mapleFU commented on PR #35731: URL: https://github.com/apache/arrow/pull/35731#issuecomment-1564390354
I use 16K as batch size now, here are benchmark under MacOS, Release(O2): ``` --------------------------------------------------------------------------------------------- Benchmark Time CPU Iterations UserCounters... --------------------------------------------------------------------------------------------- BM_ComputeHash<Int32Type> 131529 ns 131272 ns 5351 items_per_second=124.809M/s BM_ComputeHash<Int64Type> 127332 ns 127222 ns 5498 items_per_second=128.783M/s BM_ComputeHash<FloatType> 136299 ns 136063 ns 5176 items_per_second=120.414M/s BM_ComputeHash<DoubleType> 130264 ns 129839 ns 5419 items_per_second=126.187M/s BM_ComputeHash<ByteArrayType> 141495 ns 141402 ns 4932 items_per_second=115.868M/s BM_ComputeHash<FLBAType> 135159 ns 135144 ns 5181 items_per_second=121.233M/s BM_ComputeHash<Int96Type> 114941 ns 114874 ns 6118 items_per_second=142.626M/s BM_BatchComputeHash<Int32Type> 52456 ns 52429 ns 13406 items_per_second=312.497M/s BM_BatchComputeHash<Int64Type> 49498 ns 49484 ns 14178 items_per_second=331.094M/s BM_BatchComputeHash<FloatType> 58427 ns 58413 ns 11917 items_per_second=280.486M/s BM_BatchComputeHash<DoubleType> 55505 ns 55502 ns 12662 items_per_second=295.195M/s BM_BatchComputeHash<ByteArrayType> 111223 ns 111172 ns 6284 items_per_second=147.375M/s BM_BatchComputeHash<FLBAType> 106343 ns 106319 ns 6621 items_per_second=154.103M/s BM_BatchComputeHash<Int96Type> 73442 ns 73427 ns 9546 items_per_second=223.134M/s BM_InsertHash 22066188 ns 22038160 ns 100 items_per_second=743.438k/s BM_BatchInsertHash 21561561 ns 21540280 ns 100 items_per_second=760.621k/s BM_FindExistsHash 436261 ns 436161 ns 1594 items_per_second=2.29273k/s BM_FindNotExistsHash 497617 ns 497551 ns 1421 items_per_second=2.00984k/s ``` -- 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]
