Open WildCryptoFox opened 4 years ago
For fair comparison with optimized implementation for Bulletproofs, our protocols are implemented in Rust using the curve25519-dalek library for ECC operations [54] and compared with the January 2020 git version of Bulletproofs implementation in by Valence et al. [28], which is, to the best of our knowledge, one of the most optimized implementations for Bulletproofs.
They have an implementation based on dalek25519, and compare directly with this crate. I wonder whether they'll release their source code at some point. To what extent would the crate owners (@oleganza ?) be interested in having this in here? We might want a peer-reviewed paper first (or introduce a yoloyoloproofs
feature).
https://github.com/KZen-networks/bulletproofs not as fast as this library (4x slower using bulletproofs+) but a good playground if you want to play use bulletproofs+ or use as reference code (until the authors of the paper will release their code)
This is a cool development, thanks for sharing!
After the paper goes through peer review, I'd be down to include it, maybe under the yoloproofs
feature guard.
@omershlo, do you mean that your code is 4x slower than this library? Do you know what the cause of the slowdown? From a skim of the paper, it seems like Bulletproofs+ should have about comparable proving and verification times than Bulletproofs, with smaller proof sizes.
Hi @cathieyun , X = Dalek range proof computing time using Bulletproofs Y = KZen range proof computing time using Bulletproofs Z = KZen range proof computing time using Bulletproofs+
X is ~4x faster than Z X is ~6x faster than Y Y compare to Z is similar to the results in the paper (we actually get faster verification times as opposed to the paper). We will release a doc with a detailed table very soon.
Re: peer review, yous know that it usually doesn't check proofs, right? (And peer reviewers are explicitly not required to even look at "supplementary information", which the proofs are forced into because of page limits for most journals/conferences.)
Re: peer review, yous know that it usually doesn't check proofs, right? (And peer reviewers are explicitly not required to even look at "supplementary information", which the proofs are forced into because of page limits for most journals/conferences.)
Peer review has more value that just proof checking :-)
Also, the reviewers are not forced to read the proofs, but in my experience they do take a look at them.
(Just an FYI post in case you missed it. More techniques for reducing costs.)
https://eprint.iacr.org/2020/735