[mt] Implement partitioning for GPU.#11789
Merged
Merged
Conversation
work on build hist. notes. work on evaluation. build all nodes. inputs. getter. alloc. proto. apply. disable. Work on high-level tests. lint. cleanup. test policy. lazy gen. Fix scan. Cleanup. Cleanup single split.
Contributor
There was a problem hiding this comment.
Pull Request Overview
This PR implements GPU support for multi-target training with vector leaves. The key changes enable multi-target tree building on GPU by extending the histogram-based tree construction algorithm to handle multiple targets simultaneously.
Key changes:
- Removed the GPU restriction check for vector leaf training in
gbtree.cc - Implemented multi-target histogram building and split evaluation on GPU with batch processing for multiple nodes
- Refactored test code to share common multi-target test logic between CPU and GPU tests
- Added
GoLeftWrapperOpwrapper to thecuda_implnamespace for consistent row partitioning
Reviewed Changes
Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| src/gbm/gbtree.cc | Removed GPU restriction for vector leaf training |
| src/tree/updater_gpu_hist.cuh | Implemented multi-target histogram maker with node sum tracking and batch split evaluation |
| src/tree/updater_gpu_hist.cu | Moved GoLeftWrapperOp to namespace, code formatting cleanup |
| src/tree/gpu_hist/evaluate_splits.cuh | Made parent_sum const-qualified for multi-target split inputs |
| src/tree/gpu_hist/multi_evaluate_splits.cuh | Added batch split evaluation and node sum tracking methods |
| src/tree/gpu_hist/multi_evaluate_splits.cu | Refactored to support batch processing of multiple nodes, fixed gradient index calculation |
| python-package/xgboost/testing/multi_target.py | Created shared test helpers for multi-target training |
| tests/python/test_multi_target.py | Refactored to use shared test helpers |
| tests/python-gpu/test_gpu_multi_target.py | Added GPU multi-target tests using shared helpers |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Contributor
There was a problem hiding this comment.
Pull Request Overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated no new comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
trivialfis
changed the title
trivialfis
marked this pull request as ready for review
November 3, 2025 07:17
Member
Author
|
cc @rongou |
rongou
approved these changes
Nov 3, 2025
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
XGBoost can now grow a full vector-leaf tree using a GPU. However, the implementation is still minimal and has low performance.
I would like to explore some algorithmic optimizations before adding any more complexity to the code.
ref #9043