[x264-devel] add optimised functions (zigzag_scan_4x4)

[Guillaume POIRIER]

Hello,

On Dec 8, 2007 9:37 PM, Guillaume Poirier <gpoirier at mplayerhq.hu> wrote:
> Hello,
>
> Le 8 déc. 07 à 11:59, Noboru Asai a écrit :
>
>
> > These patches add the following optimised functions accelerated
> > with altivec
> > and test functions.
> >
> > o x264_zigzag_scan_4x4_frame_altivec()
> > o x264_zigzag_scan_4x4ac_frame_altivec()
> > o x264_zigzag_scan_4x4_field_altivec()
> > o x264_zigzag_scan_4x4ac_field_altivec()
> >
> > Any comments are welcome.
> >
> > Noboru Asai
> > <patch-add-altivec-zigzag_scan_4x4>
>
> That one looks Ok to me. Do you have typical speed-up over the C
> version, and typical overall speed-up for a given set of encoding
> options?

I measured them to be between 1.3 to 1.8x faster than C version.

Patch applied.


> > <patch-add-check-zigzag>
>
> That one looks Ok to me too, but I'd like to have Loren's comment on
> it since it touches common code.
>
> Unless he disagrees, I'm tempted to apply in the middle of next week.

Still no comment? Well, I assume it's Ok. Unless you disagree Loren,
I'll apply it Wednesday.

Guillaume
-- 
A soldier will fight long and hard for a bit of colored ribbon.
 -- Napoleon Bonaparte