adriangb commented on PR #19639: URL: https://github.com/apache/datafusion/pull/19639#issuecomment-3716389030
I've pushed a commit which has `filter_effectiveness_threshold = 1.0` and `pushdown_filters = true`. This should have the effect of moving filter evaluation into `ParquetOpener` but evaluate them like `FilterExec` does (by reading the columns all together with the projected columns and then applying a mask to the RecordBatch). If this is significantly slower than `main` I think it demonstrates that the issue comes from different parallelism / batch sizes, not necessarily from the I/O pattern change (at least for TPCH running on SSDs; the story may be different for ClickBench on S3). -- 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]
