Am 27.06.2017 um 21:47 schrieb Michael Niedermayer:
> 0.5% faster loop
>
> Signed-off-by: Michael Niedermayer <mich...@niedermayer.cc>
> ---
> libavcodec/utvideodec.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/utvideodec.c b/libavcodec/utvideodec.c
> index 788f4475b9..a20e28320c 100644
> --- a/libavcodec/utvideodec.c
> +++ b/libavcodec/utvideodec.c
> @@ -196,7 +196,8 @@ static int decode_plane10(UtvideoContext *c, int plane_no,
>
> prev = 0x200;
> for (j = sstart; j < send; j++) {
> - for (i = 0; i < width * step; i += step) {
> + int ws = width * step;
> + for (i = 0; i < ws; i += step) {
> pix = get_vlc2(&gb, vlc.table, VLC_BITS, 3);
This seems like a pretty obvious optimization, why doesn't gcc or any
other compiler see that it can pre-compute that before the loop, as
nothing in the loop changes both variables?
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
