PRecompute token tab

Change-Id: I85e172352dacf7a990d8d9296e071f9d603fcb7b
Write tokens with rans
2015-10-14 18:03:36 -07:00 · 2015-10-14 17:57:56 -07:00 · 2015-09-14 11:40:34 -07:00 · 2015-09-11 16:57:15 -07:00 · 2015-09-11 16:37:01 -07:00 · 2015-09-11 14:17:59 -07:00
880 changed files with 141166 additions and 60348 deletions
--- a/.gitignore
+++ b/.gitignore
@@ -30,14 +30,17 @@
 /examples/decode_with_partial_drops
 /examples/example_xma
 /examples/postproc
+/examples/resize_util
 /examples/set_maps
 /examples/simple_decoder
 /examples/simple_encoder
 /examples/twopass_encoder
 /examples/vp8_multi_resolution_encoder
 /examples/vp8cx_set_ref
+/examples/vp9_lossless_encoder
 /examples/vp9_spatial_scalable_encoder
 /examples/vpx_temporal_scalable_patterns
+/examples/vpx_temporal_svc_encoder
 /ivfdec
 /ivfdec.dox
 /ivfenc
@@ -45,12 +48,14 @@
 /libvpx.so*
 /libvpx.ver
 /samples.dox
+/test_intra_pred_speed
 /test_libvpx
 /vp8_api1_migration.dox
 /vp[89x]_rtcd.h
 /vpx.pc
 /vpx_config.c
 /vpx_config.h
+/vpx_dsp_rtcd.h
 /vpx_scale_rtcd.h
 /vpx_version.h
 /vpxdec
--- a/.mailmap
+++ b/.mailmap
@@ -1,18 +1,26 @@
 Adrian Grange <agrange@google.com>
+Alex Converse <aconverse@google.com> <alex.converse@gmail.com>
 Alexis Ballier <aballier@gentoo.org> <alexis.ballier@gmail.com>
+Alpha Lam <hclam@google.com> <hclam@chromium.org>
+Deb Mukherjee <debargha@google.com>
+Erik Niemeyer <erik.a.niemeyer@intel.com> <erik.a.niemeyer@gmail.com>
+Guillaume Martres <gmartres@google.com> <smarter3@gmail.com>
 Hangyu Kuang <hkuang@google.com>
 Jim Bankoski <jimbankoski@google.com>
-John Koleszar <jkoleszar@google.com>
 Johann Koenig <johannkoenig@google.com>
 Johann Koenig <johannkoenig@google.com> <johann.koenig@duck.com>
-Johann Koenig <johannkoenig@google.com> <johannkoenig@dhcp-172-19-7-52.mtv.corp.google.com>
+John Koleszar <jkoleszar@google.com>
+Joshua Litt <joshualitt@google.com> <joshualitt@chromium.org>
+Marco Paniconi <marpan@google.com>
+Marco Paniconi <marpan@google.com> <marpan@chromium.org>
 Pascal Massimino <pascal.massimino@gmail.com>
+Paul Wilkins <paulwilkins@google.com>
+Ralph Giles <giles@xiph.org> <giles@entropywave.com>
+Ralph Giles <giles@xiph.org> <giles@mozilla.com>
 Sami Pietilä <samipietila@google.com>
+Tamar Levy <tamar.levy@intel.com>
+Tamar Levy <tamar.levy@intel.com> <levytamar82@gmail.com>
 Tero Rintaluoma <teror@google.com> <tero.rintaluoma@on2.com>
 Timothy B. Terriberry <tterribe@xiph.org> Tim Terriberry <tterriberry@mozilla.com>
 Tom Finegan <tomfinegan@google.com>
-Ralph Giles <giles@xiph.org> <giles@entropywave.com>
-Ralph Giles <giles@xiph.org> <giles@mozilla.com>
-Alpha Lam <hclam@google.com> <hclam@chromium.org>
-Deb Mukherjee <debargha@google.com>
 Yaowu Xu <yaowu@google.com> <yaowu@xuyaowu.com>
--- a/29
+++ b/29
@@ -3,10 +3,11 @@

