Re: virtual machine efficiency

cr88192 wrote:


> "Chris Dollin" <kers@hpl.hp.com> wrote in message


>> 16-bit instructions turned out to be an interesting and occasionally
>> frustrating choice, as was the decision to try and combine stack-based
>> and register-aka-local-variable instructions. That the machine had
>> *two* stacks [1] - one for values, one for locals/return addresses - was
>> an automatic consequence of umpteen years writing Pop11 (and similar)
>> code.


> mine also has two stacks: the value stack; and a kind of opaque
> meta-stack used for control.


> locals were stored in their own space which is indirectly referenced
> by the control stack (variables and similar are accessed with their
> own opcodes).


It's important to note that the two stacks in my VM are decoupled and
not just a convenience - the value stack doesn't get cut back just
because you return from a procedure, for example, and the value stack
can grow arbitrarily large during a procedure's execution.


It's not quite as free-and-easy as Pop11's open stack, partly because
one of the points of designing Spice [no, not that one] was to get
most of the benefits of the open-stack model without the gotchas -
although most of the constraints are imposed by code generation rather
than by the instruction set. (Although one of the procedure-entry
instructions explicitly transfers values off the value stack into the
control stack.)


--
Chris "electric hedgehog" Dollin

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