Pending GCC-3.1 upgrade - do or don't

Ian Molton spyro at armlinux.org
Wed May 8 09:27:39 PDT 2002


On Wed, 8 May 2002 11:07:59 -0400
Gerard Beekmans <gerard at linuxfromscratch.org> wrote:

> On Wed, May 08, 2002 at 10:19:08AM +0100, Ian Molton wrote:
> > GCC3 may be more actively developed, but it isnt faster, doesnt
> > produce better code (often worse), and isnt stable yet.
> 
> Ian, if you say that one more time I am going to ignore anything else
> you have to say about this topic.

<sigh>

> It does produce faster code, I've given you a bunch of examples,

Actually all you said is that KDE3 'felt' faster.

> Why don't you just accept
> that GCC-3 can produce better code.

I already do. what I dont accept is that it produces /enough/ better
code to make up for its regressions on the whole (although this /is/
improving in the CVS versions). I also dont accept that its production
grade yet.

> To me it sounds like you have this
> idea in your head that gcc-3 cannot possibly create anything faster
> than gcc-2 does and you're not willing to even consider the
> possibility that we have actually tested things.

Not true.

> Let me say for the
> last time: my overall system performance increase is 25% to 50% easily

And as you never answered me the first time, HOW did you test that? or
is it still 'it feels faster' ?

and have you actually /thought/ how much faster 50% is?

> I timed it with a stopwatch the first time I finished all the
> packages and things are physically faster.

Which version of GCC ? apparently the CVS versions now have better
preprocessor speed, on a par with 2.95.x or better now).

> but I will
> this list some numbers when we're getting closer to an upgrade date
> (meaning when gcc-3.1 is closer to a release date).

I will say it again, then. WHY are we going to upgrade as soon as 3.1 is
released? (that is very much what the above statement implies). Why dont
we WAIT and make sure its all OK first?

no one has answered 'why dont we wait' yet.
-- 
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