Abhisheklearn12 commented on PR #25771: URL: https://github.com/apache/datafusion/pull/25771#issuecomment-5959714708
thanks @adriangb, i pushed the `all_false` fix: the `AND` chain is now walked in place instead of first being flattened into a list, so nothing is allocated before the first conjunct runs. results are unchanged. locally(i run locally) (x86, pinned core, 3 interleaved rounds), `all_false` went from 104.8 ns to 64.5 ns (`main`: 102.8 ns), and the other `short_circuit` and `conjunction` cases didn't change. one exception: the `cast` cases looked ~9% slower in a normal build, but that's code alignment (same instruction count, and arrow's hot comparison loop is byte-for-byte identical, just at a different address). with `-C llvm-args=-align-loops=64` on both builds the gap went away. so, the gap was smaller on my machine than on yours, so your arm run is the real check. could you please rerun `binary_op` -- 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]
