Re: [Performance] 386vs486 compilation
> > From: Jason Gunthorpe <jgg@gpu.srv.ualberta.ca>
> > > Have you ever seen numbers for running Pentium code on a 486 and 386 vs
> > > native code?
> >
> > I think the Pentium code generator now uses Pentium-specific instructions.
> > Before then, Pentium code on 386 and 486 sometimes ran faster than the 386
> > and 486 code, because it avoided some instructions that turned out to be slow.
>
> Yes, this is what I usually hear about most compilers. P5 Code runs as
> fast or faster on 486's. What GCC needs is a to have the 486 generator
> perform pentium optimizations and not use pentium/386 instructions. Then
> we could compile everything for this. The speed hit on a pentium would
> drop on alot of apps and the 486/386 would not have any hit..
Forgive me for picking a nit...
The reason P5-optimized code runs faster on a 486 than 486-optimized code
has to do with the much improved general optimizations of the P5 code
generator.
Logically, 486-optimized code will _always_ run faster on a 486 than
P5-optimized code. If that is not the case, then the 486-optimized code
wasn't optimal after all.
> I wonder what instructions gcc uses on the 586, I don't recall seeing that
> many usefull ones for general purpose stuff..
Last I heard, gcc didn't use anything but instructions available on the 386.
Brian
( bcwhite@verisim.com )
-------------------------------------------------------------------------------
Give others some insight into YOUR pages! http://www.verisim.com/insite/
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-user-request@lists.debian.org .
Trouble? e-mail to templin@bucknell.edu .
Reply to: