Date: Wed, 26 Jan 2000 15:14:26 +0100
From: Jan Hubicka <hubicka AT atrey DOT karlin DOT mff DOT cuni DOT cz>
To: pgcc AT delorie DOT com
Subject: Re: pgcc on AMD K6
Message-ID: <20000126151426.D15000@atrey.karlin.mff.cuni.cz>
References: <Pine DOT LNX DOT 4 DOT 21 DOT 0001141830350 DOT 4315-100000 AT hq DOT alert DOT sk> <Pine DOT LNX DOT 4 DOT 21 DOT 0001261105090 DOT 26171-100000 AT hq DOT alert DOT sk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 1.0i
In-Reply-To: <Pine.LNX.4.21.0001261105090.26171-100000@hq.alert.sk>; from mandos@hq.alert.sk on Wed, Jan 26, 2000 at 11:43:02AM +0100
Reply-To: pgcc AT delorie DOT com
Errors-To: dj-admin AT delorie DOT com
X-Mailing-List: pgcc AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> 
> > > Check the Makefile, if there are any refs to -mpentiumpro, (which I did see, though I think that's supposed to only enable the scheduling for Pentium Pro),
> > > -march=pentiumpro or -march=i686, change them to -march=k6. Also, don't use i686 ASM, which may also contain Pentium Pro specific opcodes. K6 doesn't support cmove, but
> > > AMD Athlon does (I just checked mine, and WCPUID says it does)
> > 
> > K6-2 tests show that more optimal combination would be
> > -march=pentium -mcpu=pentiumpro
> 
> yes, it seems that current PPro machine desc. matches K6-2 better
> than K6 description
> one of the reasons is IMHO unrealistic K6 branch costs model
> but there are others too
> i wrote some support for K6-2/3 and it seems to be promising
> but i'm not satisfied with the performance yet
> and of course i have to test it some more
> i'll send it then

If you've done some tuning on pgcc, please let me know about it.
I would like to redo the tuning for mainline gcc soon, so I would like
to know your work...

Honza