Re: a compembler for x86 that looks nearly portable

"R. L. Watkins" <RLWatkins@CompuServe.Com> wrote
> Yes. The same thing might be said for parts of the PDP-11 / WD-16
> instruction set, on which Macro 11 is based. Most useful seem to be the
> ones they have in common.
>
> That convergence is probably the result of what programmers in general need
> (e.g. the addressing modes), as opposed to "special purpose" features that
> are most useful for specific applications (e.g. the binary-to-BCD
> instructions).
>
> R. L. Watkins
>
> ----- Original Message -----
> From: "Rick Hohensee" <rickh@capaccess.org>
> > The point I personally want to repeat in groups such as this
> > is that the 386 might be a superset of a useful portable
> > virtual machine. If you look at osimplay source, the 386-ness
> > is not often evident.


Yes yes. Even Forth is getting addressing modes. Having gotten
over-familiar with the 386, it's the history of register
microprocessors in one chip. The ASCII stuff is obsolescent just in
context of the 386 itself, never mind a general register-machine
model. The registers of the 386 are representative, to me, of what
computers are trying to be, so to speak, and very similar to the 6502
in terms of basic purpose. What I think computers are trying to be is
a several-stack machine, probably with stacks for things for which the
6502 has registers. My 3-stack machine "just fits" in the 386 with
very few register spills in the VM. Just a bit more is probably right.


H3sm is Hohensee's 3-stack machine
ftp://ftp.gwdg.de/pub/cLIeNUX/interim/ABOUT


(off to look for "Macro 11")
Rick Hohensee

Return to the comp.compilers page.
Search the comp.compilers archives again.