Hattonuri commented on PR #39403: URL: https://github.com/apache/arrow/pull/39403#issuecomment-1873399763
About benchmarks Tested my program in which parquet library take ~60% of time. Data is stored on ram fs to exclude any disk problems. First is with my PR, second is what it was We can see that on average we get -1sec. which is 2% performance increase. And 2% / 0.6 ~ 3.3% performance increase in parquet library dstasenko@bench-prod15:~$ for i in `seq 5`; do echo $i; time ./parquet_playground ; done 1 real 0m46.382s user 0m46.185s sys 0m0.180s 2 real 0m46.416s user 0m46.120s sys 0m0.276s 3 real 0m46.711s user 0m46.460s sys 0m0.232s 4 real 0m46.406s user 0m46.185s sys 0m0.200s 5 real 0m46.369s user 0m46.150s sys 0m0.200s dstasenko@bench-prod15:~$ for i in `seq 5`; do echo $i; time ./parquet_playground2 ; done 1 real 0m47.161s user 0m46.967s sys 0m0.176s 2 real 0m47.344s user 0m47.083s sys 0m0.244s 3 real 0m47.233s user 0m47.009s sys 0m0.204s 4 real 0m47.300s user 0m47.060s sys 0m0.240s 5 real 0m48.072s user 0m47.109s sys 0m0.944s -- 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]
