Benchmarking scripts

[laurentsimon]

We'd like to have benchmarking scripts to improve on the current shell https://github.com/sigstore/model-transparency/blob/main/model_signing/benchmarks/run.sh

We welcome contributions; this is a really important piece to be able to compare various signing options (serialization, hash + params) and be able to cut a v1 release.

@mihaimaruseac has written some python-based script we can work off.

@font FYI

[font]

@mihaimaruseac has written some python-based script we can work off.

@mihaimaruseac Are you able to share these?

[mihaimaruseac]

I'm debating between sharing the scripts as I used them for previous benchmarks or adapting them to the new API (under work) and having them then act as CI checks

[laurentsimon]

+1 on adapting them to the new APIs. Sharing the script may speed up adapting them to the new API, if someone's interested in doing the work

[spencerschrock]

We should make sure to benchmark Python 3.10 vs Python 3.11 / 3.12. I'm curious if the hashlib.file_digest polyfill (#300) will show a meaningful difference, as there were 1-2 optimizations in the upstream that weren't included.