jorgecarleitao commented on pull request #9329: URL: https://github.com/apache/arrow/pull/9329#issuecomment-767737664
> I wouldn't have expected > 20% changes for bigger arrays (512/1024 elements) in the kernels only based on removing / not creating or cloning the Arc. Any idea why that could be the case? I remember that we had a similar issue some time ago where we were using `data` instead of `data_ref`, and that was causing a significant performance difference. My general approach here is to write simpler code in the hope that it helps the compiler. In practice, I humbly make these offerings to the gods of LLVM and pray for a sign. ---------------------------------------------------------------- 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]
