[Idea] Using static binary translation and assembly plugin to boost asterisc's performance to hundreds of times faster, even more

[yao-jz]

Idea

For fraud proof using asterisc (riscv virtual machine), we don't need to run all the simulation in the designed interpreter. Recently I want to use static binary translator to boost the performance of riscv emulation by translating all the riscv instructions to native machine codes. In this way, the original program can be run hundreds of times faster or more.

I am looking for anyone who is interested in qemu, computer architecture, and system to join me. Any idea of collaborating with optimism is also appreciated.

Another optimistic L2 platform is using wasmer to run their fraud proof process and Rust/C smart contract in nearly native speed. And I think optimism's stack may also need this, which can lower the cost for computation for the dispute process. And I think binary translation can't be a better approach to achieve this.

Thanks!