[FFmpeg-devel] [RFC]lswr/rematrix: Support s32p

[Michael Niedermayer]

On Tue, Oct 31, 2017 at 12:40:24AM +0100, Carl Eugen Hoyos wrote:
> 2017-10-30 9:47 GMT+01:00 Muhammad Faiz <mfcc64 at gmail.com>:
> > On Sun, Oct 29, 2017 at 3:55 AM, Carl Eugen Hoyos <ceffmpeg at gmail.com> wrote:
> >> Hi!
> >>
> >> Attached patch fixes a random testcase for ticket #6785 here but I
> >> don't know if this is the correct fix.
> >>
> >> Please review, Carl Eugen
> >>
> >> From a93b9309d74f5eadece371ee1e682d266af6cd83 Mon Sep 17 00:00:00 2001
> >> From: Carl Eugen Hoyos <ceffmpeg at gmail.com>
> >> Date: Sat, 28 Oct 2017 22:52:02 +0200
> >> Subject: [PATCH] lswr/rematrix: Support s32p.
> >>
> >> Fixes ticket #6785.
> >> ---
> >>  libswresample/rematrix.c |   21 +++++++++++++++------
> >>  1 file changed, 15 insertions(+), 6 deletions(-)
> >>
> >> diff --git a/libswresample/rematrix.c b/libswresample/rematrix.c
> >> index 66a43c1..a6aa6b0 100644
> >> --- a/libswresample/rematrix.c
> >> +++ b/libswresample/rematrix.c
> >> @@ -445,14 +445,23 @@ av_cold int swri_rematrix_init(SwrContext *s){
> >>          s->mix_2_1_f = (mix_2_1_func_type*)sum2_double;
> >>          s->mix_any_f = (mix_any_func_type*)get_mix_any_func_double(s);
> >>      }else if(s->midbuf.fmt == AV_SAMPLE_FMT_S32P){
> >> -        // Only for dithering currently
> >> -//         s->native_matrix = av_calloc(nb_in * nb_out, sizeof(double));
> >> -        s->native_one    = av_mallocz(sizeof(int));
> >> +        s->native_one    = av_mallocz(sizeof(int64_t));
> >>          if (!s->native_one)
> >>              return AVERROR(ENOMEM);
> >> -//         for (i = 0; i < nb_out; i++)
> >> -//             for (j = 0; j < nb_in; j++)
> >> -//                 ((double*)s->native_matrix)[i * nb_in + j] = s->matrix[i][j];
> >> +        s->native_matrix = av_calloc(nb_in * nb_out, sizeof(int));
> >> +        if (!s->native_matrix) {
> >> +            av_freep(&s->native_one);
> >> +            return AVERROR(ENOMEM);
> >> +        }
> >> +        for (i = 0; i < nb_out; i++) {
> >> +            double rem = 0;
> >> +
> >> +            for (j = 0; j < nb_in; j++) {
> >> +                double target = s->matrix[i][j] * 32768 + rem;
> >> +                ((int*)s->native_matrix)[i * nb_in + j] = lrintf(target);
> >> +                rem += target - ((int64_t*)s->native_matrix)[i * nb_in + j];
> >> +            }
> >> +        }
> >>          *((int*)s->native_one) = 32768;
> >>          s->mix_1_1_f = (mix_1_1_func_type*)copy_s32;
> >>          s->mix_2_1_f = (mix_2_1_func_type*)sum2_s32;
> >
> > The code is confusing.
> > Which is the type of native_matrix and native_one? int or int64_t?
> 
> New patch attached.
> 
> Please review, Carl Eugen

>  rematrix.c |   21 +++++++++++++++------
>  1 file changed, 15 insertions(+), 6 deletions(-)
> 9f50e61370cc72de000bb1174cbd8666388fdd9a  0001-lswr-Support-s32p.patch
> From a93b9309d74f5eadece371ee1e682d266af6cd83 Mon Sep 17 00:00:00 2001
> From: Carl Eugen Hoyos <ceffmpeg at gmail.com>
> Date: Sat, 28 Oct 2017 22:52:02 +0200
> Subject: [PATCH] lswr/rematrix: Support s32p.
> 
> Fixes ticket #6785.
> ---
>  libswresample/rematrix.c |   21 +++++++++++++++------
>  1 file changed, 15 insertions(+), 6 deletions(-)

should be ok

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Complexity theory is the science of finding the exact solution to an
approximation. Benchmarking OTOH is finding an approximation of the exact
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20171031/75e4d326/attachment.sig>