gm zk farcasters, my team was struggling to compare different ZK frameworks so we made a benchmarks website with tons of info about different tradeoffs and nuances: https://zkbench.dev

would love feedback from folks in this channel!

building polybase & @payy • zk maxi • prev pm @ cruise, eng @ apple

Love to see up-to-date bench on underlying proving sys like halo2, plonky2 gnark etc. as well as more bench on folding <3

A dumb cryptographer. Building ZKP for clients and the mass. 

Also, is there a folding optimized sha256 impl? I know Nova kinda just use bellman. If breaking all iterations into steps and fold, might be fast as hell?

We are planning to add Halo2. Noir + Barretenberg is plonky, but we could test Barretenberg directly.

co-founder and cto @payy, using zk for zk.

I don't think you'd get a speed improvement by using folding directly. Recursion is always going to be slower than having it all in a single circuit. I guess technically with folding you could have multiple separate machines doing it in parallel, but seems like overkill.