]> granicus.if.org Git - libvpx/commit
projects / libvpx / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 8894c76) | patch
Fix vp9_quantize_fp(_32x32)_neon for HBD
authorJerome Jiang <jianj@google.com>
Mon, 5 Aug 2019 22:38:39 +0000 (15:38 -0700)
committerJerome Jiang <jianj@google.com>
Mon, 5 Aug 2019 23:06:46 +0000 (16:06 -0700)
commite1f127c4b5a3b02b94006c81545233570dca8a7c
treef9bc6040242477330301ee22206e3909d064564ctree | snapshot
parent8894c766c6d9460d12e52d350e7084b143d6109bcommit | diff
Fix vp9_quantize_fp(_32x32)_neon for HBD

In high bitdepth build, Neon code would outrange because of use of
int16x8_t and vmulq_s16.
C code always truncate outrange values.

Change-Id: I33a968b8d812e3c8477f3a61d84482758a3f8b21
vp9/encoder/arm/neon/vp9_quantize_neon.c diff | blob | history
Unnamed repository; edit this file 'description' to name the repository.