Date: Fri, 12 Mar 1999 23:59:17 +0100
To: pgcc AT delorie DOT com
Subject: Re: amd and float
Message-ID: <19990312235917.K9148@cerebro.laendle>
Mail-Followup-To: pgcc AT delorie DOT com
References: <19990312222241 DOT F9148 AT cerebro DOT laendle> <Pine DOT GSO DOT 4 DOT 05 DOT 9903122241190 DOT 27432-100000 AT legolas DOT mdh DOT se>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <Pine.GSO.4.05.9903122241190.27432-100000@legolas.mdh.se>; from Henrik Berglund SdU on Fri, Mar 12, 1999 at 10:48:11PM +0100
X-Operating-System: Linux version 2.2.3 (root AT cerebro) (gcc driver version pgcc-2.93.09 19990221 (gcc2 ss-980929 experimental) executing gcc version 2.7.2.3) 
From: Marc Lehmann <pcg AT goof DOT com>
Reply-To: pgcc AT delorie DOT com
X-Mailing-List: pgcc AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On Fri, Mar 12, 1999 at 10:48:11PM +0100, Henrik Berglund SdU wrote:
> Is anyone working on 3dnow support? 3dnow should in some cases improve
> float performance much, maybe with four times. Maybe it would be easiest
> to do a math link library like libmoto.a for ppc. so one link with that
> first "gcc program.c -l3dnow -lm"  

The biggest probelm I see is that 3dnow uses different float formats than the
x86 fpu. even worse, you probably won't get improvements in the common case
anyway (doubles). also, the interoperability between the x86 fpu and 3dnow is
very bad.

the good thing is that, finally, its a sane design. a very sane design. if we
only had this design from the very beginning (and no x86 fpu).

--  
      -----==-                                             |
      ----==-- _                                           |
      ---==---(_)__  __ ____  __       Marc Lehmann      +--
      --==---/ / _ \/ // /\ \/ /       pcg AT goof DOT com      |e|
      -=====/_/_//_/\_,_/ /_/\_\       XX11-RIPE         --+
    The choice of a GNU generation                       |
                                                         |