On Thu, Feb 20, 2014 at 04:01:32PM +0100, Christophe Gisquet  wrote:
> Author:    Christophe Gisquet <[email protected]>
> Committer: Janne Grunau <[email protected]>
> Date:      Fri Feb 14 15:03:10 2014 +0000
> 
> dca: replace some memcpy by AV_COPY128
> 
> --- a/libavcodec/dcadec.c
> +++ b/libavcodec/dcadec.c
> @@ -1241,9 +1241,7 @@ static int dca_subsubframe(DCAContext *s, int 
> base_channel, int block_index)
>      /* Backup predictor history for adpcm */
>      for (k = base_channel; k < s->prim_channels; k++)
>          for (l = 0; l < s->vq_start_subband[k]; l++)
> -            memcpy(s->subband_samples_hist[k][l],
> -                   &subband_samples[k][l][4],
> -                   4 * sizeof(subband_samples[0][0][0]));
> +            AV_COPY128(s->subband_samples_hist[k][l], 
> &subband_samples[k][l][4]);

This breaks FATE on my K6-III:

tmp@silver:~/tmp/libav$ make fate-dts V=1
TEST    dts
./tests/fate-run.sh fate-dts "/home/tmp/fate-suite" "" "/home/tmp/tmp/libav" 
'pcm -i /home/tmp/fate-suite/dts/dts.ts' 'oneoff' 
'/home/tmp/fate-suite/dts/dts.pcm' '' '' '' '' '' '' '' '' ''
 /home/tmp/tmp/libav/avconv -nostats -cpuflags all -threads 1 -thread_type 
frame+slice -i /home/tmp/fate-suite/dts/dts.ts -vn -f s16le -
stddev:17145.22 PSNR: 11.65 MAXDIFF:40087 bytes:  1554432/  1554432
make: *** [fate-dts] Error 1

Diego
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to