 Aaron Watry <awatry@gmail.com>
 Abo Talib Mahfoodh <ab.mahfoodh@gmail.com>
+Adam Xu <adam@xuyaowu.com>
 Adrian Grange <agrange@google.com>
 Ahmad Sharif <asharif@google.com>
 Alexander Voronov <avoronov@graphics.cs.msu.ru>
-Alex Converse <alex.converse@gmail.com>
+Alex Converse <aconverse@google.com>
 Alexis Ballier <aballier@gentoo.org>
 Alok Ahuja <waveletcoeff@gmail.com>
 Alpha Lam <hclam@google.com>
@@ -14,44 +15,58 @@ A.Mahfoodh <ab.mahfoodh@gmail.com>
 Ami Fischman <fischman@chromium.org>
 Andoni Morales Alastruey <ylatuya@gmail.com>
 Andres Mejia <mcitadel@gmail.com>
+Andrew Russell <anrussell@google.com>
 Aron Rosenberg <arosenberg@logitech.com>
 Attila Nagy <attilanagy@google.com>
 changjun.yang <changjun.yang@intel.com>
+Charles 'Buck' Krasic <ckrasic@google.com>
 chm <chm@rock-chips.com>
 Christian Duvivier <cduvivier@google.com>
 Daniel Kang <ddkang@google.com>
 Deb Mukherjee <debargha@google.com>
+Dim Temp <dimtemp0@gmail.com>
 Dmitry Kovalev <dkovalev@google.com>
 Dragan Mrdjan <dmrdjan@mips.com>
-Erik Niemeyer <erik.a.niemeyer@gmail.com>
+Ehsan Akhgari <ehsan.akhgari@gmail.com>
+Erik Niemeyer <erik.a.niemeyer@intel.com>
 Fabio Pedretti <fabio.ped@libero.it>
 Frank Galligan <fgalligan@google.com>
 Fredrik Söderquist <fs@opera.com>
 Fritz Koenig <frkoenig@google.com>
 Gaute Strokkenes <gaute.strokkenes@broadcom.com>
 Giuseppe Scrivano <gscrivano@gnu.org>
+Gordana Cmiljanovic <gordana.cmiljanovic@imgtec.com>
 Guillaume Martres <gmartres@google.com>
 Guillermo Ballester Valor <gbvalor@gmail.com>
 Hangyu Kuang <hkuang@google.com>
+Hanno Böck <hanno@hboeck.de>
 Henrik Lundin <hlundin@google.com>
 Hui Su <huisu@google.com>
 Ivan Maltz <ivanmaltz@google.com>
+Jacek Caban <cjacek@gmail.com>
+JackyChen <jackychen@google.com>
 James Berry <jamesberry@google.com>
+James Yu <james.yu@linaro.org>
 James Zern <jzern@google.com>
+Jan Gerber <j@mailb.org>
 Jan Kratochvil <jan.kratochvil@redhat.com>
 Janne Salonen <jsalonen@google.com>
 Jeff Faust <jfaust@google.com>
 Jeff Muizelaar <jmuizelaar@mozilla.com>
 Jeff Petkau <jpet@chromium.org>
+Jia Jia <jia.jia@linaro.org>
 Jim Bankoski <jimbankoski@google.com>
 Jingning Han <jingning@google.com>
+Joey Parrish <joeyparrish@google.com>
 Johann Koenig <johannkoenig@google.com>
 John Koleszar <jkoleszar@google.com>
+John Stark <jhnstrk@gmail.com>
 Joshua Bleecher Snyder <josh@treelinelabs.com>
 Joshua Litt <joshualitt@google.com>
 Justin Clift <justin@salasaga.org>
 Justin Lebar <justin.lebar@gmail.com>
 KO Myung-Hun <komh@chollian.net>
+Lawrence Velázquez <larryv@macports.org>
 Lou Quillio <louquillio@google.com>
 Luca Barbato <lu_zero@gentoo.org>
 Makoto Kato <makoto.kt@gmail.com>
@@ -65,6 +80,7 @@ Michael Kohler <michaelkohler@live.com>
 Mike Frysinger <vapier@chromium.org>
 Mike Hommey <mhommey@mozilla.com>
 Mikhal Shemer <mikhal@google.com>
+Minghai Shang <minghai@google.com>
 Morton Jonuschat <yabawock@gmail.com>
 Parag Salasakar <img.mips1@gmail.com>
 Pascal Massimino <pascal.massimino@gmail.com>
@@ -72,6 +88,8 @@ Patrik Westin <patrik.westin@gmail.com>
 Paul Wilkins <paulwilkins@google.com>
 Pavol Rusnak <stick@gk2.sk>
 Paweł Hajdan <phajdan@google.com>
