[FFmpeg-devel] [PATCH 2/2] libavcodec: Implementation of AC3 fixed point decoder

Christophe Gisquet christophe.gisquet at gmail.com
Wed Apr 10 12:43:57 CEST 2013


Hi,

2013/4/10 Nedeljko Babic <Nedeljko.Babic at imgtec.com>:
> Can someone please have a look at them? Should I make some additional changes to them?

Some questions, but that is far from a review.

1) If I haven't misread, you use a 20.12 FP format. It looks somewhat
realistic. Do you think this format will still fit other codecs you
might be working on?
Because otherwise, the FP dsp functions would probably work only for
that format, in which case the FP DSP context could hold somewhere
info about the format used, so that proper dsp functions are set.

2) Some places use it seems 15bits of decimal precision. Why is that?

3) Speed-wise, how does it compare to the float version on your target device?
You'd probably need to write optimized dsp functions first, I guess

4) Third, quality-wise, how do you validate the quality?
Something like outputting both float and FP decoders to 16bits signed,
and computing the MSE/max error?
Having some figures could be interesting.

5) Are there still some non-deterministic parts, like still using
floats, besides the precomputed tables?
ie, can the output be checked for exact matching via MD5?
If one is to write optimized dsp functions besides yours, this could
help check the validity of the optimizations.

I'm surprised only vector_fmul_window needs to be implemented in a
fixed-point way.

Overall, this looks like of limited intrusiveness and ok.

Thanks,
Christophe


More information about the ffmpeg-devel mailing list