emkornfield commented on issue #37559: URL: https://github.com/apache/arrow/issues/37559#issuecomment-1712111390
Two answer some questions: > are we talking about BMI or BMI2 here? Those are two different things, BMI is far more widespread than BMI2. BMI2, `pext` and `pdep` I believe are what is used in the paper (I'd have to double check if there are other instructions used). > do we have to call intrinsics explicitly? are the corresponding operations exposed by https://github.com/xtensor-stack/xsimd and, if not, can we perhaps contribute them? Per above, I think we already have the necessary instructions modelled with dynamic dispatch in parquet (we might want o revisit settings for AMD CPUs, as I think later versions now support the instructions efficiently > perhaps some compiler optimizers can even produce those instructions automatically given the right code patterns? Maybe, this would be an area to look into. > Are we considering row-level filter evaluation or just pruning based on page stats? It sounds like both. -- 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]
