lidavidm commented on pull request #10813: URL: https://github.com/apache/arrow/pull/10813#issuecomment-887521774
Ah, and for min_max, the scalar kernel becomes much faster if it calls std::max instead of std::fmax: ``` ---------------------------------------------------------------------------------------------------- Benchmark Time CPU Iterations UserCounters... ---------------------------------------------------------------------------------------------------- MinMaxKernelDouble/1048576/10000 192 us 192 us 7333 bytes_per_second=5.09496G/s null_percent=0.01 size=1048.58k MinMaxKernelDoubleAggregate/1048576/10000 347 us 347 us 3754 bytes_per_second=2.81357G/s null_percent=0.01 size=1048.58k MinMaxKernelInt64/1048576/10000 74.8 us 74.8 us 19546 bytes_per_second=13.0498G/s null_percent=0.01 size=1048.58k MinMaxKernelInt64Aggregate/1048576/10000 301 us 301 us 4941 bytes_per_second=3.24722G/s null_percent=0.01 size=1048.58k ``` So in all cases the hash aggregate kernel is considerably slower. -- 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]
