MNT use new threadpoolctl API (global threadpool controller) #21206
Conversation
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
I will have a look now. |
ogrisel
approved these changes
Oct 1, 2021
jjerphan
approved these changes
Oct 1, 2021
|
threadpoolctl 3.0.0 has been released. This PR is ready for review |
|
To avoid merging code that isn't used or tested, could you please make |
|
Since there is a performance benefit, what do you think of updating the required version for threadpoolctl to 3.0 in scikit-learn 1.1? |
glemaitre
pushed a commit
to glemaitre/scikit-learn
that referenced
this pull request
Oct 23, 2021
glemaitre
pushed a commit
that referenced
this pull request
Oct 25, 2021
samronsin
pushed a commit
to samronsin/scikit-learn
that referenced
this pull request
Nov 30, 2021
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Outdated suggestions cannot be applied.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.
The new API of threadpoolctl allows to not look for the shared libraries at each call to
threadpool_(limits/info). The idea is that in scikit-learn the BLAS and OpenMP libraries are loaded during the import and won't change during the lifetime of a program using sklearn. It means that we can reuse their handlers for all subsequent call to threadpoolctl.The goal is to reduce some unnecessary overhead, which can be as high as check_array in some extreme situations like in #20254 when running neighbors prediction on a single test sample.
@jjerphan could you try the new threadpoolctl api in your PR, via the fixes introduced in this PR ?
It's already available in the master branch of https://github.com/joblib/threadpoolctl
You should now import threadpool_limits from utils._threadpoolctl_fixes.
This PR is in draft state. Its current purpose is to check that the new api solves the perf issues in sklearn. Then, if it does, we'll need to release a new version of threadpoolctl before considering merging this.