Update nvForest integration to use CUDA streams by chyunsu3 · Pull Request #8601 · NVIDIA/cuml

chyunsu3 · 2026-09-10T02:03:22Z

Updates cuML’s nvForest integration to use CUDA streams following rapidsai/nvforest#219.

Passes streams directly when exporting RandomForest and IsolationForest models to nvForest.
Updates C++ RF tests to reuse the same stream for model import and inference.
Forwards cuML’s static Treelite setting when building nvForest from source.

coderabbitai · 2026-09-10T02:08:21Z

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Last reviewed commit: 213ce94
Latest commit, not reviewed: 64ca6e3

Use the following commands to manage reviews:

@coderabbitai resume to resume automatic reviews.
@coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

▶️ Resume reviews
🔍 Trigger review

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 0c1189ac-45d1-4f00-95f4-b733a4910710

📥 Commits

Reviewing files that changed from the base of the PR and between 1a45939 and 213ce94.

📒 Files selected for processing (2)

ci/build_docs.sh
python/cuml/cuml/ensemble/isolation_forest.pyx

Included review availability: Your plan provides up to 12 included reviews per hour; 11 remain after this review.

📝 Summary

Summary by CodeRabbit

Bug Fixes
- Improved CUDA stream handling when exporting random forest and isolation forest models to NVForest for more reliable predictions and model operations.
Chores
- Updated automated build and test workflows to support pull-request versions of NVForest Conda packages and wheels.
- Refined dependency channel prioritization across C++, Python, documentation, notebook, and wheel workflows.
- Updated the NVForest dependency source and refreshed copyright year ranges.

Changes

CI scripts retrieve pull-request nvforest Conda packages and wheels. Build and test flows consume them. nvforest prediction and model conversion now pass explicit CUDA streams. The nvforest dependency uses a pinned fork commit.

nvforest integration

Layer / File(s)	Summary
Conda package selection and consumers `ci/use_conda_packages_from_prs.sh`, `ci/build_*.sh`, `ci/test_cpp.sh`, `ci/test_notebooks.sh`, `ci/test_python_common.sh`	The Conda helper retrieves pull-request artifacts and configures channels for builds, documentation, and tests.
Wheel selection and consumers `ci/use_wheels_from_prs.sh`, `ci/build_wheel_.sh`, `ci/test_wheel.sh`	The wheel helper retrieves pull-request wheels and writes pip constraints for wheel builds and tests.
Explicit CUDA stream usage `cpp/tests/sg/rf_test.cu`, `python/cuml/cuml/ensemble/*.pyx`, `cpp/cmake/thirdparty/get_nvforest.cmake`	Prediction tests and Python model conversion pass explicit CUDA streams to nvforest. The dependency configuration selects a pinned fork commit.

Priority: ⬇️ Low

Estimated code review effort: 3 (Moderate) | ~25 minutes

Change: Other

Suggested reviewers: betatim

Merge Risk: ⚪ Minimal · up to 213ce

The updated CI artifact selection and CUDA stream integration have no demonstrated actionable defect. The change is ready for normal validation.

Pre-merge checks: 3 · 2

Failed checks: 2 warnings

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 3 functions across 14 files. (1 skipped: 1…	Write docstrings for the functions missing them to satisfy the coverage threshold.
Description check		The description explains the main changes and implementation choices, but it does not link an issue with a GitHub closing keyword as required by the template.	Add an issue reference using a GitHub closing keyword, such as `Closes #1234`. Include validation details if applicable.

Passed checks: 3 passed

Check name	Status	Explanation
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.
Title check		The title clearly describes the primary change: updating nvForest integration to use CUDA streams.

Full details: Docstring Coverage

Explanation

Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 3 functions across 14 files. (1 skipped: 1 unsupported.)

Resolution

Write docstrings for the functions missing them to satisfy the coverage threshold.

Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

_{Comment @coderabbitai help to get the list of available commands.}

chyunsu3 · 2026-10-05T19:07:23Z

 export RAPIDS_PACKAGE_VERSION

 # populates `RATTLER_CHANNELS` array and `RATTLER_ARGS` array
+source ./ci/use_conda_packages_from_prs.sh


TODO: Remove this line after rapidsai/nvforest#219 is merged. Marking this PR as draft as the reminder.

chyunsu3 · 2026-10-16T01:39:00Z

/ok to test 81c38a2

csadorf · 2026-10-16T20:43:33Z

/ok to test aa05bfa

csadorf · 2026-10-16T21:20:50Z

/ok to test f1fef78

csadorf · 2026-10-16T21:23:17Z

Applied DO NOT MERGE to hold this until next week, when we can coordinate the landing with rapidsai/nvforest#219. Merging these independently could disrupt cuML CI. We should also remove the temporary PR-artifact plumbing and finalize the nvForest pin before merging.

chyunsu3 · 2026-10-19T19:59:55Z

/ok to test ba0f03b

**Companion PR: NVIDIA/cuml#8601. To minimize the disruption in the cuML CI, we should merge this PR together with NVIDIA/cuml#8601 . * Remove the use of `raft::handle_t` and `nvforest::handle_t` and use CUDA streams directly. * Create `predict()` overloads that takes in either a single CUDA stream or a pool of multiple CUDA streams. * Use `cuda::buffer` instead of `rmm::device_buffer`. Since `cuda::buffer` requires NVCC to build, we need to use PIMPL with type erasure. * Update all C++ tests. * Update the Python layer. The `handle` parameter is now deprecated and its use will trigger a `FutureWarning`. Users are directed to use `stream` instead. * Remove `cpp/cmake/thirdparty/get_rmm.cmake`. For now, we retain `cpp/cmake/thirdparty/get_raft.cmake`, since we need to wait one release cycle until we can remove the deprecated `handle` parameter. * Add test coverage for using stream with wrong device. After this PR, #193 will update `dependencies.yaml` to remove the RMM dependency and update the docs. History * This PR is part of #193 * This is a re-submission of #195 (which was previously reverted to unblock the cuML CI). * #195 originally targeted 26.10, whereas this PR targets 26.12. Authors: - Philip Hyunsu Cho (https://github.com/chyunsu3) - Simon Adorf (https://github.com/csadorf) Approvers: - Bradley Dice (https://github.com/bdice) - Simon Adorf (https://github.com/csadorf) URL: #219

…to pr-8601

csadorf · 2026-10-19T20:04:25Z

/ok to test 64ca6e3

jcrist · 2026-10-19T21:23:21Z

/merge

[DO NOT MERGE] Test rapidsai/nvforest#219

ae81310

chyunsu3 requested a review from a team as a code owner September 10, 2026 02:03

chyunsu3 requested a review from KyleFromNVIDIA September 10, 2026 02:03

github-actions Bot assigned chyunsu3 Sep 10, 2026

github-actions Bot added the ci label Sep 10, 2026

chyunsu3 added improvement Improvement / enhancement to an existing function non-breaking Non-breaking change labels Sep 10, 2026

Use stream arg instead of handle when using nvForest

1e7ddd0

chyunsu3 requested review from a team as code owners September 10, 2026 02:54

chyunsu3 requested a review from jcrist September 10, 2026 02:54

github-actions Bot added Cython / Python Cython or Python issue CUDA/C++ labels Sep 10, 2026

chyunsu3 added 3 commits September 10, 2026 06:52

Merge branch 'main' into update_nvforest

bc091a1

Update cpp/cmake/thirdparty/get_nvforest.cmake

02e0162

Fix tests

26ca86e

chyunsu3 requested a review from a team as a code owner September 11, 2026 02:23

github-actions Bot added the CMake label Sep 11, 2026

chyunsu3 added 8 commits September 10, 2026 19:25

Merge branch 'main' into update_nvforest

6494313

Merge branch 'main' into update_nvforest

1a45939

Merge branch 'main' into update_nvforest

213ce94

Merge branch 'main' into update_nvforest

e827b44

Pass CUML_USE_TREELITE_STATIC to nvForest

8c645d0

Adapt to cuda::stream_ref adaoption in RAFT

ef27049

Merge branch 'main' into update_nvforest

2a10756

Merge branch 'main' into update_nvforest

6b61217

chyunsu3 changed the title ~~[DO NOT MERGE] Test rapidsai/nvforest#219~~ Update nvForest usage according to rapidsai/nvforest#219 Oct 5, 2026

chyunsu3 mentioned this pull request Oct 5, 2026

Remove RMM dependency and prepare for RAFT removal, Part 1 rapidsai/nvforest#219

Merged

chyunsu3 marked this pull request as draft October 5, 2026 19:06

chyunsu3 commented Oct 5, 2026

View reviewed changes

chyunsu3 added 3 commits October 6, 2026 15:13

Merge branch 'main' into update_nvforest

9e91506

Merge remote-tracking branch 'origin/main' into update_nvforest

afa7717

Update nvforest

8625182

chyunsu3 marked this pull request as ready for review October 14, 2026 03:42

chyunsu3 marked this pull request as draft October 14, 2026 21:56

Merge branch 'main' into update_nvforest

81c38a2

Test nvForest integration against reviewed stream-pool fixes

aa05bfa

Refresh nvForest integration pin after main merge

f1fef78

csadorf added the DO NOT MERGE Hold off on merging; see PR for details label Oct 16, 2026

Merge branch 'main' into update_nvforest

ba0f03b

Remove temporary nvForest PR artifact overrides

71996d2

csadorf added 2 commits October 19, 2026 15:02

Merge remote-tracking branch 'origin/main' into pr-8601

2c898d3

Merge branch 'update_nvforest' of https://github.com/chyunsu3/cuml in…

64ca6e3

…to pr-8601

csadorf marked this pull request as ready for review October 19, 2026 20:20

csadorf removed request for a team and KyleFromNVIDIA October 19, 2026 20:20

csadorf approved these changes Oct 19, 2026

View reviewed changes

csadorf removed the DO NOT MERGE Hold off on merging; see PR for details label Oct 19, 2026

csadorf changed the title ~~Update nvForest usage according to rapidsai/nvforest#219~~ Update nvForest integration to use CUDA streams Oct 19, 2026

rapids-bot Bot merged commit 210dff7 into NVIDIA:main Oct 19, 2026
168 of 171 checks passed

chyunsu3 deleted the update_nvforest branch October 19, 2026 23:20

Navigation Menu

Repository navigation

Update nvForest integration to use CUDA streams - #8601

Update nvForest integration to use CUDA streams#8601
rapids-bot[bot] merged 23 commits into
NVIDIA:mainfrom
chyunsu3:update_nvforest

chyunsu3 commented Sep 10, 2026 •

edited by csadorf

Loading

Uh oh!

coderabbitai Bot commented Sep 10, 2026 •

edited

Loading

Reviews paused

Summary by CodeRabbit

Changes

Uh oh!

chyunsu3 Oct 5, 2026

Uh oh!

chyunsu3 commented Oct 16, 2026

Uh oh!

csadorf commented Oct 16, 2026

Uh oh!

csadorf commented Oct 16, 2026

Uh oh!

csadorf commented Oct 16, 2026

Uh oh!

chyunsu3 commented Oct 19, 2026

Uh oh!

csadorf commented Oct 19, 2026

Uh oh!

jcrist commented Oct 19, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

chyunsu3 commented Sep 10, 2026 • edited by csadorf Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot commented Sep 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Reviews paused

Summary by CodeRabbit

Changes

Uh oh!

chyunsu3 Oct 5, 2026

Choose a reason for hiding this comment

Uh oh!

chyunsu3 commented Oct 16, 2026

Uh oh!

csadorf commented Oct 16, 2026

Uh oh!

csadorf commented Oct 16, 2026

Uh oh!

csadorf commented Oct 16, 2026

Uh oh!

chyunsu3 commented Oct 19, 2026

Uh oh!

csadorf commented Oct 19, 2026

Uh oh!

jcrist commented Oct 19, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

chyunsu3 commented Sep 10, 2026 •

edited by csadorf

Loading

coderabbitai Bot commented Sep 10, 2026 •

edited

Loading