[FFmpeg-devel] [FFmpeg-cvslog] avfilter/af_acrossover: use fdsp for inverter

James Almer jamrial at gmail.com
Mon Nov 30 15:18:04 EET 2020


> ffmpeg | branch: master | Paul B Mahol <onemda at gmail.com  <https://ffmpeg.org/mailman/listinfo/ffmpeg-cvslog>> | Mon Nov 30 13:45:08 2020 +0100| [6100a01a7ad939d8069b011a29eb8db90aa68aba] | committer: Paul B Mahol
>
> avfilter/af_acrossover: use fdsp for inverter
>
> >/http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=6100a01a7ad939d8069b011a29eb8db90aa68aba 
> <http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=6100a01a7ad939d8069b011a29eb8db90aa68aba> 
> /---
>
>   libavfilter/af_acrossover.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/libavfilter/af_acrossover.c b/libavfilter/af_acrossover.c
> index 9b4e79a653..7788251d3a 100644
> --- a/libavfilter/af_acrossover.c
> +++ b/libavfilter/af_acrossover.c
> @@ -367,9 +367,9 @@ static int filter_channels_## name(AVFilterContext *ctx, void *arg, int jobnr, i
>           for (int band = 0; band < ctx->nb_outputs && s->first_order; band++) {              \
>               if (band & 1) {                                                                 \
>                   type *dst = (type *)frames[band]->extended_data[ch];                        \
> -                                                                                            \
> -                for (int n = 0; n < nb_samples; n++)                                        \
> -                    dst[n] *= -one;                                                         \
> +                s->fdsp->vector_## ff ##mul_scalar(dst, dst, -one,                          \
> +                                                   FFALIGN(nb_samples, sizeof(type)));      \
> +                emms_c();                                                                   \

The only SIMD implementation for this function is SSE. No need to call 
emms for it.
Only vector_fmul_window needs it since there's a 3dnow implementation 
(That maybe we could just remove).



More information about the ffmpeg-devel mailing list