perf: batch contiguous extend calls in array_replace#22119
Merged
Conversation
github-actions
Bot
added
sqllogictest
Member
|
This patch optimizes cases when the element to replace is sparse. Maybe for the opposite scenario where elements to retain are sparse can also benefit from a similar optimization. Like only materialize an |
Member
|
Just mention it, we can explore this in a follow up PR |
Contributor
|
Thanks everyone! |
Contributor
Author
gstvg
pushed a commit
to gstvg/arrow-datafusion
that referenced
this pull request
May 14, 2026
## Which issue does this PR close? <!-- We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes apache#123` indicates that this PR will close issue apache#123. --> - Closes #. ## Rationale for this change <!-- Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed. Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes. --> array_replace / array_replace_n / array_replace_all currently call `MutableArrayData::extend` once per non-matching element. This pr batches consecutive non-matching elements into a single extend call, reducing call count. ## What changes are included in this PR? <!-- There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR. --> - Track contiguous non-matching runs and flush them as one extend call. - Add Criterion benchmark and SLT edge-case tests for n = 0 / -1 / > match_count. ### Benchmarks ``` group main perf ----- ---- ---- array_replace_all_int64/replace/list size: 10, num_rows: 4000 1.06 1149.7±23.40µs ? ?/sec 1.00 1080.3±71.22µs ? ?/sec array_replace_all_int64/replace/list size: 100, num_rows: 10000 1.70 11.2±0.19ms ? ?/sec 1.00 6.6±0.36ms ? ?/sec array_replace_all_int64/replace/list size: 500, num_rows: 10000 1.94 44.2±0.71ms ? ?/sec 1.00 22.8±0.48ms ? ?/sec array_replace_all_int64_nested/replace/list size: 10, num_rows: 4000 1.03 7.4±0.05ms ? ?/sec 1.00 7.2±0.26ms ? ?/sec array_replace_all_int64_nested/replace/list size: 100, num_rows: 3000 1.14 42.1±0.57ms ? ?/sec 1.00 36.9±0.22ms ? ?/sec array_replace_all_int64_nested/replace/list size: 300, num_rows: 1500 1.14 60.5±2.57ms ? ?/sec 1.00 52.8±0.60ms ? ?/sec array_replace_boolean/replace/list size: 10, num_rows: 4000 1.04 1087.5±25.31µs ? ?/sec 1.00 1042.4±18.85µs ? ?/sec array_replace_boolean/replace/list size: 100, num_rows: 10000 1.20 4.2±0.13ms ? ?/sec 1.00 3.5±0.06ms ? ?/sec array_replace_boolean/replace/list size: 500, num_rows: 10000 1.11 7.0±0.28ms ? ?/sec 1.00 6.3±0.06ms ? ?/sec array_replace_fixed_size_binary/replace/list size: 10, num_rows: 4000 1.03 1187.3±13.63µs ? ?/sec 1.00 1155.9±26.39µs ? ?/sec array_replace_fixed_size_binary/replace/list size: 100, num_rows: 10000 1.01 7.0±0.08ms ? ?/sec 1.00 6.9±0.67ms ? ?/sec array_replace_fixed_size_binary/replace/list size: 500, num_rows: 10000 1.00 23.6±0.61ms ? ?/sec 1.01 23.8±0.92ms ? ?/sec array_replace_int64/replace/list size: 10, num_rows: 4000 1.07 1071.8±25.30µs ? ?/sec 1.00 1003.3±24.68µs ? ?/sec array_replace_int64/replace/list size: 100, num_rows: 10000 1.16 4.0±0.07ms ? ?/sec 1.00 3.4±0.09ms ? ?/sec array_replace_int64/replace/list size: 500, num_rows: 10000 1.10 7.2±0.31ms ? ?/sec 1.00 6.6±0.34ms ? ?/sec array_replace_int64_nested/replace/list size: 10, num_rows: 4000 1.02 7.2±0.07ms ? ?/sec 1.00 7.0±0.21ms ? ?/sec array_replace_int64_nested/replace/list size: 100, num_rows: 3000 1.01 36.1±0.38ms ? ?/sec 1.00 35.8±1.62ms ? ?/sec array_replace_int64_nested/replace/list size: 300, num_rows: 1500 1.00 51.2±0.48ms ? ?/sec 1.00 51.1±0.67ms ? ?/sec array_replace_n_int64/replace/list size: 10, num_rows: 4000 1.09 1119.2±24.61µs ? ?/sec 1.00 1022.4±20.53µs ? ?/sec array_replace_n_int64/replace/list size: 100, num_rows: 10000 1.45 7.2±0.22ms ? ?/sec 1.00 5.0±0.16ms ? ?/sec array_replace_n_int64/replace/list size: 500, num_rows: 10000 1.68 25.3±0.38ms ? ?/sec 1.00 15.1±1.32ms ? ?/sec array_replace_n_int64_nested/replace/list size: 10, num_rows: 4000 1.03 7.4±0.08ms ? ?/sec 1.00 7.1±0.07ms ? ?/sec array_replace_n_int64_nested/replace/list size: 100, num_rows: 3000 1.05 38.5±0.36ms ? ?/sec 1.00 36.5±1.38ms ? ?/sec array_replace_n_int64_nested/replace/list size: 300, num_rows: 1500 1.04 55.0±0.41ms ? ?/sec 1.00 52.6±1.99ms ? ?/sec array_replace_strings/replace/list size: 10, num_rows: 4000 1.07 1443.5±18.35µs ? ?/sec 1.00 1348.6±41.07µs ? ?/sec array_replace_strings/replace/list size: 100, num_rows: 10000 1.09 10.9±0.17ms ? ?/sec 1.00 9.9±0.17ms ? ?/sec array_replace_strings/replace/list size: 500, num_rows: 10000 1.00 36.2±0.82ms ? ?/sec 1.04 37.5±0.91ms ? ?/sec ``` ## Are these changes tested? Yes — existing SLT tests and new edge-case coverage. <!-- We typically require tests for all PRs in order to: 1. Prevent the code from being accidentally broken by subsequent changes 2. Serve as another way to document the expected behavior of the code If tests are not included in your PR, please explain why (for example, are they covered by existing tests)? --> ## Are there any user-facing changes? No. <!-- If there are user-facing changes then we may require documentation to be updated before approving the PR. --> <!-- If there are any breaking changes to public APIs, please add the `api change` label. -->
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
3 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.
Which issue does this PR close?
Rationale for this change
array_replace / array_replace_n / array_replace_all currently call
MutableArrayData::extendonce per non-matching element. This pr batches consecutive non-matching elements into a single extend call, reducing call count.What changes are included in this PR?
Benchmarks
Are these changes tested?
Yes — existing SLT tests and new edge-case coverage.
Are there any user-facing changes?
No.