On 10/16/18, Moritz Barsnick <barsn...@gmx.net> wrote:
> On Sat, Oct 13, 2018 at 23:14:59 +0200, Paul B Mahol wrote:
>
> I see a few promotions from float to double arithmetic in here. Note
> that initializations don't matter, the preprocessor handles assigning a
> double to a float and downscales (casts) automatically. But
> float_var = float_var * double_const
> promotes. If that's undesired, a few constants need an 'f' qualifier.
>
> Example:
>
>> + float term = 1, sum = 1, last_sum, x2 = x / 2;
>
> The first two don't matter, but the latter turns into a CPU
> calculation.
>
>> + float mult = scale / bessel_I_0(beta), mult1 = 1 / (.5 * m + rho);
>
> -> 1.f / (.5f * m + rho)
>
>> + .description = NULL_IF_CONFIG_SMALL("Generate a sinc
>> kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR
>> coefficients."),
>
> Apart from the length of the string, I don't understand "Generate a ...
> coefficients". Is the "a" too much?
>
> Does it generate coefficients, or does it use them for something?
>
Once can use this coefficients with afir filter.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
