On Tue, Feb 27, 2018 at 9:35 PM, David Murmann <david.murm...@btf.de> wrote:
> Quantization scaling seems to be a slight bottleneck,
> this change allows the compiler to more easily vectorize
> the loop. This improves total encoding performance in my
> tests by about 10-20%.
>
> Signed-off-by: David Murmann <da...@btf.de>
> ---
> libavcodec/proresenc_anatoliy.c | 12 ++++++++----
> 1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/libavcodec/proresenc_anatoliy.c
> b/libavcodec/proresenc_anatoliy.c
> index 0516066163..8b296f6f1b 100644
> --- a/libavcodec/proresenc_anatoliy.c
> +++ b/libavcodec/proresenc_anatoliy.c
> @@ -232,14 +232,18 @@ static const uint8_t lev_to_cb[10] = { 0x04, 0x0A,
> 0x05, 0x06, 0x04, 0x28,
> static void encode_ac_coeffs(AVCodecContext *avctx, PutBitContext *pb,
> int16_t *in, int blocks_per_slice, int *qmat)
> {
> + int16_t block[64];
> int prev_run = 4;
> int prev_level = 2;
> int run = 0, level, code, i, j;
> - for (i = 1; i < 64; i++) {
> - int indp = progressive_scan[i];
> - for (j = 0; j < blocks_per_slice; j++) {
> - int val = QSCALE(qmat, indp, in[(j << 6) + indp]);
> + for (j = 0; j < blocks_per_slice; j++) {
> + for (i = 0; i < 64; i++) {
> + block[i] = (float)in[(j << 6) + i] / (float)qmat[i];
> + }
> +
> + for (i = 1; i < 64; i++) {
> + int val = block[progressive_scan[i]];
> if (val) {
> encode_codeword(pb, run, run_to_cb[FFMIN(prev_run, 15)]);
Usually, using float is best avoided. Did you test re-factoring the
loop structure without changing it to float?
- Hendrik
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
