ponderous 'make world' times post GCC 3.3...

[Kris Kennaway]

On Wed, Aug 06, 2003 at 11:44:26PM -0700, David O'Brien wrote:
> On Wed, Aug 06, 2003 at 11:32:10PM -0700, Kris Kennaway wrote:
> > On Wed, Aug 06, 2003 at 11:25:36PM -0700, David O'Brien wrote:
> > > Am I the only one that saw 'make world' go from almost 3 hours with GCC
> > > 3.2 to:
> > 
> > Nope, that's just gcc 3.3.  Someone had a reference to some benchmarks
> > the gcc people did that showed progressive slowdowns over the course
> > of the gcc 3.x branch.
> 
> It has been argued by Richard Henderson (urber GCC hacker) that 3.3 is no
> slower than 3.2.  I'm now thinking that this is a cache-size issue.  I
> wonder if GCC is blowing out my tiny 256KB cache.

I havent benchmarked make world myself recently, but the benchmark I
remember was compiling some reference version of gcc with different
gcc compiler versions (from 2.7.x to 3.4.x), and it demonstrated
something like a factor of 3 slowdown from gcc 3.2 to 3.3 on i386.
I'm paraphrasing from memory though, so I could be way off.

Kris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-sparc64/attachments/20030807/18501f0f/attachment.bin