+Pengchong Jin <pengchong@google.com>
+Peter de Rivaz <peter.derivaz@gmail.com>
 Philip Jägenstedt <philipj@opera.com>
 Priit Laes <plaes@plaes.org>
 Rafael Ávila de Espíndola <rafael.espindola@gmail.com>
@@ -79,22 +97,29 @@ Rafaël Carré <funman@videolan.org>
 Ralph Giles <giles@xiph.org>
 Rob Bradford <rob@linux.intel.com>
 Ronald S. Bultje <rbultje@google.com>
+Rui Ueyama <ruiu@google.com>
 Sami Pietilä <samipietila@google.com>
 Scott Graham <scottmg@chromium.org>
 Scott LaVarnway <slavarnway@google.com>
+Sean McGovern <gseanmcg@gmail.com>
+Sergey Ulanov <sergeyu@chromium.org>
 Shimon Doodkin <helpmepro1@gmail.com>
 Stefan Holmer <holmer@google.com>
 Suman Sunkara <sunkaras@google.com>
 Taekhyun Kim <takim@nvidia.com>
 Takanori MATSUURA <t.matsuu@gmail.com>
 Tamar Levy <tamar.levy@intel.com>
+Tao Bai <michaelbai@chromium.org>
 Tero Rintaluoma <teror@google.com>
 Thijs Vermeir <thijsvermeir@gmail.com>
+Tim Kopp <tkopp@google.com>
 Timothy B. Terriberry <tterribe@xiph.org>
 Tom Finegan <tomfinegan@google.com>
 Vignesh Venkatasubramanian <vigneshv@google.com>
 Yaowu Xu <yaowu@google.com>
+Yongzhe Wang <yongzhe@google.com>
 Yunqing Wang <yunqingwang@google.com>
+Zoe Liu <zoeliu@google.com>
 Google Inc.
 The Mozilla Foundation
 The Xiph.Org Foundation
--- a/28
+++ b/28
@@ -1,3 +1,31 @@
+xxxx-yy-zz v1.4.0 "Changes for next release"
+  vpxenc is changed to use VP9 by default.
+  Encoder controls added for 1 pass SVC.
+  Decoder control to toggle on/off loopfilter.
+
+2015-04-03 v1.4.0 "Indian Runner Duck"
+  This release includes significant improvements to the VP9 codec.
+
+  - Upgrading:
+    This release is ABI incompatible with 1.3.0. It drops the compatibility
+    layer, requiring VPX_IMG_FMT_* instead of IMG_FMT_*, and adds several codec
+    controls for VP9.
+
+  - Enhancements:
+    Faster VP9 encoding and decoding
+    Multithreaded VP9 decoding (tile and frame-based)
+    Multithreaded VP9 encoding - on by default
+    YUV 4:2:2 and 4:4:4 support in VP9
+    10 and 12bit support in VP9
+    64bit ARM support by replacing ARM assembly with intrinsics
+
+  - Bug Fixes:
+    Fixes a VP9 bitstream issue in Profile 1. This only affected non-YUV 4:2:0
+    files.
+
+  - Known Issues:
+    Frame Parallel decoding fails for segmented and non-420 files.
+
 2013-11-15 v1.3.0 "Forest"
  This release introduces the VP9 codec in a backward-compatible way.
  All existing users of VP8 can continue to use the library without
--- a/2
+++ b/2
@@ -17,7 +17,7 @@ or agree to the institution of patent litigation or any other patent
 enforcement activity against any entity (including a cross-claim or
 counterclaim in a lawsuit) alleging that any of these implementations of WebM
 or any code incorporated within any of these implementations of WebM
-constitutes direct or contributory patent infringement, or inducement of
+constitute direct or contributory patent infringement, or inducement of
 patent infringement, then any patent rights granted to you under this License
 for these implementations of WebM shall terminate as of the date such
 litigation is filed.
--- a/20
+++ b/20
@@ -1,4 +1,4 @@
-README - 30 May 2014
+README - 23 March 2015

 Welcome to the WebM VP8/VP9 Codec SDK!

@@ -59,15 +59,10 @@ COMPILING THE APPLICATIONS/LIBRARIES:
    armv7-none-rvct
    armv7-win32-vs11
    armv7-win32-vs12
+    armv7-win32-vs14
    armv7s-darwin-gcc
    mips32-linux-gcc
    mips64-linux-gcc
-    ppc32-darwin8-gcc
-    ppc32-darwin9-gcc
-    ppc32-linux-gcc
-    ppc64-darwin8-gcc
-    ppc64-darwin9-gcc
-    ppc64-linux-gcc
    sparc-solaris-gcc
    x86-android-gcc
    x86-darwin8-gcc
@@ -78,6 +73,7 @@ COMPILING THE APPLICATIONS/LIBRARIES:
    x86-darwin11-gcc
    x86-darwin12-gcc
    x86-darwin13-gcc
+    x86-darwin14-gcc
    x86-iphonesimulator-gcc
    x86-linux-gcc
    x86-linux-icc
@@ -90,11 +86,14 @@ COMPILING THE APPLICATIONS/LIBRARIES:
    x86-win32-vs10
    x86-win32-vs11
    x86-win32-vs12
+    x86-win32-vs14
+    x86_64-android-gcc
    x86_64-darwin9-gcc
    x86_64-darwin10-gcc
    x86_64-darwin11-gcc
    x86_64-darwin12-gcc
    x86_64-darwin13-gcc
+    x86_64-darwin14-gcc
    x86_64-iphonesimulator-gcc
    x86_64-linux-gcc
    x86_64-linux-icc
@@ -105,12 +104,7 @@ COMPILING THE APPLICATIONS/LIBRARIES:
    x86_64-win64-vs10
    x86_64-win64-vs11
    x86_64-win64-vs12
-    universal-darwin8-gcc
-    universal-darwin9-gcc
-    universal-darwin10-gcc
-    universal-darwin11-gcc
-    universal-darwin12-gcc
-    universal-darwin13-gcc
+    x86_64-win64-vs14
    generic-gnu

  The generic-gnu target, in conjunction with the CROSS environment variable,
--- a/args.c
+++ b/args.c
@@ -14,9 +14,7 @@
 #include <limits.h>
 #include "args.h"

-#ifdef _MSC_VER
-#define snprintf _snprintf
-#endif
+#include "vpx_ports/msvc.h"

 #if defined(__GNUC__) && __GNUC__
 extern void die(const char *fmt, ...) __attribute__((noreturn));
--- a/build/make/Android.mk
+++ b/build/make/Android.mk
@@ -67,6 +67,8 @@ else ifeq  ($(TARGET_ARCH_ABI),arm64-v8a)
  LOCAL_ARM_MODE := arm
 else ifeq ($(TARGET_ARCH_ABI),x86)
  include $(CONFIG_DIR)libs-x86-android-gcc.mk
+else ifeq ($(TARGET_ARCH_ABI),x86_64)
+  include $(CONFIG_DIR)libs-x86_64-android-gcc.mk
 else ifeq ($(TARGET_ARCH_ABI),mips)
  include $(CONFIG_DIR)libs-mips-android-gcc.mk
 else
@@ -158,24 +160,32 @@ LOCAL_CFLAGS += \

