Athlon64 3200+ make.conf settings

[Andrew Pantyukhin]

On 11/3/06, Eric <heli at mikestammer.com> wrote:
> Andrew Pantyukhin wrote:
> >
> > Yes, very nice :-)
> >
> > Beware that in some rare cases you might experience
> > software failures because of CPUTYPE being set.
> > It's not very clear whether it's worth having a
> > marginal boost in performance along with some
> > instability. I usually gamble on performance, but
> > YMMV.
>
> ok cool.
>
> One more question and I will leave you alone. when i do a test compile i
> am seeing this:
>
> Build iozone for FreeBSD
>
> cc -c -O2 -fno-strict-aliasing -pipe -march=athlon-mp   -Dunix -Dbsd4_2
> -DHAVE_ANSIC_C  -DNAME='"freebsd"' -DSHARED_MEM iozone.c -o iozone_freebsd.o
> iozone.c: In function `initfile':
> iozone.c:16548: warning: passing arg 2 of `mmap' makes integer from
> pointer without a cast
> cc -c -O2 -fno-strict-aliasing -pipe -march=athlon-mp  -Dunix -Dbsd4_2
> -DHAVE_ANSIC_C  -DSHARED_MEM libbif.c -o libbif.o
>
> Building fileop for FreeBSD
>
> cc -c -O -O2 -fno-strict-aliasing -pipe -march=athlon-mp  fileop.c -o
> fileop_freebsd.o
>
>
> is march correct? did the compiler determine athlon-mp due to what i
> added or should that say athlon64?

It happens, some cputypes are just aliases, though
I thought athlon-mp and athlon64 are in different
groups. More info on type groups here:
http://gcc.gnu.org/onlinedocs/gcc/i386-and-x86_002d64-Options.html