]> granicus.if.org Git - libvpx/commit
projects / libvpx / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 4626faf) | patch
Optimize 16x16 dequant and idct
authorYunqing Wang <yunqingwang@google.com>
Wed, 7 Nov 2012 00:06:22 +0000 (16:06 -0800)
committerYunqing Wang <yunqingwang@google.com>
Thu, 8 Nov 2012 04:04:09 +0000 (20:04 -0800)
commit6c17c9fae0f54942abe480414b4c0db7b2476412
treec9fc78a93e43b05213608c45f0232f51ffa8af83tree | snapshot
parent4626faf1e75ab229552d626db07c18e122971096commit | diff
Optimize 16x16 dequant and idct

As suggested by Yaowu, simplified 16x16 dequant and idct. In decoder,
after detoken step, we know the number of non-zero dct coefficients
(eobs) in a macroblock. Idct calculation can be skipped or simplified
based on eobs, which improves the decoder performance.

Change-Id: I9ffa1cb134bcb5a7d64fcf90c81871a96d1b4018
vp9/common/generic/systemdependent.c diff | blob | history
vp9/common/idct.h diff | blob | history
vp9/common/idctllm.c diff | blob | history
vp9/common/rtcd_defs.sh diff | blob | history
vp9/decoder/decodframe.c diff | blob | history
vp9/decoder/dequantize.c diff | blob | history
Unnamed repository; edit this file 'description' to name the repository.