cloud-fan commented on pull request #27013: URL: https://github.com/apache/spark/pull/27013#issuecomment-705324814
Seems like we don't support `subexpressionElimination` in whole-stage-codegen (at least not in `ProjectExec.doConsume`), and thus introduce perf regression. I'm a bit hesitant to revert this, as it doesn't fix the fundamental issue that whole-stage-codegen can be slower because of the lack of `subexpressionElimination`. @viirya how hard it is to support `subexpressionElimination` in whole-stage-codegen? IIRC you tried to do it before. ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
