xfs
[Top] [All Lists]

Re: gcc-2.96-nn status

To: Seth Mos <knuffie@xxxxxxxxx>, Alan Eldridge <alane@xxxxxxxxxxxx>
Subject: Re: gcc-2.96-nn status
From: Joseph Fannin <jhf@xxxxxxxxxxxxxx>
Date: Tue, 18 Sep 2001 21:15:55 -0400
Cc: SGI XFS Dev List <linux-xfs@xxxxxxxxxxx>
In-reply-to: <4.3.2.7.2.20010918192539.0375bbd0@pop.xs4all.nl>
References: <4.3.2.7.2.20010918190751.03312448@pop.xs4all.nl> <4.3.2.7.2.20010918192539.0375bbd0@pop.xs4all.nl>
Sender: owner-linux-xfs@xxxxxxxxxxx
On Tuesday 18 September 2001 13:27, Seth Mos wrote:
> At 13:19 18-9-2001 -0400, Alan Eldridge wrote:
> >On Tue, Sep 18, 2001 at 07:14:47PM +0200, Seth Mos wrote:
> > >At 12:40 18-9-2001 -0400, Alan Eldridge wrote:
<snip>
> > >I can, altough it does not have as many optimizations. So I only think
> > > that the runtime speed would be affected. The linux kernel already
> > > knows what a Athlon processor is and what to do with it. I think this
> > > makes a larger difference then the compiler.
> >
> >Umm, how? rpm --target athlon causes gcc to be invoked with -march=athlon,
> >which kgcc knows nothing about so it dies.
>
> You can specify a target of i686 and have a kernel config for athlon/duron
> systems

    kgcc / egcs doesn't have k6 optimizations either. :-(  Yet the kbuild 
scripts somehow know this and use -march=i586 when -march=k6 is not 
available.  I'm sure the same test could be created for the athlon, and would 
be accepted into the kernel, but no one has bothered to do it.

--
Joseph Fannin
jhf@xxxxxxxxxxxxxx

"Bull in pure form is rare; there is usually some contamination by data."
    -- William Graves Perry Jr.


<Prev in Thread] Current Thread [Next in Thread>