Closed Boshen closed 1 year ago
How does the compile time compare to compiling without codspeed (using "plain" criterion)?
Have you tried caching release
artifacts?
We're doing it for the self-benchmarks of this repository, and it helps a lot. We're building all our benches in ~6secs. For example, in this run: https://github.com/CodSpeedHQ/codspeed-rust/actions/runs/6160795381/job/16718554517
We use https://github.com/moonrepo/setup-rust but using any other tools it should work if you cache the release target. https://github.com/CodSpeedHQ/codspeed-rust/blob/1621b664db0e47c8a63b55b9cde64e1f7e9cdf9a/.github/workflows/ci.yml#L32-L34
I'm using https://github.com/Swatinem/rust-cache due to the 10G GitHub cache limit.
I'll look into faster compilation methods on my end first. Thank you.
How does the compile time compare to compiling without codspeed (using "plain" criterion)?
I quick experimentation showed the same compilation duration.
For my project, https://github.com/web-infra-dev/oxc/actions/runs/6095539763/job/16539311128, it is currently taking 4 minutes to compile 5 benchmarks. I'd like to explore ways to speed things up a little bit because I am planning to add more benchmarks.
You can give me things to try or tell me the things you have tried so I don't waste too much time on this.