In this blog series, the author details their work on improving the performance of programming language interpreters in Rust. Contrary to popular belief, AST interpreters showed impressive performance compared to bytecode interpreters. The focus is on optimizing interpreters for the SOM language, where method calls on objects are prevalent. The challenge lies in optimizing loops, specifically the `to:do:` method, which required implementing specialized bytecode and introducing performance-enhancing modifications to the interpreter. While some solutions led to code complexity and minor slowdowns, the overall results showed significant speed improvements. The blog concludes with the ongoing efforts to further optimize interpreter performance through innovative solutions.
https://octavelarose.github.io/2024/05/29/to-do-inlining.html