Increase ByteViewMap block size to 2MB #11674
Merged
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 the last piece to get the initial StringView support, testing on my machine shows that it can increase the string-intensive ClickBench performance by 20%-200%. |
Dandandan
approved these changes
Jul 27, 2024
alamb
approved these changes
Jul 27, 2024
| @@ -149,7 +149,7 @@ where | |||
| output_type, | |||
| map: hashbrown::raw::RawTable::with_capacity(INITIAL_MAP_CAPACITY), | |||
| map_size: 0, | |||
| builder: GenericByteViewBuilder::new(), | |||
| builder: GenericByteViewBuilder::new().with_block_size(2 * 1024 * 1024), | |||
There was a problem hiding this comment.
After apache/arrow-rs#6136 is merged, will this be needed anymore?
alamb
added a commit
that referenced
this pull request
Jul 29, 2024
… some ClickBench queries (not on by default) (#11667) * Pin to pre-release version of arrow 52.2.0 * Update for deprecated method * Add a config to force using string view in benchmark (#11514) * add a knob to force string view in benchmark * fix sql logic test * update doc * fix ci * fix ci only test * Update benchmarks/src/util/options.rs Co-authored-by: Andrew Lamb <[email protected]> * Update datafusion/common/src/config.rs Co-authored-by: Andrew Lamb <[email protected]> * update tests --------- Co-authored-by: Andrew Lamb <[email protected]> * Add String view helper functions (#11517) * add functions * add tests for hash util * Add ArrowBytesViewMap and ArrowBytesViewSet (#11515) * Update `string-view` branch to arrow-rs main (#10966) * Pin to arrow main * Fix clippy with latest arrow * Uncomment test that needs new arrow-rs to work * Update datafusion-cli Cargo.lock * Update Cargo.lock * tapelo * merge * update cast * consistent dep * fix ci * add more tests * make doc happy * update new implementation * fix bug * avoid unused dep * update dep * update * fix cargo check * update doc * pick up the comments change again --------- Co-authored-by: Andrew Lamb <[email protected]> * Enable `GroupValueBytesView` for aggregation with StringView types (#11519) * add functions * Update `string-view` branch to arrow-rs main (#10966) * Pin to arrow main * Fix clippy with latest arrow * Uncomment test that needs new arrow-rs to work * Update datafusion-cli Cargo.lock * Update Cargo.lock * tapelo * merge * update cast * consistent dep * fix ci * avoid unused dep * update dep * update * fix cargo check * better group value view aggregation * update --------- Co-authored-by: Andrew Lamb <[email protected]> * Initial support for regex_replace on `StringViewArray` (#11556) * initial support for string view regex * update tests * Add support for Utf8View for date/temporal codepaths (#11518) * Add StringView support for date_part and make_date funcs * run cargo update in datafusion-cli * cargo fmt --------- Co-authored-by: Andrew Lamb <[email protected]> * GC `StringViewArray` in `CoalesceBatchesStream` (#11587) * gc string view when appropriate * make clippy happy * address comments * make doc happy * update style * Add comments and tests for gc_string_view_batch * better herustic * update test * Update datafusion/physical-plan/src/coalesce_batches.rs Co-authored-by: Andrew Lamb <[email protected]> --------- Co-authored-by: Andrew Lamb <[email protected]> * [Bug] fix bug in return type inference of `utf8_to_int_type` (#11662) * fix bug in return type inference * update doc * add tests --------- Co-authored-by: Andrew Lamb <[email protected]> * Fix clippy * Increase ByteViewMap block size to 2MB (#11674) * better default block size * fix related test * Change `--string-view` to only apply to parquet formats (#11663) * use inferenced schema, don't load schema again * move config to parquet-only * update * update * better format * format * update * Implement native support StringView for character length (#11676) * native support for character length * Update datafusion/functions/src/unicode/character_length.rs --------- Co-authored-by: Andrew Lamb <[email protected]> * Remove uneeded patches * cargo fmt --------- Co-authored-by: Xiangpeng Hao <[email protected]> Co-authored-by: Xiangpeng Hao <[email protected]> Co-authored-by: Andrew Duffy <[email protected]>
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.
Targets the string-view2 branch
Which issue does this PR close?
Closes #.
Rationale for this change
Increase the default block size from 8KB to 2MB, this significantly improves the aggregation performance with high cardinality.
More details see apache/arrow-rs#6094
What changes are included in this PR?
Are these changes tested?
Are there any user-facing changes?