SSE conversion optimization

H.J. Lu hjl@lucon.org
Sun Sep 9 00:13:00 GMT 2007


On Sun, Sep 09, 2007 at 01:35:07AM +0200, Jan Hubicka wrote:
> > On Sun, Sep 09, 2007 at 12:02:28AM +0200, Jan Hubicka wrote:
> > > Hi,
> > > Amdfam10 preffers doing packed conversions destinating SSE register rather than scalar.
> > > This means basically following replacments:
> > > 
> > > -      cvtsi2sd -> movd + cvtdq2pd
> > > -      cvtsi2ss -> movd + cvtdq2ps
> > 
> > Can you disable them for -mtune=generic if an extra pair of
> > memory load/store is added?
> 
> You mean doing it only for conversion from memory operand_  Yes, that
> should be quite easy, I will send you updated patch.
> 

Yes. I will test your new patch on SPEC CPU 2000/2006. I should
have the results within a week.

Thanks.


H.J.



More information about the Gcc-patches mailing list