[x264-devel] Re: [patch] faster MMXEXT SATD

[Loren Merritt]

On Wed, 21 Sep 2005, Christian Heine wrote:

> the attached patch contains new MMXEXT SATD functions that are slightly 
> faster than the original ones on Athlon XP.

> x264_pixel_satd_16x16_mmxext:
...
> pxor        mm3, mm3
> pshufw      mm1, mm0, 01001110b
> paddw       mm0, mm1
> punpcklwd   mm0, mm3
> pshufw      mm1, mm0, 01001110b
> paddw       mm0, mm1

paddd, right? The output can take up to 17 bits.

> movd        eax, mm0
> pop         ebx
> ret

--Loren Merritt

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html