On Fri, Jun 02, 2000 at 12:19:51PM +1000, Martin Bishop wrote: > > If I understand you correctly, adding a switch (-march=pentiumpro / i686) > to gcc at compile time is meaningless for 99.9% of programs? AFAIK there is no optimization switch for anything but plain pentium, which does not buy much for newer chips (or for non-intel chips) > It doesn't make sense for the gcc people to add a switch that does > nothing 99.9% of the time, plus the increased unreliability factor. > > Can you explain a little bit more or point to docs that can on this > issue? This is one of the thing I'm curious about. see list archives for (maybe) this list and definitly debian-devel there has been much discussion (flame war) about it in more detail then i can give. keywords to look are are pgcc the package that automatically enables pentium optimiations in the local compiler. pgcc is known to cause some software to break, (namly the linux kernel) due to the ways the optimiztion interacts with some C code. -- Ethan Benson http://www.alaska.net/~erbenson/
Attachment:
pgprdS7bPiBkW.pgp
Description: PGP signature