Peter Cordes wrote:
>
> On Fri, Jun 15, 2001 at 01:56:45PM +0200, Michel D?nzer wrote:
> >
> >
> > My $CFLAGS looks like this: -pipe -O2 -Wall -mcpu=750 -fstrict-aliasing
> >
> > It's not entirely clear from the gcc docs if code generated this way runs
> > on older CPUs than G3s though. Should I rather use -mtune=750? What would
> > be optimal for -mcpu for all CPUs we support?
> >
> >
> > I wonder if anyone running on something like 604 or 603 has had strange
> > problems like illegal instructions with packages I built?
>
> gcc for IA32 uses -march=... to control what instructions it is
> allowed to use, and what arch to tune for. You can override the
> tuning target with -mcpu=.... Specifying -mcpu=... without -march
> still generates code that will run on a 386.
>
> I don't know how things are for PPC, but unless things are way
> different, -mcpu=... shouldn't change what instructions gcc restricts
> itself to.
>From reading the gcc texinfo, I understand it might:
`-mcpu=CPU_TYPE'
Set architecture type, register usage, choice of mnemonics, and
instruction scheduling parameters for machine type CPU_TYPE.
It would be great if we could put this straight:
What set of CPUs do we officially support? I know of 603, 604, G3, G4,
probably 601. More?
What's the 'greatest common divider' -mcpu option for these? -mcpu=powerpc?
--
Earthling Michel Dänzer (MrCooper) \ Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast \ XFree86 and DRI project member
