Jason Garrett-Glaser <[email protected]> writes:

> On Wed, Jun 27, 2012 at 12:29 PM, Daniel Kang <[email protected]> wrote:
>> Now, memory is allocated for each macroblock in the frame.
>>
>> Previously, there was an optimization to only need width+height*2+1
>> macroblocks allocated.

That sounds like it could easily hurt cache performance.

> I strongly dislike this patch.  Can we implemented sliced threading --
> a rarely-used feature -- without slowing down the most common case?

+1

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

Reply via email to