perf: improve precompute performance by optimizing updates#1977
Merged
triceo merged 2 commits intoTimefoldAI:mainfrom Dec 12, 2025
Merged
perf: improve precompute performance by optimizing updates#1977triceo merged 2 commits intoTimefoldAI:mainfrom
triceo merged 2 commits intoTimefoldAI:mainfrom
Conversation
triceo
requested review from
Christopher-Chianelli and
Copilot
and removed request for
Copilot
Contributor
There was a problem hiding this comment.
Pull request overview
This PR optimizes the performance of precompute operations by eliminating intermediate data structures in the update path. The optimization removes the updateQueue from RecordAndReplayPropagator and delegates deduplication directly to StaticPropagationQueue, which already handles duplicate updates efficiently via tuple state checking.
Key Changes
- Removed the intermediate
updateQueuefield and its associated processing overhead - Modified
update()method to directly propagate updates to the tuple-level queue instead of collecting them at the object level - Simplified
propagateUpdates()to directly delegate to the propagation queue
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
RecordAndReplayPropagator.java |
Removed updateQueue field and refactored update handling to bypass object-level queuing and deduplication, directly calling propagationQueue.update() for each tuple |
PropagationQueue.java |
Added Javadoc clarifying that duplicate updates of the same item are deduplicated |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
core/src/main/java/ai/timefold/solver/core/impl/bavet/common/PropagationQueue.java
Outdated
Show resolved
Hide resolved
…ropagationQueue.java Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
|
Christopher-Chianelli
approved these changes
Dec 12, 2025
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
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.
In my benchmarks, I see speedups of 20-25 %.
This should significantly improve the viability of
precompute.Looking at the insert/retract logic, we probably over-designed this. The de-duplication is very heavy, for something that barely ever happens. But retracts and inserts also barely ever happen themselves, so this code doesn't show up in profiling at all and therefore I'm not touching it.