Skip to content

Use cuda::stream_ref for compute APIs - #23649

Merged
vyasr merged 5 commits into
NVIDIA:mainfrom
vyasr:codex/cuda-stream-ref-batch-3
Aug 14, 2026
Merged

Use cuda::stream_ref for compute APIs#23649
vyasr merged 5 commits into
NVIDIA:mainfrom
vyasr:codex/cuda-stream-ref-batch-3

Conversation

@vyasr

@vyasr vyasr commented Aug 13, 2026

Copy link
Copy Markdown
Contributor

Description

This third batch migrates compute libcudf joins, groupby, reductions, rolling, sorting, transform, unary, and binary operation APIs and tests from rmm::cuda_stream_view to cuda::stream_ref. It also includes small follow-up documentation fixes from the previous stream-ref batch.

Contributes to #23636

Checklist

  • I am familiar with the Contributing Guidelines.
  • New or existing tests cover these changes.
  • The documentation is up to date with these changes.

@vyasr
vyasr requested a review from a team as a code owner August 13, 2026 17:09
@vyasr
vyasr requested review from abigalekim and lamarrr August 13, 2026 17:09
@vyasr vyasr added 3 - Ready for Review Ready for review by team improvement Improvement / enhancement to an existing function libcudf Affects libcudf (C++/CUDA) code. non-breaking Non-breaking change labels Aug 13, 2026 — with ChatGPT Codex Connector
Comment thread cpp/tests/join/join_tests.cpp Outdated
@@ -2431,7 +2432,7 @@ TEST_F(JoinTest, HashJoinInnerMatchContext)

auto const host_match_counts =
cudf::detail::make_host_vector_async(*match_context._match_counts, stream);
stream.synchronize();
stream.wait();

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please use .sync() instead. xref: #23653

@vyasr vyasr Aug 14, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done in 6eba12f, thanks for the reminder.

@nirandaperera nirandaperera left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@nirandaperera

nirandaperera commented Aug 14, 2026

Copy link
Copy Markdown
Contributor

@vyasr @bdice Is this applicable to this PR as well? (replacing cuda/stream_ref with cuda/stream)

@vyasr

vyasr commented Aug 14, 2026

Copy link
Copy Markdown
Contributor Author

Oh boy, didn't realize that. Yes it is. Since CI is basically passing on this PR and hasn't run at all on Bradley's, I'll merge this PR then expand the scope of his to also cover the changes in this PR.

@vyasr

vyasr commented Aug 14, 2026

Copy link
Copy Markdown
Contributor Author

/merge

@vyasr
vyasr merged commit 5b6960d into NVIDIA:main Aug 14, 2026
136 of 140 checks passed
@vyasr
vyasr deleted the codex/cuda-stream-ref-batch-3 branch August 14, 2026 20:32
cursor Bot pushed a commit that referenced this pull request Aug 14, 2026
Rebase onto main and switch the additional includes introduced by
the compute-API stream_ref migration.

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
cursor Bot pushed a commit that referenced this pull request Aug 14, 2026
This file already included <cuda/stream>; the stream_ref rename
from #23649 left a second identical include that clang-format rejects.

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
rapids-bot Bot pushed a commit that referenced this pull request Aug 15, 2026
CCCL deprecated the `<cuda/stream_ref>` header; `cuda::stream_ref` is now provided by `<cuda/stream>`. This PR switches every `#include <cuda/stream_ref>` to `#include <cuda/stream>` so libcudf builds against newer CCCL.

Rebased onto main after #23649 (`Use cuda::stream_ref for compute APIs`) so the additional includes from that migration are covered as well. Also drops a duplicate `<cuda/stream>` include in `top_k.cu` that the rename introduced.

The `cuda::stream_ref` type is unchanged. This matches the same change in RMM (rapidsai/rmm#2513) and was observed in rapids-cmake CI: https://github.com/rapidsai/rapids-cmake/actions/runs/31745925779/job/94600794610?pr=1073

Authors:
  - Bradley Dice (https://github.com/bdice)
  - Cursor Agent (https://github.com/cursoragent)

Approvers:
  - Vyas Ramasubramani (https://github.com/vyasr)

URL: #23664
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

3 - Ready for Review Ready for review by team improvement Improvement / enhancement to an existing function libcudf Affects libcudf (C++/CUDA) code. non-breaking Non-breaking change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants