GCC 2.95, 2.96 and 3.0 on linear algebra (was RE: 2 GB file limitation)

["M. Edward Borasky" <znmeb@aracnet.com>]

I have heard from the Atlas linear algebra folks the following:

1. For compiling Atlas, both on Athlons and Pentia, GCC 2.95.x produces
*significantly* faster operation than either 3.0.x or 2.96.x
2. For IA64, the reverse is true: GCC 3.0.x produces significantly faster
code.

I can dig up the URL for the mailing list if anyone cares for the details.

--
M. Edward (Ed) Borasky, Chief Scientist, Borasky Research
http://www.borasky-research.net  http://www.aracnet.com/~znmeb
mailto:znmeb@borasky-research.net  mailto:znmeb@aracnet.com

Q: How do you tell when a pineapple is ready to eat?
A: It picks up its knife and fork.

> -----Original Message-----
> From: linux-kernel-owner@vger.kernel.org
> [mailto:linux-kernel-owner@vger.kernel.org]On Behalf Of Gábor Lénárt
> Sent: Sunday, September 30, 2001 1:24 AM
> To: Luigi Genoni
> Cc: Linux Kernel
> Subject: Re: 2 GB file limitation

[snip]

> > > I think you can get >2GB support if you've Gcc 3.0. Even with
> the latest
> > >
> > ???
> > I am using it and I am using gcc 2.95.3 for normal things,
> > and to compiled my kernel and my libc, because gcc
> > 3.0.1 produces slower binaries on my Athlons (yes, with athlon
> > optimizzations turned on), at less for my programs, and it is better to
> > avoid it for glibc compilation because of back compatibility issues.
>
> Yes, gcc3 is (well at least NOW) a piece of shit. It produces BIGGER and
> SLOWER binaries ... Checked on: Athlon, AMD K6-2.
> With the same gcc command line ...