amoeba commented on PR #40647: URL: https://github.com/apache/arrow/pull/40647#issuecomment-2011218123
Is that as simple as adding `>ThreadRange(1, 32);` to the relevant benchmark's args? When I do that to the `AllocateDeallocate` benchmark (memory_pool_benchark.cc), I see about a 10x improvement at 32 threads when I comment out the entire body of `UpdateAllocatedBytes` and about half that if I just comment out what was added in Arrow 13 (https://github.com/apache/arrow/commit/ddfa8eed9b188fcc7b38767d1858c2588c588f05#diff-2111aac8ee579e238fb15d4380f2ea1e2f3e2830da939ec3f07c61dc68038d1f). I put the full results in a Gist: https://gist.github.com/amoeba/b95102829280dbe2b1f64e6c23a5f594. A [branch on my fork](https://github.com/amoeba/arrow/commits/exp/memory-pool-bench-threads/) shows the changes I made to the benchmark and code. -- 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]
