Re: SUMMARY: pentium optimized debian discussion
On 30 Jul 1998, Guy Maor wrote:
> Joey Hess <joey@kitenet.net> writes:
>
> > Pentium optmized binaries are about 10% larger than i386 binaries (the
> > compiler probably inserts NOPs for pipelining purposes). If debian switched
> > to pentium optimized binaires, it would increase the installed size of
> > debian a lot,
>
> Why "a lot"? Let's say we compile only 10% of the distribution that
> matters and that binaries account for about half of a package's
> installed size. That's an 0.5% increase in distribution size?
The size gain in pentium optimized binaries is not from nops but from
alignment constrains. Structure members are aligned on paragraphs and
some function elements are aligned on pargraphs too.
> > Pentium optimized binaries will run slower on 386's than normal i386 binaries.
>
> Because of the 10% size presumably? They can't run more than 10%
> slower. And probably much less as NOPs take minimal time to execute.
I actually don't belive there will be a measuable slowdown on 486's. (386s
you might see a small reduction) The selection of instructions does not
change that dramatically, the main effects are ordering and alignment,
both of which have no effect on a 486.
Jason
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: