Long-term L1 execution layer proposal: replace the EVM with RISC-V

(not sure how to edit posts here, so just write some additional thoughts)

  1. LLVM has a lot of compiler bugs. Solidity, in comparison, has excellent track record. This is largely thanks to much simpler pipeline.
  2. None of RISC-V vms are production ready, and their complexity is significant. Their teams might falsely claim otherwise, but, for example, SP1 recommends their customers to use a trusted prover. It came in handy when we have recently uncovered a soundness bug allowing to prove arbitrary statements.
  3. @pcaversaccio this is correct intuition. The likely bottleneck is stack management (due to larger granularity of instructions), and, potentially, jumps. That being said, of course going from interpretation to execution will have huge advantages anyway, but this requires to figure out things written in my previous post.
  4. @AdamCochran strongest possible support of enshrined aggregation of stateless computations.
3 Likes