 LOCAL_MODULE := libvpx

-LOCAL_LDLIBS := -llog
-
 ifeq ($(CONFIG_RUNTIME_CPU_DETECT),yes)
  LOCAL_STATIC_LIBRARIES := cpufeatures
 endif

 # Add a dependency to force generation of the RTCD files.
+define rtcd_dep_template
+rtcd_dep_template_SRCS := $(addprefix $(LOCAL_PATH)/, $(LOCAL_SRC_FILES))
+rtcd_dep_template_SRCS := $$(rtcd_dep_template_SRCS:.neon=)
 ifeq ($(CONFIG_VP8), yes)
-$(foreach file, $(LOCAL_SRC_FILES), $(LOCAL_PATH)/$(file)): vp8_rtcd.h
+$$(rtcd_dep_template_SRCS): vp8_rtcd.h
 endif
 ifeq ($(CONFIG_VP9), yes)
-$(foreach file, $(LOCAL_SRC_FILES), $(LOCAL_PATH)/$(file)): vp9_rtcd.h
+$$(rtcd_dep_template_SRCS): vp9_rtcd.h
 endif
-$(foreach file, $(LOCAL_SRC_FILES), $(LOCAL_PATH)/$(file)): vpx_scale_rtcd.h
+ifeq ($(CONFIG_VP10), yes)
+$$(rtcd_dep_template_SRCS): vp10_rtcd.h
+endif
+$$(rtcd_dep_template_SRCS): vpx_scale_rtcd.h
+$$(rtcd_dep_template_SRCS): vpx_dsp_rtcd.h

-ifeq ($(TARGET_ARCH_ABI),x86)
-$(foreach file, $(LOCAL_SRC_FILES), $(LOCAL_PATH)/$(file)): vpx_config.asm
+ifneq ($(findstring $(TARGET_ARCH_ABI),x86 x86_64),)
+$$(rtcd_dep_template_SRCS): vpx_config.asm
 endif
+endef
+
+$(eval $(call rtcd_dep_template))

 .PHONY: clean
 clean:
@@ -184,7 +194,11 @@ clean:
 	@$(RM) -r $(ASM_CNV_PATH)
 	@$(RM) $(CLEAN-OBJS)

-include $(BUILD_SHARED_LIBRARY)
+ifeq ($(ENABLE_SHARED),1)
+  include $(BUILD_SHARED_LIBRARY)
+else
+  include $(BUILD_STATIC_LIBRARY)
+endif

 ifeq ($(CONFIG_RUNTIME_CPU_DETECT),yes)
 $(call import-module,cpufeatures)
--- a/build/make/Makefile
+++ b/build/make/Makefile
@@ -22,8 +22,10 @@ clean:: .DEFAULT
 exampletest: .DEFAULT
 install:: .DEFAULT
 test:: .DEFAULT
+test-no-data-check:: .DEFAULT
 testdata:: .DEFAULT
 utiltest: .DEFAULT
+exampletest-no-data-check utiltest-no-data-check: .DEFAULT


