Re: genetic compilation (David Starner)
3 Apr 2000 04:08:48 -0400

          From comp.compilers

Related articles
genetic compilation (Christophe Grand) (2000-03-23)
Re: genetic compilation (Joachim Durchholz) (2000-03-28)
Re: genetic compilation (Pratap Subrahmanyam) (2000-04-01)
Re: Re: genetic compilation (Brock) (2000-04-01)
Re: genetic compilation (Dr. Andy Nisbet) (2000-04-01)
Re: genetic compilation (Jean-Luc Nagel) (2000-04-01)
Re: genetic compilation (Chris Fraser) (2000-04-01)
Re: genetic compilation (2000-04-03)
Re: genetic compilation (2000-04-03)
Re: genetic compilation (2000-04-20)
| List of all articles for this month |

From: (David Starner)
Newsgroups: comp.compilers
Date: 3 Apr 2000 04:08:48 -0400
Organization: Oklahoma State University
References: 00-04-019
Keywords: optimize, architecture

On 1 Apr 2000 14:09:15 -0500,
  Brock <> wrote:
> Why wouldn't it work for cross-compilation? It seems that the
> performance measurement could be done just as well using a table of
> measured performance values instead of doing the actual tests,
> thereby making it quite possible.

The Intel Pentium and some clone may process the same ISA, but they
won't run instructions at the same relative speeds. If you're
compiling to a particular chip, it's a lot easier to have that chip
than find a set of values for that chip - which may not be
accurate. You also have to simulate exactly how it does its pipelining
and several other internal details.

So, yes, it could be done, but it's not nearly as simple as what was
being done.
David Starner -

Post a followup to this message

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