[FFmpeg-devel] Once again: Multithreaded H.264 decoding withffmpeg?

[Robert Swain]

2008/5/30 M?ns Rullg?rd <mans at mansr.com>:
> Goga777 <goga777 at bk.ru> writes:
>
>>> > Could it be realisitic to achieve lets say a speedup between 30% and 50%
>>> > on a modern dual core CPU or are there limitations that make this a
>>> > dream? What about using NVIDIA's CUDA or AMD's stream computing to make
>>> > use of modern GPUs (given that the "native" acceleration APIs are not
>>> > available under Linux)?
>>
>> yes, VAAPI under Linux doesn't finish yet. Nobody knows when VAAPI
>> will finish http://www.freedesktop.org/wiki/Software/vaapi
>>
>>> There is definitively room for 50% decode speed improvement in
>>> ffmpeg's h264 decoder but it would take some work. CUDA et al have
>>> been discussed in the past (again see archives) but the kist of it is
>>> that while very fast, the latency can be quite significant so you
>>> would have to do some fancy coding to hide that. The h264 acceleration
>>> 'featured' by most modern GPUs is actually a HW decoder and they are
>>> generally limited to Level 4.1 decode
>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>
>> what's mean ? with the HW decoding the pictures quality will not the best ?
>
> It means that they are unable to decode streams using certain
> features.

Actually, levels correspond to VBV sizes, macroblocks per second,
maximum MV lengths, etc., rather than codec features. :) Off the top
of my head, level 4.1 is adequate for 1080p30 at somewhere around Blu
Ray bit rates.

Rob