On 07/22/2012 05:30 AM, Anton Khirnov wrote: > It internally decodes as float and then converts to s16 by a call to > float_to_int16(). The caller can do this just as well by using lavr. > --- > libavcodec/nellymoserdec.c | 31 ++++--------------------------- > 1 file changed, 4 insertions(+), 27 deletions(-)
Such basic conversions in lavr are not optimized for ARM yet, but float_to_int16() in dsputil is, so it might be a good idea to wait until these are ported/implemented unless you can test the performance impact on ARM. As a side note, there might be a better place to apply the scale bias, like in the MDCT or the sine window. But the patch looks ok in general in terms of functionality. -Justin _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
