Re: libc6 optimizations

[Neilen <brick@adept.co.za>]

Hi 

On Sun, 2003-08-24 at 05:24, Arnt Karlsen wrote:
> On Sat, 23 Aug 2003 22:33:11 +0200, 
> Pierre THIERRY <pierre.thierry@moine-fou.org> wrote in message 
> <[🔎] 20030823203310.GW24470@imperatrice.arcanes>:
> >
> > Benchmark have been made between Debian, Mandrake and Gentoo, with
> > identical version of programs, and there is no signficant speedup with
> > Gentoo.
> > 
> > As a matter of fact, it is the slowest of the three distros in all but
> > one test of this bench, and never the fastest...
> 
> ..they mention cpu and compile flags as possible isses on Gentoo.
> Such a race should be set up with a team from each distro, doing 
> the same tasks, inside a set cost and time frame.

Also, if I remember correctly, they were using a P4 based Celeron. My
feeling is that a chip with only 128k cache might in fact do worse
running code optimised for a 512k cache CPU, that the generic code.  Or
does GCC not make any assumptions about the cache size?  Either way, I'd
like to see some data points with different CPUs (eg. Athlon 256k, 512k
cache, P4 256k, 512k cache).  

Cheers
Neilen


> 
> -- 
> ..med vennlig hilsen = with Kind Regards from Arnt... ;-)
> ...with a number of polar bear hunters in his ancestry...
>   Scenarios always come in sets of three: 
>   best case, worst case, and just in case.
-- 
Neilen <brick@adept.co.za>