GoKawiilHow To Make a Fast Dynamic Language Interpreter
This post is about optimizing an extremely simple AST-walking interpreter for a dynamic language called Zef that I created for fun to the point where it is competitive with the likes of Lua, QuickJS, and CPython.
Why?
Most of what gets written about making language implementations fast focuses on the work you'd do when you already have a stable foundation, like writing yet another JIT (just in time) compiler or fine tuning an already pretty good garbage collector. I've written a lot of posts about crazy optimizations in a mature JS runtime. This post is different. It's about the case where you're starting from scratch, and you're nowhere near writing a JIT and your GC isn't your top problem.
The techniques in this post are easy to understand - there's no SSA, no GC, no bytecodes, no machine code - yet they achieve a massive 16x speed-up (67x if you include the incomplete port to Yolo-C++) and bring my tiny interpreter into the ballpark of QuickJS, CPython, and Lua.
The techniques I'll focus on in this post are:
Value representation.
Inline caching.
Object model.
Watchpoints.
... continue reading