On 4/12/16, Paul B Mahol <one...@gmail.com> wrote:
> On 4/12/16, Carl Eugen Hoyos <ceho...@ag.or.at> wrote:
>> Paul B Mahol <onemda <at> gmail.com> writes:
>>
>>>
>>> Hi,
>>>
>>> patch attached.
>>
>> Very cool!
>>
>>> -        pred += s->dsp.scalarproduct_and_madd_int16(s->cdlms[ch]
>>> +        pred += scalarproduct_and_madd_int(s->cdlms[ch][ilms].coefs,
>>
>> Doesn't this lead to a speedloss for 16 bit decoding?
>
> It does, havent checked how much.

40x vs 72x realtime.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Reply via email to