On Mon, Oct 19, 2015 at 10:00 PM, Timothy Gu wrote:
> About 16% faster on large clips (>1200px width), more than 2x slower on small
> clips
> (352px).
The reason is for this is likely the fact that you fall back to scalar
as soon as you have less than 2*mmsize bytes left to process which
leads t
On Mon, Oct 19, 2015 at 1:44 PM Ganesh Ajjanagadde wrote:
> On Mon, Oct 19, 2015 at 4:41 PM, Hendrik Leppkes
> wrote:
> > On Mon, Oct 19, 2015 at 10:36 PM, Ganesh Ajjanagadde
> wrote:
> >> On Mon, Oct 19, 2015 at 4:00 PM, Timothy Gu
> wrote:
> >>> About 16% faster on large clips (>1200px width
On Mon, Oct 19, 2015 at 4:41 PM, Hendrik Leppkes wrote:
> On Mon, Oct 19, 2015 at 10:36 PM, Ganesh Ajjanagadde wrote:
>> On Mon, Oct 19, 2015 at 4:00 PM, Timothy Gu wrote:
>>> About 16% faster on large clips (>1200px width), more than 2x slower on
>>> small clips
>>> (352px). So using a heurist
On Mon, Oct 19, 2015 at 4:00 PM, Timothy Gu wrote:
> About 16% faster on large clips (>1200px width), more than 2x slower on small
> clips
> (352px). So using a heuristic to select with one to use.
What system, what compiler, etc? Without any such information, numbers
are meaningless. Please eit
On Mon, Oct 19, 2015 at 10:36 PM, Ganesh Ajjanagadde wrote:
> On Mon, Oct 19, 2015 at 4:00 PM, Timothy Gu wrote:
>> About 16% faster on large clips (>1200px width), more than 2x slower on
>> small clips
>> (352px). So using a heuristic to select with one to use.
>
> What system, what compiler, e
About 16% faster on large clips (>1200px width), more than 2x slower on small
clips
(352px). So using a heuristic to select with one to use.
---
libavcodec/huffyuvenc.c| 6 +++---
libavcodec/huffyuvencdsp.c | 4 ++--
libavcodec/huffyuvencdsp.h | 4 ++--
libavcodec/pnge
