[FFmpeg-devel] [PATCH] DCA floating point output

[Kostya]

On Mon, Apr 26, 2010 at 09:32:30AM +0800, FB2000 wrote:
> OK. Corrected.
 
Looking at the discussion here, patch is rejected until there's some
more or less unanimous opinion on floating-point output for generic
decoders.

And conversion loop like:

> -        s->dsp.float_to_int16_interleave(samples, s->samples_chanptr, 256,
> channels);
> -        samples += 256 * channels;
> +        for (j = 0; j < 256; j++) {
> +            for (k = 0; k < channels; k++)
> +                samples[k] = s->samples[j + k * 256] / 32768.0f;
> +            samples += channels;
> +        }

tells me that something is not quite right - I'd prefer having memcpy()
at most instead of this loop. Scaling can be merged somewhere and maybe
planar audio output will become a reality one day.

For now, sorry, no luck.