Michael wrote: > > On Mon, Mar 04, 2002 at 06:51:10PM +0100, Dieter Nützel wrote: > > On Monday, 4. März 2002 16:30:33, Thomas Dodd wrote: > > > Dieter Nützel wrote: > > > > On Sonntag, 3. März 2002 18:09:18, Gareth Hughes wrote: > > > >> Dieter Nützel wrote: > > > > > > >> > CDEBUGFLAGS = -O -mcpu=k6 -pipe -mpreferred-stack-boundary=2 > > > >> > -malign-functions=4 -fschedule -insns2 -fexpensive-optimizations > > > >> > > > >> Hmmm, I'm not really sure that's legal (in terms of having reportable > > > >> results with a binary compiled like that) :-) > > > > > > > > He asked about self compiling...;-))) > > > > > > Are these with the default CFLAGS or the K7 optimized? > > > > My K7. I found them during "heavy" Mesa testing since 26. August 1999...;-) > > Running all Linux kernels with them. > > What speedup do you get? I'd have thought that in both cases > (mesa and kernel) that the biggest gains are already done explicitly, > with things like 3dnow and fast_copy/clear_page? > > Have you tried gcc-3's --march=athlon? > > (which as an aside intrigued me, is MESA_NO_3DNOW required to run > veiwperf 'legally'? What exactly is cheating about such general purpose > optimisations like mcpu=k6?)
I don't really see it as cheating either. I'm sure that when compaq runs it on alpha they use uptodate instruction sets, likewise for Sun on sparc. Keith _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel