RISC Zero在这里做得非常好,我可能更担心Rust标准库或LLVM的bug,而不是他们电路的健全性。不幸的是,这仍然留下了很多担忧的空间! 编译器bug很常见,而32位RISC-V是一个相当小众的目标。
Jacob
Jacob2025年4月22日
Is it hard? Yes Our soundness bugs devastating? Yes Are we closer than you might think? Yes From Day 1, yes the same day we released our proof system, @RiscZero began the long process of formally verifying our zkVM. Recently we launched "Picus" which mathmatically proves determinism in our circuits during CI! This is the cause of 96% of soundness bugs in ZK systems (Per ZK-Security). This effort plus some more we are working on, shows we can get to a provably sound RISC-V zkVM faster than you most think.
1.64K