jayzhan211 commented on issue #11680: URL: https://github.com/apache/datafusion/issues/11680#issuecomment-2262949557
I found I got performance boost for clickbench Q17 just by enforcing single mode for group by, interesting ```rust Comparing main and single-multi-groupby -------------------- Benchmark clickbench_1.json -------------------- ┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Query ┃ main ┃ single-multi-groupby ┃ Change ┃ ┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ QQuery 0 │ 2009.51ms │ 1435.19ms │ +1.40x faster │ │ QQuery 1 │ 7270.86ms │ 4033.52ms │ +1.80x faster │ └──────────────┴───────────┴──────────────────────┴───────────────┘ ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓ ┃ Benchmark Summary ┃ ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩ │ Total Time (main) │ 9280.37ms │ │ Total Time (single-multi-groupby) │ 5468.70ms │ │ Average Time (main) │ 4640.19ms │ │ Average Time (single-multi-groupby) │ 2734.35ms │ │ Queries Faster │ 2 │ │ Queries Slower │ 0 │ │ Queries with No Change │ 0 │ └─────────────────────────────────────┴───────────┘ ``` https://github.com/apache/datafusion/pull/11762 I think what I need to do is find a query that is currently slower in single mode, and find a way to optimize it like partial/final way in single execution node? 🤔 Does the result shows that what we really need is storing values in one large hash table + spilling? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
