FACTS instead of FUD about gcc3

David Lautenschlager david.lautenschlager at mcdata.com
Wed May 8 10:32:12 PDT 2002


A bug which has been fixed, and now GCC 3.XX gives a performance increase in
all the tests that that poster has run. So far we have one claim with
numbers that 3.XX produces code that is about 8% faster, one claim by the
father of LFS that his system is 25-50% faster with GCC 3.XX(without
numbers), and one claim that GCC 3.XX is horrible and slow (without
specifics let alone numbers). I find the first too claims far more credible
than yours. Especially since you have repeated yours 50 times without adding
any details.

The standard policy of LFS is to update to the latest version of each
package unless it can be shown to break the LFS system. Even then if the
work arounds are minor, or on the same scale as the workarounds for the
previous version, the change will still be made.

I think you have a misconception of what LFS is. LFS is not a better
distribution, it is not a distribution at all. It isn't even about
stability, even though that is very important. LFS is about teaching people
to build their own linux system. GCC 3.XX does not seem to threaten the
ability of LFS to teach, and it fits with the LFS upgrade policy. Certainly
if it does not fit your needs, don't use it. But don't project your needs
onto LFS as a whole. LFS is not about my needs, or your needs, or even
Gerard's needs. It is about teaching.

Later,

Dave

> -----Original Message-----
> From: Ian Molton [mailto:spyro at armlinux.org]
> Sent: Wednesday, May 08, 2002 11:19 AM
> To: lfs-dev at linuxfromscratch.org
> Subject: Re: FACTS instead of FUD about gcc3
> 
> 
> On Wed, 08 May 2002 18:29:50 +0200
> johan <j.h.veenstra at stud.tue.nl> wrote:
> 
> > 
> > As it turns out, the performance regression had nothing to do with 
> > integer based operations, it was all about not-inlining 
> memset() from 
> > libc: 
> 
> It was still a bug, though.
> -- 
> Unsubscribe: send email to listar at linuxfromscratch.org
> and put 'unsubscribe lfs-dev' in the subject header of the message
> 
-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



More information about the lfs-dev mailing list