[MPlayer-dev-eng] [PATCH] yadif SSE2/SSSE3 optimization

Baptiste Coudurier baptiste.coudurier at gmail.com
Sat Nov 15 01:31:40 CET 2008


Hi Michael,

Michael Niedermayer wrote:
> On Thu, Nov 13, 2008 at 08:22:50PM +0800, Zhou, Zongyi wrote:
>> Hi all, 
>>
>> I made this patch from ffdshow-mt branch. The original SSE2/SSSE3 codes are written by h.yamagata.
>> I ported those codes back to mplayer and replaced all movdqa/movdqu with movaps/movups.
>>
>> My tests show that on Intel CPUs (except Yonah), SSE2 is ~20% faster than MMX2 and SSSE3 is 30% faster than MMX2.
>> However on AMD CPUs (except Socket 754 Semprons), SSE2 is ~5% slower than MMX2.
>> So now SSE2 function is used only on Intel CPUs.
>>

When will we have yadif in libavfilter ? This filter is awesome !
What's left to do ?

[...]

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
checking for life_signs in -lkenny... no



More information about the MPlayer-dev-eng mailing list