Hi,

On Wed, Apr 13, 2011 at 4:08 PM, Alexander Strange
<[email protected]> wrote:
> On Apr 13, 2011, at 11:35 AM, Ronald Bultje wrote:
>> @@ -1151,6 +1163,7 @@ static av_always_inline void 
>> prefetch_motion(VP8Context *s, VP8Macroblock *mb, i
>>          int my = (mb->mv.y>>2) + y_off;
>>          uint8_t **src= s->framep[ref]->data;
>>          int off= mx + (my + (mb_x&3)*4)*s->linesize + 64;
>> +        // FIXME does this need a ff_thread_await_progress()?
>>          s->dsp.prefetch(src[0]+off, s->linesize, 4);
>>          off= (mx>>1) + ((my>>1) + (mb_x&7))*s->uvlinesize + 64;
>>          s->dsp.prefetch(src[1]+off, src[2]-src[1], 2);
>
> Yes (if the reference thread is in another CPU). So you might want to follow 
> the h264 pattern.

I don't follow, that code is complex. What should it/I do exactly?

Ronald
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to