This code is in a performance critical section. AV_QSORT is substantially faster due to the inlining of the comparison callback. Thus, the increase in performance is worth the increase in binary size.
Signed-off-by: Ganesh Ajjanagadde <gajjanaga...@gmail.com> ------------------------------------------------------------------------------- Benchmarks not provided, since FATE does not seem to exercise this. Left to the maintainer/any interested. --- libavcodec/motion_est_template.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/libavcodec/motion_est_template.c b/libavcodec/motion_est_template.c index 37ff110..327a24b 100644 --- a/libavcodec/motion_est_template.c +++ b/libavcodec/motion_est_template.c @@ -24,6 +24,7 @@ * Motion estimation template. */ +#include "libavutil/qsort.h" #include "mpegvideo.h" //Let us hope gcc will remove the unused vars ...(gcc 3.2.2 seems to do it ...) @@ -723,7 +724,7 @@ static int sab_diamond_search(MpegEncContext * s, int *best, int dmin, j++; } - qsort(minima, j, sizeof(Minima), minima_cmp); + AV_QSORT(minima, j, Minima, minima_cmp); for(; j<minima_count; j++){ minima[j].height=256*256*256*64; -- 2.6.2 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel