projects / platform / upstream / libvpx.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b800ec1) | patch
Optimize the scaling calculation
authorYunqing Wang <yunqingwang@google.com>
Thu, 4 Apr 2013 16:56:02 +0000 (09:56 -0700)
committerYunqing Wang <yunqingwang@google.com>
Tue, 16 Apr 2013 15:52:40 +0000 (08:52 -0700)
commit148eb803bbb3de3a8548445fe50c120ea872a882
tree6e5fdcab8fda42c1c4fef48884cc429b6837c91dtree | snapshot
parentb800ec1772b7d84c9c8272f4dab4661c9a696dedcommit | diff
Optimize the scaling calculation

In decoder, the scaling calculation, such as (mv * x_num / x_den),
is fairly time-consuming. In this patch, we check if the scaling
happens or not at frame level, and then decide which function to
call to skip scaling calculation when no scaling is needed. Tests
showed a 3% decoder performance gain.

Change-Id: I270901dd0331048e50368cfd51ce273dd82b8733
vp9/common/vp9_blockd.h diff | blob | history
vp9/common/vp9_reconinter.c diff | blob | history
vp9/common/vp9_reconinter.h diff | blob | history
Domain: Multimedia / Codec;