Recently, Somebody Somewhere wrote these words
> I still find myself interested in what optimization does besides make
> the file bigger. I have seen some small (1 or 2%) differences in the
> times of a tar command of a big file for instance. I guess I'll have
> to take the new gcc4 lfs and replicate it with heavy optimization and
> see if there is any difference.
>
It's about the processors. PCs have been made with i88(8088) i86(8086,
80186) i286, i386, i386, i486, i586, before all the variants took hold
in their own right, and the 64 bit flavours.
As each cpu generation added instructions, the optimisations allows the
compiler to use these. That doesn't mean that's what the programmer
wrote. Few save much. i386 - i586 is one decent step, and i686 - i786
is, I am told, another. MMX instructions save also. But the differences
are minimal for a fast pc. Where you see the biggest speed difference
is between the early generations of cpu. Evrything has been squeezed
out of the hardware already, and compatability is a curse on the pc
from which it will never recover.
BTW, when I was running i686, the advice was to compile for i586 since
the i686 optimisations actually _lost_ speed in some instances
--
With best Regards,
Declan Moriarty.
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
