High performance interpreter for UTM SE

[meokit]

Hi everyone,

I’ve noticed that some high-performance interpreters (even without JIT), such as libriscv, can achieve 10x the performance of QEMU TCI.

I also developed my own in C++, which achieves about 1/3 the performance of QEMU’s JIT. While JIT is clearly superior, QEMU TCI is not particularly fast. Given this, is it feasible to replace the current TCI backend with something more efficient, like libriscv, within the existing UTM architecture?

[osy]

I've had an idea to use the new webasm backend for QEMU but like any change it'll take a lot of time and effort to develop. Feel free to try it yourself and send a PR if it works.