Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Hack for GPU ID config #5086

Merged
merged 13 commits into from
Dec 7, 2019
Merged

Hack for GPU ID config #5086

merged 13 commits into from
Dec 7, 2019

Conversation

trivialfis
Copy link
Member

@trivialfis trivialfis commented Dec 4, 2019

Continues #5072 . The JSON PR works out everything, but being too huge. So continue the band-aid for rapids' release. @RAMitchell @JohnZed

  • Define prediction cache in GBTree.
  • Initialise prediction cache on first iteration.
  • Save gpu_id into binary model when CUDA is enabled.

src/learner.cc Outdated Show resolved Hide resolved
@trivialfis
Copy link
Member Author

I might implement a state machine for this configuration in next PR.

@trivialfis
Copy link
Member Author

@RAMitchell I will revert the changes in Learner if I can't get JSON in soon enough.

@trivialfis
Copy link
Member Author

Seems on Windows pipeline, if I stop the previous run, the next run will fail at getting sources.

@trivialfis trivialfis changed the title [WIP] Hack for GPU ID config Hack for GPU ID config Dec 5, 2019
@trivialfis
Copy link
Member Author

trivialfis commented Dec 5, 2019

Back ported the changes in prediction cache from JSON PR. Added a small Python test.

@trivialfis trivialfis merged commit 608ebbe into dmlc:master Dec 7, 2019
@trivialfis trivialfis deleted the gpu-id-config branch December 23, 2019 12:06
@lock lock bot locked as resolved and limited conversation to collaborators Mar 22, 2020
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants