Given Athlon's success there is no way RedHat could use -march=pentium4 without alienating a _large_ share of their user base. But they could use -mcpu=pentium4 instead of -mcpu=i686 as the Pentium3s are being phased out. This would be faster on P4 and would stll run athlons. Also given that the only remaining "useful" processors have MMX capabilities (except for the PentiumPro who never was around in large numbers) RedHat could and should use: -mcpu=pentium4 -mmmx Of course all the discussion above assumes: a) that Gcc P4 or MMX code generation is not as buggy as hell b) that it makes a real difference on the P4 (don't assume it beforehand: it depnds on how much manpower has gone into that part of Gcc: at one point I found that the code genartion for the K6 was so bad that programs "optimized" for it were slower than whne compiled for the 686) c) That it does not take an unordinate performance hit on the PIII and Athlons. Probable since AFAIK both are far less picky than the P4 On Sat, 2003-07-19 at 16:16, Steve Snyder wrote: > On Saturday 19 July 2003 6:53 am, Chris Adams wrote: > > Once upon a time, Audioslave - 7M3 - Live <creed7m3live@xxxxxxxxxxxxxxx> > said: > > > I hope that the programs start to be compiled for at least 586 and > > > higher applications. (maybe more efficient programs) > > > > Red Hat has been compiling the packages that matter for specific > > architectures for years. There isn't any point in compiling > > _everything_ for anything other than i386; do you think "ls" or "vi" > > will run faster somehow? Also, everything is built optimized for i686 > > but using only the i386 instruction set. That way, it is optimized for > > higher CPUs (affects memory and instruction layout), but will run on > > any x86 CPU. > > I was just about to write the same thing. > > [RANT] > Now then: where's my Pentium4-optimized RPMs? The 1990s are long gone and > GCC v2.x with it. It's a shame that the "-march=pentium4" and "-msse2" > switches in GCC v3.x are going unused in a major distro like Red Hat's. > [/RANT] -- Jean Francois Martinez <jfm512@xxxxxxx> -- Shrike-list mailing list Shrike-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/shrike-list
