-
Notifications
You must be signed in to change notification settings - Fork 187
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
p256 point multiplication got ~25% slower in 0.12 #882
Comments
The biggest change was probably switching to the generic curve arithmetic implementation in the |
I didn't have much luck bisecting so I wrote a script that just benchmarks every commit
where
My results from this seem to implicate especially cea8f60, across two runs there is clear jump there
and
|
Nice sleuthing! This makes me want to set up something like https://github.com/bencherdev/bencher to watch for these sorts of regressions. I'll have to see if I can spot how this impacted codegen. |
I opened #885 which might address this issue |
I think this is addressed now? |
We just noticed in some internal benchmarks that
p256
performance for point multiplication and point addition both became notably slower in the 0.12 release. This is replicated by the benchmarks inp256
as well, at least for point multiplication (seems point addition is not benchmarked):0.11.1
Current
master
Both run on the same x86-64 Linux laptop, otherwise idle, Rust 1.65.
We looked at the release notes but don't see any reference to a change that suggests a performance slowdown, I also reviewed the diff and don't see any obvious culprits. Wanted to open an issue to check if this performance change was expected, and if so if there is any way to improve the situation to reach parity with 0.11 performance.
The text was updated successfully, but these errors were encountered: