We have just ran the verification of a pairing computation of arkworks using SP1 of
@SuccinctLabs. We did not have to make any modifications to the Arkworks code. I am testing ideas for
@alignedlayer. We’re trying different ideas to aggregate proofs and post them in Ethereum after a few hours.
Proof time was around 18 minutes, and proof size was 568 MB.
The advantages of SP1 are the fact that it is open-source, has relatively fast proof generation. The problem is that it has larger proof sizes.
On the other hand,
@RiscZero has continuations (which makes proof smaller and breaks down computations into smaller pieces that can be handled separately), and smaller proof sizes, but it is has some closed-source part and it seems slower in CPU (we didn’t run benchmarks).
We believe that the best solution will take approaches from both worlds.