[FFmpeg-devel] [PATCH] mmx implementation of vc-1 inverse transformations

[Michael Niedermayer]

On Fri, Aug 01, 2008 at 06:40:08PM +0200, Victor Pollex wrote:
> Michael Niedermayer schrieb:
>> On Thu, Jul 31, 2008 at 02:50:44PM +0200, Victor Pollex wrote:
>>   
>>> Michael Niedermayer schrieb:
>>>     
>
> [...]
>
>>>>> +void ff_vc1dsp_init_sse2(DSPContext* dsp, AVCodecContext *avctx) {
>>>>> +    if(!(mm_flags & MM_SSE2))
>>>>> +        return;
>>>>> +
>>>>> +    dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_sse2;
>>>>> +    dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_sse2;
>>>>> +    dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_sse2;
>>>>> +}
>>>>>             
>>>> are all of the SSE2 variants faste than mmx?
>>>>         
>>> For me the 8x8 sse2 variant is faster than the mmx one, but as I I #
>>> metioned in an earlier post, the 4x8 isn't and the 8x4 is only a bit 
>>> faster, that is why I asked if someone else could benchmark them, to see 
>>> if they behave like that just for me.
>>>     
>>
>> In the anbsense of anyone posting benchmark we only have yours that says
>> that one isnt faster so it should be disabled, it could be kept in the 
>> file
>> under #if 0 for others to test ...
>>   
>
> I put it under #if 0

Could you please ommit patches that have already been approved, it makes
review easier if i do not have to check the thread and compare patches

Also the 8x8 transform either has to be optimal or removed from the patch.
The suboptimal one, that does a redundant transpose will not be applied
to svn.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If you really think that XML is the answer, then you definitly missunderstood
the question -- Attila Kinali
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20080826/b5f8ccad/attachment.pgp>