Date:	Wed, 22 Dec 1999 21:35:26 +0100
From:	Jan Hubicka <hubicka AT atrey DOT karlin DOT mff DOT cuni DOT cz>
To:	pgcc AT delorie DOT com
Subject:	Re: K7/Athlon, Thoughts anyone?
Message-ID:	<19991222213526.A13457@atrey.karlin.mff.cuni.cz>
References:	<3860FD5C DOT BDD0C7DF AT lycosmail DOT com>
Mime-Version:	1.0
X-Mailer:	Mutt 1.0i
In-Reply-To:	<3860FD5C.BDD0C7DF@lycosmail.com>; from ajschrotenboer@lycosmail.com on Wed, Dec 22, 1999 at 11:33:32AM -0500
Reply-To:	pgcc AT delorie DOT com
X-Mailing-List:	pgcc AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

> As a (feeble?) attempt to put some more life into this otherwise dead
> list (not really dead, but very little is going on. Maybe everybody's on
> vacation[Vacation, what's that?]) I'd like to hear anyone's thoughts on
> the AMD Athlon. It's strengths, weaknesses, who wants one, who doesn't.
> 
Well, I would surely love to have one. Almost everything is better than P/60
> It should probably center around the possibility for optimizing
> GCC(Formerly EGCS)/PGCC for it.
> 
The latest CVS snapshots of gcc already support Athlon. I have some extra
patches pending in this area, I will send them to the egcs-patches list as soon as I will have some
free time.

Problem with commenting the Athlon architecture is, that AMD failed to write documentation
\good enought to discover it's weakness.
It has very good performance IMO, but during the tunning
I found it prety touchy about hte code it executes and making mysterious speedups/slowdowns
in the benchmarks.
Also unlike K6 it is better in FPU than in the integer..

Thats probably all I can say..
Honza
> Maybe as a sideline, some comments on the IA-64, and EPIC. Apparrently
> some compiler writers are excited, but maybe we should hear more, beyond
> the docs, to what it could actually mean.
> 
> Anyone, Anyone?

- Raw text -