Skip to content

Add RMM patch to update include paths and allocation interface#110

Merged
hcho3 merged 4 commits intorapidsai:mainfrom
bdice:rmm-25.12-updates
Nov 25, 2025
Merged

Add RMM patch to update include paths and allocation interface#110
hcho3 merged 4 commits intorapidsai:mainfrom
bdice:rmm-25.12-updates

Conversation

@bdice
Copy link
Copy Markdown

@bdice bdice commented Nov 21, 2025

This makes xgboost compatible with changes in RMM 25.12 for headers and allocation interfaces.

This also rebuilds with CCCL 3.1.3 (in progress). Depends on NVIDIA/cccl#6727 and rapidsai/rapids-cmake#944, followed by a fresh nightly build of RMM 25.12.

Copy link
Copy Markdown
Member

@trivialfis trivialfis left a comment

Choose a reason for hiding this comment

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

The patch looks good to me.

@hcho3
Copy link
Copy Markdown

hcho3 commented Nov 24, 2025

The windows job is failing as it's picking up CCCL 3.1.2, not CCCL 3.1.3

@jakirkham
Copy link
Copy Markdown
Member

That's probably because CCCL 3.1.3 hadn't been packaged yet. It now is: conda-forge/cccl-feedstock#41

@bdice
Copy link
Copy Markdown
Author

bdice commented Nov 25, 2025

@jakirkham @hcho3 @trivialfis CI is now passing. Should we merge this with the R package disabled, so that we can unblock cuML's testing?

@hcho3 hcho3 mentioned this pull request Nov 25, 2025
@hcho3 hcho3 merged commit af693a9 into rapidsai:main Nov 25, 2025
14 of 16 checks passed
@hcho3
Copy link
Copy Markdown

hcho3 commented Nov 25, 2025

Merging this for now to unblock cuML testing. I created the follow-up issue at #111

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants