Justin Ruggles <[email protected]> writes:
> Eliminates the need for vector_fmul_scalar() in each frame.
> ---
> libavcodec/binkaudio.c | 8 ++++----
> 1 files changed, 4 insertions(+), 4 deletions(-)
I'd prefer something like "coeff scale" instead of s->root in the commit
message.
> diff --git a/libavcodec/binkaudio.c b/libavcodec/binkaudio.c
> index 31a6a7c..915e7aa 100644
> --- a/libavcodec/binkaudio.c
> +++ b/libavcodec/binkaudio.c
> @@ -47,7 +47,6 @@ static float quant_table[96];
> typedef struct {
> AVFrame frame;
> GetBitContext gb;
> - DSPContext dsp;
> FmtConvertContext fmt_conv;
> int version_b; ///< Bink version 'b'
> int first;
> @@ -79,7 +78,6 @@ static av_cold int decode_init(AVCodecContext *avctx)
> int i;
> int frame_len_bits;
>
> - ff_dsputil_init(&s->dsp, avctx);
> ff_fmt_convert_init(&s->fmt_conv, avctx);
>
> /* determine frame length */
> @@ -112,7 +110,10 @@ static av_cold int decode_init(AVCodecContext *avctx)
> s->overlap_len = s->frame_len / 16;
> s->block_size = (s->frame_len - s->overlap_len) * s->channels;
> sample_rate_half = (sample_rate + 1) / 2;
> - s->root = 2.0 / sqrt(s->frame_len);
> + if (avctx->codec->id == AV_CODEC_ID_BINKAUDIO_RDFT)
> + s->root = 2.0 / sqrt(s->frame_len);
> + else
> + s->root = s->frame_len / sqrt(s->frame_len);
> for (i = 0; i < 96; i++) {
> /* constant is result of 0.066399999/log10(M_E) */
> quant_table[i] = expf(i * 0.15289164787221953823f) * s->root;
> @@ -259,7 +260,6 @@ static int decode_block(BinkAudioContext *s, int16_t
> *out, int use_dct)
> if (CONFIG_BINKAUDIO_DCT_DECODER && use_dct) {
> coeffs[0] /= 0.5;
> s->trans.dct.dct_calc(&s->trans.dct, coeffs);
> - s->dsp.vector_fmul_scalar(coeffs, coeffs, s->frame_len / 2,
> s->frame_len);
> }
> else if (CONFIG_BINKAUDIO_RDFT_DECODER)
> s->trans.rdft.rdft_calc(&s->trans.rdft, coeffs);
> --
LGTM
--
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
