My own version of his instruction set. http://m.imgur.com/GajGpjQ With two versions. On that should run a bit faster but is simpler. And a more complicated full version.
The IS is slowly changing. And so is the CPU. It now will have a 64 deep stack and also 8 registers for some stuff. And it also has a port to get stuff into and out of the system.
some suggestions,
1. Do you have subtract both ways (A-B and B-A) or atleast a swap function to swap top 2 elements of the stack?
2. A bit to signify a pop/ no pop (sometimes you want to hold the second of stack for example processing on data streams) default is to always pop.
3. 64 deep stack seems a lot for mc considering chunk updates. you can also use a pointer or a combination of both. (8 registers in stack setup)
Wrubbel also did make a nice IS on a stack machine.