[FFmpeg-devel] FATE & Regressions (and PPC is broken)
Sat Mar 15 18:23:45 CET 2008
On Sat, 15 Mar 2008, Diego Biurrun wrote:
> On Sat, Mar 15, 2008 at 02:24:41AM +0100, Michael Niedermayer wrote:
>> We have to use lrint* because its faster see:
> Hmm, does speed really matter that much in that part of the WMA encoder?
> Or is that just a comment for the general case and Mans' patch is OK?
Speed does matter there, so dsp.float_to_int16 would be better than lrint.
More information about the ffmpeg-devel