what's wrong?
PanosYang opened this issue · 1 comments
Spartan offers the fastest prover with speedups of 36–152× depending on the baseline, produces proofs that are shorter by 1.2–416×。
Does it mean that the size of the proofs shorter by 1.2-416x?
If not, this means the time to generate the proof of generation shorter by 1.2x . 36x speedups of prover, but only with 1.2x shorter produces proofs , why the difference so big?
There is no (direct) relationship between prover time speedups and proof size reductions. Spartan's prover is faster than baselines because its underlying algorithms are fast (e.g., the sum-check protocol) and also because it avoids certain superlinear algorithms used in other proof systems (e.g., FFT). Our paper (https://eprint.iacr.org/2019/550.pdf) provides more details.