If you like this, you might like Lisp interpreter written in assembly in a single file. It is one of the best commented code ever written imo.
Also, another, more interesting Lisp by the same author: (http://piumarta.com/software/maru/). Maru is basically a lisp where some of the core functions like eval and apply are extensible from user code. There's basically a global map of types to evaluators and applicators, with some functions for you to register your own types and get the evaluation behavior you want.
Yes, jonesforth definitely inspired and influenced me; that's why Jones is first in the Acknowledgements section.
If you're curious, I keep a list of single-file implementations of programming languages (including jonesforth):
and their opinion was that it was easier to do the lisp than it was to do the FORTH. Although right now they are trying to improve their C Compiler prototype https://github.com/oriansj/M2-Planet before they convert it to assembly
If you were to implement the same Lisp in C then compare, then maybe the assembly variant would be faster for the reasons you mention. Or maybe not.
Also, I modeled arpilisp after the original Lisp. That's barely a first step, and possibly the wrong first step, for anything non-trivial, including applications requiring a "super fast lisp that can compete with Go."
But I didn't write arpilisp for performance. I wrote it to learn and share. Enjoy!