Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 29 Jul 2004 20:58:48 -0700
From:      Steve Kargl <sgk@troutmask.apl.washington.edu>
To:        Paul Seniura <pdseniura@techie.com>
Cc:        Scott Long <scottl@freebsd.org>
Subject:   Re: gcc 3.5 (Re: about the gcc 3.4.x problems)
Message-ID:  <20040730035848.GA1700@troutmask.apl.washington.edu>
In-Reply-To: <20040729201439.BFB685C11@techpc04.okladot.state.ok.us>
References:  <20040729173132.GA917@troutmask.apl.washington.edu> <20040729201439.BFB685C11@techpc04.okladot.state.ok.us>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 29, 2004 at 03:14:39PM -0500, Paul Seniura wrote:
> 
> Steve Kargl said:
> >On Thu, Jul 29, 2004 at 11:47:38AM -0500, Paul Seniura wrote:
> >> 
> >> At any rate, GCC 3.5 has been available for some time. 
> >
> >3.5 won't be released for at least another year.  You can
> >get the GCC cvs branch that is used for 3.5 development,
> >but it is certainly not ready for prime time.  I build the
> >3.5 branch several times a week and there are some severe
> >problems with it (e.g., excessive compile times and excessive
> >use of memory).
> 
> I might try a (quasi-)weekly tarball sometime (it's on a
> mirror I can reach; can't do CVS here). 
> Where might your builds be at, if I may ask?

I don't distribute the builds.  I only build the C and
gfortran compilers, because I'm actively contributing to
gfortran.  g77 has been removed from GCC.  Bootstrapping
C and gfortran takes 2 to 3 hours on my 1.2 GHz athlon
with 1.5 GB memory and SCSI disks.  I've never built
g++ or java, but I'm willing to bet that it takes a long,
long time.

> 
> >For those that do not know, the entire middle-end and much
> >of the back-end of GCC has been re-written in 3.5.
> 
> This is interesting.  I do know Apple & IBM are coordinating
> work on GCC -- they *really* want the G5 chip to be the best
> supported fastest system going.

To learn more about tree-ssa, which is the guts of 3.5, see

http://gcc.gnu.org/projects/tree-ssa/

It's my understanding that once tree-ssa settles down, it
will offer many more opportunities for high level optimizations.

-- 
Steve



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040730035848.GA1700>