Introduce Benches To Measure Performance Impact of Changes to the Scanner

[0xNeshi]

Resources:

• Rust Performance Book: https://nnethercote.github.io/perf-book/benchmarking.html
• Stylus smart contracts (inspiration): https://github.com/OpenZeppelin/rust-contracts-stylus/tree/main/benches

[yug49]

GM @0xNeshi,

I have analyzed these resources you shared. Among all the tools I read on the Rust Performance Book you shared, I think using criterion.rs for this project would be best. Unlike the Stylus benches, which are specialized for measuring EVM Gas costs, this project needs something that can measure execution time and throughput for async Rust code. Criterion handles async runtimes natively and provides the statistical rigor needed here.

In this benchmark system, I would also like to integrate bencher into the CI pipeline, to track performance regressions, every time a code is pushed into this repo. So you can see if a PR slows things down.

What do think about this plan? Open to any suggestions that would make it more efficient.

Also, if no one has started working on this, can you please assign this to me? I would love to work on this system.

[0xNeshi]

Hey @yug49 , handling this would be really helpful, as this is a higher priority item for us!

Your plan sounds good, feel free to start working on it.
Ping me if you need to clarify anything 👍

[yug49]

Sure @0xNeshi
Thanks for allowing me to work on this :)

[0xNeshi]

A couple of guidelines:

1. Benches should run on-demand (manual dispatch)
2. Benches are necessary for historic mode and latest events mode
   • it's uncertain what exactly we'd be benchmarking for other modes, but if you have ideas, we're open to suggestions
3. Benches should be heavy load tests that measure the time to fetch all the expected events
   • you can use this load test as inspiration.