Dandandan commented on PR #21886: URL: https://github.com/apache/datafusion/pull/21886#issuecomment-4334353948
Even if it doesn't improve perf measurable, I think we can take the changes. I think aggregation is very sensitive to memory / cache usage, so the more we can do to reduce the state the better and at some point it will pay off as aggregation state moves some level up (RAM - > L3, L3 -> L2, etc.) I think there is more low hanging fruit in this area * usize -> u32 group ids if possible * memorizing u64 hash not always needed * reducing double storage for `u128` views -- 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]
