release the GIL in denoise_tv_bregman and denoise_bilateral #5400
Conversation
This file contains 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
measured approximately 6x better performance on a 10-core system with util.apply_parallel after this change
jni
approved these changes
May 17, 2021
|
Thank you @grlee77! |
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Outdated suggestions cannot be applied.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.
Description
I tried using
apply_parallelwith the variousdenoise_*functions inskimage.restoration. Most showed a substantial speedup, butdenoise_tv_bregmananddenoise_bilateralshowed worse performance when calling withapply_parallelthan without. This turns out to be because the GIL is not being released by these two functions.As in #5399, this PR just adds a
with nogilcontext block into each of these functions so the GIL will be released, allowing multithreading. I confirmed locally that this leads to the expected improvement.The somewhat large diff is misleading as nearly all line differences are just the change in indentation.
Timing
For example for
denoise_tv_bregmanon an input of shape (2048, 4096) it takes:normal call (single thread): 2.33 s
call with apply_parallel (prior to this PR, GIL not released): 3.2 s
call with apply_parallel (after this PR): 0.53 s
The result for
denoise_bilateralis very similar.