Skip to content

How to use a data structure that supports GPU in R? #11263

Closed as not planned
Closed as not planned

Description

@travis-leith

I am using the xgboost R package, with GPU enabled, running with a local GPU. I get a noticeable speedup compared to CPU only.

I also get the message "Falling back to prediction using DMatrix due to mismatched devices.". Same as mentioned here: #9791

The solutions mentioned in the other issue are useful for python. Is there anything I can do in R to better support the GPU?

Activity

  1. trivialfis commented on Feb 19, 2025

    @trivialfis
    Member

    I'm not aware of GPU based inputs for the R package yet. If you want XGBoost to support a specific R project (say, a GPU-enabled datatable implementation), please let us know.

  2. travis-leith commented on Feb 19, 2025

    @travis-leith
    Author

    Hmm, looking around, there appears not to be any such R package out there yet. I'll leave it up to you to decide if this issue should be closed or not.

  3. trivialfis commented on Feb 19, 2025

    @trivialfis
    Member

    Let's close this issue for now. We can open a new one for a more specific feature request when there's a GPU-based structure.

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions