]> git.sesse.net Git - x264/commit
projects / x264 / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 89a893a) | patch
Much faster CAVLC residual coding
authorFiona Glaser <fiona@x264.com>
Mon, 8 Dec 2008 21:44:23 +0000 (13:44 -0800)
committerFiona Glaser <fiona@x264.com>
Thu, 11 Dec 2008 06:48:18 +0000 (22:48 -0800)
commit99448f6c98289e74f1234e38b9ed2c945f2bdfca
tree71972910b2a4b6f889c02054dc71396d028cf15btree | snapshot
parent89a893a0b153cbc9fa5143aad15b17581b9b448bcommit | diff
Much faster CAVLC residual coding
Use a VLC table for common levelcodes instead of constructing them on-the-spot
Branchless version of i_trailing calculation (2x faster on Nehalem)
Completely remove array_non_zero_count and instead use the count calculated in level/run coding.  Note: this slightly changes output with subme > 7 due to different nonzero counts being stored during qpel RD.
common/bs.h diff | blob | history
common/common.h diff | blob | history
common/macroblock.h diff | blob | history
common/vlc.c diff | blob | history
common/x86/util.h diff | blob | history
encoder/cabac.c diff | blob | history
encoder/cavlc.c diff | blob | history
encoder/encoder.c diff | blob | history
encoder/macroblock.c diff | blob | history
Unnamed repository; edit this file 'description' to name the repository.