hpx::copy_t and hpx::copy_if_t within the hpx::segmented namespace.
• Header Integration: Hooked the algorithms into hpx/parallel/segmented_algorithm.hpp and exposed them globally via hpx/include/parallel_copy.hpp.
• Testing: Expanded partitioned_vector_copy.cpp to validate cross-node data movement, multi-partition behavior, and sequential/parallel policy integration for both standard and _async variants.

Testing:

• Successfully compiled all test targets with ninja tests.unit.modules.segmented_algorithms.partitioned_vector_copy.
• Tested across both sequenced (hpx::execution::seq) and parallel (hpx::execution::par) execution policies.

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}

[Copilot]

Pull request overview

This PR adds segmented (distributed-aware) implementations of copy and copy_if intended to execute across hpx::partitioned_vector segments/localities, and wires them into the segmented algorithms umbrella headers and public include entry point.

Changes:

• Added a new segmented algorithms header implementing tag_invoke overloads for hpx::copy_t and hpx::copy_if_t.
• Integrated the new header into hpx/parallel/segmented_algorithm.hpp and hpx/include/parallel_copy.hpp.
• Expanded the partitioned_vector_copy.cpp unit test to cover copy_if (sync/async, seq/par policies).

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 5 comments.

File	Description
libs/full/segmented_algorithms/tests/unit/partitioned_vector_copy.cpp	Adds new copy_if tests for partitioned_vector (sync + task-based async).
libs/full/segmented_algorithms/include/hpx/parallel/segmented_algorithms/copy.hpp	New segmented implementations and CPO registrations for copy/copy_if.
libs/full/segmented_algorithms/include/hpx/parallel/segmented_algorithm.hpp	Includes the new segmented copy.hpp in the umbrella segmented algorithm header.
libs/full/include/include/hpx/include/parallel_copy.hpp	Exposes the segmented copy/copy_if implementation via the public include header.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 5 comments.

[hkaiser]

Implementing copy_if the way you propose will expose extremely bad performance. I'd rather not have an implementation at all than one that is slow and the user can't improve things. May I suggest removing the copy_if from this PR?

[arpittkhandelwal]

Implementing copy_if the way you propose will expose extremely bad performance. I'd rather not have an implementation at all than one that is slow and the user can't improve things. May I suggest removing the copy_if from this PR?

I agree a slow sequential fallback would be worse than nothing here. I've removed copy_if from this PR so we can just focus on getting segmented_copy merged cleanly

[hkaiser]

The LLM got you (again) ;-)

/libs/full/segmented_algorithms/include/hpx/parallel/segmented_algorithms/copy.hpp:
    Non-ASCII: line 244