>
> I would imagine that just compiling ghostscript for i686 isn't enough;
> one would also have to recompile things it depends on (like libc) in
> order to really tell whether there was an improvement.
>
But when you compile benchmarks (even those who make minimal use of
the library) you find benefits are small.
> In any case, if I go through the work to get the source RPMs to produce
> i586 and i686 code, does anyone know any particular reason why RedHat
> *wouldn't* accept that change into their source-base or whatever?
> (Assuming my job is professional, of course. :-) I would like future
> source-RPMs to have the *option*, at least.
>
Some kremlinology I made based on an Alan Cox mail makes me think
RedHat will do the work when, with gcc 2.96 and later, compiling with
processor-specific optimizations will give a real performance boost.
With egcs 1.1 it is not worth the trouble.
--
Jean Francois Martinez
Project Independence: Linux for the Masses
http://www.independence.seul.org
--
To unsubscribe:
mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
