Improve kd tree performance during initialization #18
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
|
This is the artificial benchmark dataset that triggers the problem: Up to a million spots are organized in a perfect circle (see screenshot). Without this PR Mastodon needs 15min to open the dataset. With the fix in this PR the dataset is opened in < 1 min. |
|
@tpietzsch look at this! |
tinevez
approved these changes
Oct 17, 2024
|
Yeah, we had the same problem in imglib2 a while ago imglib/imglib2#333 and it was fixed tor imglib2 However, I never got around to fixing the imglib2 kthElement implementation (it's not used by the revised KDTree anymore). @maarzt if you have time to fix |
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Outdated suggestions cannot be applied.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.
Mastodon opens sometimes very slowly on artificial datasets. Profiling with Java Visual VM showed that
KDTreeinitialization takes very long. More precisely calls to the methodKDTree.kthElement(...)take up 90% of CPU time.The method
kthElementaims to partially sort a list. The implementation is very efficient with a complexity of O(N) on randomized list. But strategy for selecting the pivot element is very inefficient when applied to an already sorted list, such that the complexity increases to O(N²). Or when applied to a list that contains several copies of the same constant value. This PR fixing the implementation to work efficient for these special cases.Unit tests are added to ensure that the new method works as intended.