Repository navigation
Update ClickBench benchmarks with DataFusion 45.0.0 (When Published) #14246
Description
Is your feature request related to a problem or challenge?
- Follow on to Update ClickBench benchmarks with DataFusion
44.0.0#13983 - Requires Release DataFusion
45.0.0#14008
Describe the solution you'd like
Once DataFusion 45.0.0 is released, It would be great to update ClickBench https://benchmark.clickhouse.com/ with the newest version
ClickBench is a benchmark heavy on filtering and aggregation that we have used as an optimization target for the last several releases.
Describe alternatives you've considered
- Here is the PR that @pmcgleenon made for DataFusion 43.0.0: updated for datafusion release 43.0.0 ClickHouse/ClickBench#251
Additional context
Improvements I know of:
- @Dandandan 's work to improve parquet UTF8 Validation speed: Speed up Parquet utf8 validation arrow-rs#6667
Activity
Dandandan commentedon Jan 23, 2025
Would be nice to get this in as well #13681
Rachelint commentedon Jan 23, 2025
I added fuzzy tests for it in my local, and found some failed cases for
Decimaltoday ...I am fixing it, and working for benchmark, I think it can be finished today or tomorrow.
Dandandan commentedon Jan 23, 2025
Thank you very much @Rachelint
alamb commentedon Jan 23, 2025
I agree (though I don't think it will help ClickBench as clickbench doesn't have a median aggregate)
Added to release checklist on
45.0.0#14008Rachelint commentedon Jan 26, 2025
#13681 is ready for review
pmcgleenon commentedon Feb 8, 2025
Here are the initial ClickBench results for datafusion
45.0.0clickbench-latest 3.html.zip
If it looks reasonable, I can create a ClickBench PR to update the results
alamb commentedon Feb 8, 2025
It does look reasonable to me. Thank you @pmcgleenon 🙏
pmcgleenon commentedon Feb 8, 2025
Makes sense 👍
I've created a PR ClickHouse/ClickBench#304
pmcgleenon commentedon Feb 10, 2025
FYI the Clickbench PR has been merged and the latest Datafusion
45.0.0results have been published on the site https://benchmark.clickhouse.com/alamb commentedon Feb 10, 2025
Thank you so much @pmcgleenon ❤
Dandandan commentedon Feb 10, 2025
Nice! Looks we have some more competition now from DuckDB: results
alamb commentedon Feb 10, 2025
alamb commentedon Feb 10, 2025
Filed a ticket for running this on 46