Optimize SlidingTimeWindowMovingAverages sumBuckets#4943
Merged
joschi merged 1 commit intorelease/5.0.xfrom Sep 15, 2025
Merged
Optimize SlidingTimeWindowMovingAverages sumBuckets#4943joschi merged 1 commit intorelease/5.0.xfrom
joschi merged 1 commit intorelease/5.0.xfrom
Conversation
`SlidingTimeWindowMovingAverages` `sumBuckets()` method could be optimized to perform indexed list access and remove allocations as it currently allocates a `LongAdder` as well as one or two streams. If one is using the 1, 5, or 15 minute rates on hot paths, these can be unnecessary expensive allocations as well as less optimized computations. We can avoid the allocations completely and accumulate the sum directly to a `long` via optimized direct indexed list access. [MovingAverageBenchmarks](https://github.com/palantir/tritium/blob/davids/OptimizedSlidingTimeWindowMovingAverages/tritium-jmh/src/jmh/java/com/palantir/tritium/microbenchmarks/MovingAverageBenchmarks.java) demonstrates the difference in implementations in both execution time (22x faster) and allocations: ``` Benchmark (recordings) (type) Mode Cnt Score Error Units MovingAverageBenchmarks.getM1Rate 10 SlidingTimeWindowMovingAverages avgt 20 1364.969 ± 12.040 ns/op MovingAverageBenchmarks.getM1Rate:gc.alloc.rate.norm 10 SlidingTimeWindowMovingAverages avgt 20 688.037 ± 0.001 B/op MovingAverageBenchmarks.getM1Rate 10 OptimizedSlidingTimeWindowMovingAverages avgt 20 59.182 ± 0.343 ns/op MovingAverageBenchmarks.getM1Rate:gc.alloc.rate.norm 10 OptimizedSlidingTimeWindowMovingAverages avgt 20 0.002 ± 0.001 B/op MovingAverageBenchmarks.getM1Rate 1000 SlidingTimeWindowMovingAverages avgt 20 1401.864 ± 107.134 ns/op MovingAverageBenchmarks.getM1Rate:gc.alloc.rate.norm 1000 SlidingTimeWindowMovingAverages avgt 20 688.038 ± 0.003 B/op MovingAverageBenchmarks.getM1Rate 1000 OptimizedSlidingTimeWindowMovingAverages avgt 20 61.157 ± 1.393 ns/op MovingAverageBenchmarks.getM1Rate:gc.alloc.rate.norm 1000 OptimizedSlidingTimeWindowMovingAverages avgt 20 0.002 ± 0.001 B/op ```
|
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
2 participants
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.
SlidingTimeWindowMovingAveragessumBuckets()method could be optimized to perform indexed list access and remove allocations as it currently allocates aLongAdderas well as one or two streams. If one is using the 1, 5, or 15 minute rates on hot paths, these can be unnecessary expensive allocations as well as less optimized computations.We can avoid the allocations completely and accumulate the sum directly to a
longvia optimized direct indexed list access.MovingAverageBenchmarks demonstrates the difference in implementations in both execution time (22x faster) and allocations: