[PATCH][libgcc-math] Vectorized intrinsics for x86_64
Fri Apr 7 15:37:00 GMT 2006
On Fri, 7 Apr 2006, H. J. Lu wrote:
> On Fri, Apr 07, 2006 at 03:56:02PM +0200, Richard Guenther wrote:
> > On Fri, 7 Apr 2006, H. J. Lu wrote:
> > > The code is specific to Opteron. Shouldn't it be neutral to Opteron
> > > and Nocona?
> > Can you elaborate?
> My first quick glance shows that at least there is an issue of choices
> of movlpd vs. movsd and movsd vs. movapd, which are very different
> bewteen Opteron and Nocona.
Ah yes, that difference. We probably can change this if the performance
hit is not too high.
More information about the Gcc-patches