• Yunqing Wang's avatar
    Improve sad3x16 SSE2 function · e7cd8071
    Yunqing Wang authored
    Vp9_sad3x16_sse2() is heavily called in decoder, in which the
    unaligned reads consume lots of cpu cycles. When CONFIG_SUBPELREFMV
    is off, the unaligned offset is 1. In this situation,
    we can adjust the src_ptr to be 4-byte aligned, and then do the
    aligned reads. This reduced the reading time significantly. Tests
    on 1080p clip showed over 2% decoder performance gain with
    CONFIG_SUBPELREFM off.
    
    Change-Id: I953afe3ac5406107933ef49d0b695eafba9a6507
    e7cd8071
Name
Last commit
Last update
build Loading commit data...
examples Loading commit data...
libmkv Loading commit data...
nestegg Loading commit data...
test Loading commit data...
third_party Loading commit data...
tools Loading commit data...
vp8 Loading commit data...
vp9 Loading commit data...
vpx Loading commit data...
vpx_mem Loading commit data...
vpx_ports Loading commit data...
vpx_scale Loading commit data...
.gitattributes Loading commit data...
.gitignore Loading commit data...
.mailmap Loading commit data...
AUTHORS Loading commit data...
CHANGELOG Loading commit data...
LICENSE Loading commit data...
PATENTS Loading commit data...
README Loading commit data...
args.c Loading commit data...
args.h Loading commit data...
configure Loading commit data...
docs.mk Loading commit data...
example_xma.c Loading commit data...
examples.mk Loading commit data...
keywords.dox Loading commit data...
libs.doxy_template Loading commit data...
libs.mk Loading commit data...
mainpage.dox Loading commit data...
md5_utils.c Loading commit data...
md5_utils.h Loading commit data...
solution.mk Loading commit data...
tools_common.c Loading commit data...
tools_common.h Loading commit data...
usage.dox Loading commit data...
usage_cx.dox Loading commit data...
usage_dx.dox Loading commit data...
vp8_multi_resolution_encoder.c Loading commit data...
vp8_scalable_patterns.c Loading commit data...
vpxdec.c Loading commit data...
vpxenc.c Loading commit data...
y4minput.c Loading commit data...
y4minput.h Loading commit data...