alamb commented on issue #20601: URL: https://github.com/apache/datafusion/issues/20601#issuecomment-3986917766
I am now confident I have a reasonable ClickBench testing setup and can reproduce the If you want to check it out, here is the result: [result-53.zip](https://github.com/user-attachments/files/25694276/result-53.zip) (unzip and open index.html) Some findings: ## DataFusion 47 results are reproducible: <img width="1680" height="267" alt="Image" src="https://github.com/user-attachments/assets/3ded8f03-d142-480a-bc6a-9d3991682e3d"; /> ## DataFusion vs DuckDB Here is what the results for DataFusion 53 will look like compared to DuckDB: <img width="663" height="995" alt="Image" src="https://github.com/user-attachments/assets/d5ccde8f-e945-49a8-8dca-9588fd9b4fe0"; /> The most obvious problem is Q23, our old friend https://github.com/apache/datafusion/blob/a5f490e4c31125b8126105385c6c1120cb2fa2de/benchmarks/queries/clickbench/queries/q23.sql#L4 <img width="841" height="285" alt="Image" src="https://github.com/user-attachments/assets/5d2bbb58-f357-43fb-931e-de6790590db6"; /> This will be resolved by turning on filter pushdown (w/ dynamic filters - https://github.com/apache/datafusion/issues/3463 The next slowest is Q29 https://github.com/apache/datafusion/blob/a5f490e4c31125b8126105385c6c1120cb2fa2de/benchmarks/queries/clickbench/queries/q29.sql#L4 <img width="559" height="154" alt="Image" src="https://github.com/user-attachments/assets/c0e4fdb1-ee60-4beb-ae90-b2bdc1913e9a"; /> Where I think we will just have to hold our nose and implement the ClickBench special optimization 💩 - https://github.com/apache/datafusion/issues/15524 I am not going to spend a lot more time reviewing results until we get [morsel driven scan](https://github.com/apache/datafusion/pull/20481) from @Dandandan over the line -- 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]
