[MPlayer-dev-eng] possible bugs in vf_decimate filter

[Guillaume POIRIER]

Hi,

On 10/17/06, Trent Piepho <xyzzy at speakeasy.org> wrote:


> I removed the "emms" from the asm block in the diff_MMX() function.  This code
> ended up slowing the filter down by around 20%!  It has nothing do with the
> extra variable 'r' or the goto.  If I add an extra emms right after the call to
> diff at (1), it speeds up to around the original speed (even with the extra emms
> as (2), the goto, and the variable 'r').

Aren't you using an intel P4 by any chance? On these, emms instruction
is very slow.
On AMD CPUs, it's a lot fast, all the more faster if you use femms.

Guillaume
-- 
With DADVSI (http://en.wikipedia.org/wiki/DADVSI), France finally has
a lead on USA on selling out individuals right to corporations!
Vive la France!