Just for clarity here, I'll eat my words (the benchmark was modified to make them comparable) and we are still discussing (I think this is another case of gRPC/Java's long standing issue with a small fixed backpressure threshold).
On Fri, Aug 26, 2022, at 07:43, David Li wrote: > I replied on Github, but in short, the two benchmarks aren't measuring > anything remotely near the same thing. They are not comparable at all. > > The Java version could use work to be on par with the C++ version. It > does report figures in a misleading way. > > On Fri, Aug 26, 2022, at 06:04, Игорь Скоков wrote: >> Hello, >> I created the issue https://github.com/apache/arrow/issues/13980 with a >> question about the performance difference of Arrow Flight C++ and Java. >> Will be glad to hear suggestions from you >> >> Thanks
