[power-gnu-discuss] Support for the MPC5554 (e200z6 core)

[Nathan Froyd]

On Tue, Jun 02, 2009 at 11:59:02PM +0200, David Brown wrote:
> Nathan Froyd wrote:
> >GCC does support generating SPE instructions for hardware floating
> >point.  You need to compile with the "-te500v1" flag, which enables the
> >compiler to use single-precision floating-point SPE instructions.  You
> >should not use additional "-mcpu=" flags on the command-line along with
> >"-te500v1".
> 
> Marvellous!  This looks like gcc is now generating exactly the code I 
> want - SPE instructions for single precision, and library code for 
> double precision (unless -fshort-double is in effect).

Glad to hear it.

> I can't see the -te500v1 flag documented anywhere - is it a new flag 
> that hasn't yet made it into the documentation?

The flag is at least mentioned in the Getting Started Guide that comes
with each release, although I see that the documentation for the option
is rather sparse (e.g. no mention is made of generating SPE instructions
for -te500v1).  We'll work on fixing that.

-Nathan