]> granicus.if.org Git - libvpx/commit
projects / libvpx / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 72522db) | patch
Correctly initialize "ones" value in neon quantize
authorJohann <johannkoenig@google.com>
Thu, 20 Nov 2014 21:24:55 +0000 (13:24 -0800)
committerJohann <johannkoenig@google.com>
Fri, 21 Nov 2014 02:34:45 +0000 (18:34 -0800)
commit08ad7e4db57ab5be05a5304ad518c0886e18ad17
tree41e9909024032eb9a9f071de4e6aff5a1cb91230tree | snapshot
parent72522dbc86673569c03127f37f0f42d244ed7b73commit | diff
Correctly initialize "ones" value in neon quantize

By using 0xff for a short it was not setting the high bits. When
comparing the output with vtst to find non-zero elements it was skipping
vaules which had no low bits set such as -512 / 0xFE00.

Using -8191 as the first element of coeff will generate this condition.

BUG=883

Change-Id: Ia1e10fb809d1e7866f28c56769fe703e6231a657
test/quantize_test.cc diff | blob | history
vp8/encoder/arm/neon/fastquantizeb_neon.c diff | blob | history
Unnamed repository; edit this file 'description' to name the repository.