Optimized distro for i686

Christopher Browne cbbrowne-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
Fri Dec 30 02:04:33 UTC 2005


On 12/29/05, Steve <bassix-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org> wrote:
> I guess I am curious as to what is involved in "optimizing" a distro
> for a certain CPU/platform.

This tends to be about setting up compiler flags and possibly
compiling things using some fairly exactly chosen version of GCC.

A lot of it is snakeoil; generic 386/486 code tends to be
"self-optimized" by running it on those fancy new processors.

The potential "big wins" tend to come if you can take specific
advantage of extra registers, stuff like Intel's MMX extensions and
similar DSP-like systems on the recent chips.  Unfortunately, the
patterns that could be mapped onto MMX are things that compilers don't
generally get written in such a  way that they can detect them.  So
the way you could take advantage of MMX, 3DNow, AltiVec and such is if
someone has written a library (e.g. - for playing MP3 or Ogg Vorbis)
that uses those special instructions.

There would definitely be merit to recompiling for AMD64 even if you
could still run 32 bit code there; AMD64 offers a bunch more
registers, which can certainly lead to faster code as more data can
stay on the CPU and not be spilled to slow memory.
--
http://www3.sympatico.ca/cbbrowne/linux.html
"The true  measure of a  man is how he treats  someone who can  do him
absolutely no good." -- Samuel Johnson, lexicographer (1709-1784)
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml





More information about the Legacy mailing list