]> granicus.if.org Git - libvpx/commit
quantize 32x32: saturate dqcoeff on x86
authorJohann <johann.koenig@duck.com>
Fri, 30 Nov 2018 23:42:57 +0000 (15:42 -0800)
committerJohann <johann.koenig@duck.com>
Sat, 1 Dec 2018 00:27:14 +0000 (16:27 -0800)
commit5fbc7a286b4d72883392fdbb10ec52bace662f66
treebdb6cf66e8eb26819fb1a8ddb602cd75471608cc
parent356174583506fb6654a3de7264348fbbfb7ca62c
quantize 32x32: saturate dqcoeff on x86

This slows down low bitdepth builds but is necessary to obtain correct
values.

BUG=webm:1448

Change-Id: I4ca9145f576089bb8496fcfeedeb556dc8fe6574
test/vp9_quantize_test.cc
vpx_dsp/quantize.c
vpx_dsp/x86/quantize_ssse3.h