Hello Arrow community, We (first at Ursa Computing, now Voltron Data) have been orchestrating Apache Arrow C++, Python, R, Java, and JavaScript benchmarks and logging results into Conbench [1] since May 2021 (huge shout out and thank you to Diana Clarke who was the driving force behind this with me). You might have already interacted with Arrow Benchmarks CI via `@ursabot please benchmark` comments or noticed ursabot's comments with benchmarks results (like [2]) on Apache Arrow pull requests. This data and interface has been invaluable in reviewing performance-related pull requests already.
In order to make it easy for other Arrow developers to provide machines to run benchmarks on, we have made the orchestration part of this project open source [3]. In particular, there is now a process for adding new benchmark machines [4] and public Buildkite pipelines to https://buildkite.com/apache-arrow. If you would like to add a new machine for running new benchmarks (I know there was interest to do this from the Rust folks a few months ago), you are welcome to follow the above process. Feel free to reach out to me directly (via Issues and Pull Requests on [3]) for any help you might need. -Elena [1]: https://conbench.ursa.dev/ [2]: https://github.com/apache/arrow/pull/12085#issuecomment-1008899897 [3]: https://github.com/ursacomputing/arrow-benchmarks-ci [4]: https://github.com/ursacomputing/arrow-benchmarks-ci/blob/main/docs/how-to-add-new-benchmark-machine.md
