HuaHuaY commented on PR #50749: URL: https://github.com/apache/arrow/pull/50749#issuecomment-5191842973
There are significant improvements in some benchmarks, such as `BM_ReadBinaryColumn` and `BM_ReadBinaryViewColumn` at `cpp/src/parquet/arrow/reader_writer_benchmark.cc`. However, there are also some performance regressions, maybe forcing the inlining of destructors affected the instruction layout. machine | z-score | z < -3 | z > 3 | Not significant -- | -- | -- | -- | -- m5 / Skylake | 4704 | 202 | 500 | 4002 c6a / Zen 3 | 4614 | 464 | 563 | 3587 t4g / Neoverse N1 | 4487 | 276 | 249 | 3962 Mac / M1 | 4467 | 29 | 18 | 4420 total | 18272 | 971（5.31%） | 1330（7.28%） | 15971（87.41%） I use AI to summarize the results. @pitrou Please take a look and help analyze whether these benchmark results are acceptable. -- 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]
