Date: Thu, 6 May 1999 23:47:32 +0200
To: pgcc AT delorie DOT com
Subject: Re: pgcc problem with AMD K6
Message-ID: <19990506234732.E20732@cerebro.laendle>
Mail-Followup-To: pgcc AT delorie DOT com
References: <14129 DOT 63360 DOT 491937 DOT 17038 AT hectic DOT hectic DOT net> <19990506230451 DOT D20732 AT cerebro DOT laendle> <14130 DOT 2529 DOT 79466 DOT 367962 AT hectic DOT hectic DOT net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <14130.2529.79466.367962@hectic.hectic.net>; from Alex Shnitman on Fri, May 07, 1999 at 12:37:45AM +0300
X-Operating-System: Linux version 2.2.7 (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, May 07, 1999 at 12:37:45AM +0300, Alex Shnitman wrote:
>  > Because egcs added these switches as -mk6. The rationale was that no -m
>  > option includes the vendor name (-mintelpentium?).
> 
> Oh, I didn't know that.

How could you? Maybe it should belong to the faq (hint, hint...)

> I'm sure. I'll tell you the exact flow of events. I compiled it with
> -mk6. It segfaulted. I compiled it with -mk6 -march=i486. It worked. I 
> inferred that -mk6 uses -march=pentium by default. So I compiled with
> -mk6 -march=pentium and it segfaulted again, which I took as a proof
> to my assumption.
> 
> Am I misunderstanding something?

No, I just wanted to be sure about the options ;)

> 
>  > > Why does the K6 qualify as i486 architecture? It's supposed to be
>  > > totally Pentium-compatible, isn't it?
>  > 
>  > The pentium, the i486 and the amdk6 share the same instruction set
>  > architecture (with regards to a compiler). Both pentium and amdk6 use the
>  > same instructions (i.e. -march=i486)
> 
> So what uses -march=pentium?

This is strange. -mk6 expands to "-mcpu=k6 -march=i486". However, some
of the scheduling parameters change depending on the -march= switch
(they shouldn't). I have never looked into that, but maybe gcc changes
scheduling parameters with the -march=, which it shouldn't do. I'll
investigate.

> You have to download a source archive and a patch (a patched source
> can't be distributed because of patents on mp3 encoding). The source
> archive (the ISO sample code) is at http://geek.rcc.se/mp3encoder/ ,
> there are many links to mirrors there. The patch (and patching
> instructions) are at http://www.uq.net.au/~zzmcheng/lame/download.html .

Thanks, I'll try to look into this. (Don't expect this too quickly,
though)

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