AdamGS commented on issue #21080: URL: https://github.com/apache/datafusion/issues/21080#issuecomment-4510522263
@alamb I did some digging into this performance regression, here is what I've found: 1. It doesn't reproduce in DataFusion's clickbench benchmark, because the files we use in Vortex went through some pre-processing and they are different (different encoding, different row-group chunking). 1. There's a real regression in some of the stats code paths (I think its a combination of the NDV code and some changes around min/max for strings), I'll open issues/PRs once I have a better handle of exactly where it is, but the higher number of row groups we have in our setup makes it worse. -- 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]
