gabotechs commented on issue #11680: URL: https://github.com/apache/datafusion/issues/11680#issuecomment-5104206424
I've been trying it some more, and ended up closing the PR as it did not prove any meaningful performance improvements, not even being smart and just apply hash reusing for non-primitive columns. Left some conclusion here: - https://github.com/apache/datafusion/pull/23720 At least with the public benchmarks, hashing occupies a very small fraction of the profiles, and even avoiding half the hashing operations in a query only shows some improvements that could perfectly fall within noise. My take is that this is a high complexity / low impact piece of work that could probably be worth exploring once other clearer performance optimizations have been addressed. -- 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]
