Re: P-Code and virtual machines

pardo@cs.washington.edu (David Keppel)
Fri, 25 Mar 1994 02:41:08 GMT

          From comp.compilers

Related articles
P-Code and virtual machines ghmilmei@iiic.ethz.ch (1994-03-23)
Re: P-Code and virtual machines pardo@cs.washington.edu (1994-03-25)
Re: P-Code and virtual machines koopman@casun2.res.utc.com (1994-03-28)
Re: P-Code and virtual machines rau@slapshot.pa.dec.com) (1994-04-05)
Re: P-Code and virtual machines pardo@cs.washington.edu (1994-04-08)
| List of all articles for this month |

Newsgroups: comp.compilers
From: pardo@cs.washington.edu (David Keppel)
Keywords: interpreter, design
Organization: Computer Science & Engineering, U. of Washington, Seattle
References: 94-03-104
Date: Fri, 25 Mar 1994 02:41:08 GMT

ghmilmei@iiic.ethz.ch (Gerard Henri Rene Milmeister) writes:
>[Books or papers on the instruction set of an interpreter for a
> procedural language? I have experience with stack machines and they
> seem inappropriate for e.g., C.]


I can't say how appropriate they are, but several older machines (e.g., HP
and Tandem) used *hardware* stack machines and ran e.g., FORTRAN. Stack
machines have the advantage that the code is compact, easy to decode and
easy to convert to threaded code (for fast decode) and can be implemented
with similar efficiency across a fair range of machines.


Virtual machines that have a modest number of (virtual) registers may work
well on host machines with, say, double the number of hardware registers
but may perform worse on machines with a limited number of registers.


An alternative is to (dynamically) compile the v-code to n-code or a
machine-specific v-code in order to improve register allocation.


;-D on ( Virtual vegetables ) Pardo
--


Post a followup to this message

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