 # Note: md5sum is not installed on OS X, but openssl is. Openssl may not be
@@ -56,13 +58,10 @@ dist:
        fi
 endif

+# Since we invoke make recursively for multiple targets we need to include the
+# .mk file for the correct target, but only when $(target) is non-empty.
 ifneq ($(target),)
-# Normally, we want to build the filename from the target and the toolchain.
-# This disambiguates from the $(target).mk file that exists in the source tree.
-# However, the toolchain is part of the target in universal builds, so we
-# don't want to include TOOLCHAIN in that case. FAT_ARCHS is used to test
-# if we're in the universal case.
-include $(target)$(if $(FAT_ARCHS),,-$(TOOLCHAIN)).mk
+include $(target)-$(TOOLCHAIN).mk
 endif
 BUILD_ROOT?=.
 VPATH=$(SRC_PATH_BARE)
@@ -116,6 +115,9 @@ test::
 testdata::
 .PHONY: utiltest
 utiltest:
+.PHONY: test-no-data-check exampletest-no-data-check utiltest-no-data-check
+test-no-data-check::
+exampletest-no-data-check utiltest-no-data-check:

 # Add compiler flags for intrinsic files
 ifeq ($(TOOLCHAIN), x86-os2-gcc)
@@ -138,6 +140,8 @@ $(BUILD_PFX)%_avx.c.d: CFLAGS += -mavx $(STACKREALIGN)
 $(BUILD_PFX)%_avx.c.o: CFLAGS += -mavx $(STACKREALIGN)
 $(BUILD_PFX)%_avx2.c.d: CFLAGS += -mavx2 $(STACKREALIGN)
 $(BUILD_PFX)%_avx2.c.o: CFLAGS += -mavx2 $(STACKREALIGN)
+$(BUILD_PFX)%vp9_reconintra.c.d: CFLAGS += $(STACKREALIGN)
+$(BUILD_PFX)%vp9_reconintra.c.o: CFLAGS += $(STACKREALIGN)

 $(BUILD_PFX)%.c.d: %.c
 	$(if $(quiet),@echo "    [DEP] $@")
@@ -313,18 +317,15 @@ $(1):
        $$(filter %.o,$$^) $$(extralibs)
 endef

-
-
-define lipo_lib_template
-$(1): $(addsuffix /$(1),$(FAT_ARCHS))
-	$(if $(quiet),@echo "    [LIPO] $$@")
-	$(qexec)libtool -static -o $$@ $$?
-endef
-
-define lipo_bin_template
-$(1): $(addsuffix /$(1),$(FAT_ARCHS))
-	$(if $(quiet),@echo "    [LIPO] $$@")
-	$(qexec)lipo -output $$@ -create $$?
+define dll_template
+# Not using a pattern rule here because we don't want to generate empty
+# archives when they are listed as a dependency in files not responsible
+# for creating them.
+$(1):
+	$(if $(quiet),@echo "    [LD] $$@")
+	$(qexec)$$(LD) -Zdll $$(LDFLAGS) \
+        -o $$@ \
+        $$(filter %.o,$$^) $$(extralibs) $$(EXPORTS_FILE)
 endef


@@ -383,8 +384,9 @@ LIBS=$(call enabled,LIBS)
 .libs: $(LIBS)
 	@touch $@
 $(foreach lib,$(filter %_g.a,$(LIBS)),$(eval $(call archive_template,$(lib))))
-$(foreach lib,$(filter %so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH),$(LIBS)),$(eval $(call so_template,$(lib))))
-$(foreach lib,$(filter %$(VERSION_MAJOR).dylib,$(LIBS)),$(eval $(call dl_template,$(lib))))
+$(foreach lib,$(filter %so.$(SO_VERSION_MAJOR).$(SO_VERSION_MINOR).$(SO_VERSION_PATCH),$(LIBS)),$(eval $(call so_template,$(lib))))
+$(foreach lib,$(filter %$(SO_VERSION_MAJOR).dylib,$(LIBS)),$(eval $(call dl_template,$(lib))))
+$(foreach lib,$(filter %$(SO_VERSION_MAJOR).dll,$(LIBS)),$(eval $(call dll_template,$(lib))))

 INSTALL-LIBS=$(call cond_enabled,CONFIG_INSTALL_LIBS,INSTALL-LIBS)
 ifeq ($(MAKECMDGOALS),dist)
--- a/build/make/configure.sh
+++ b/build/make/configure.sh
@@ -390,7 +390,7 @@ write_common_config_banner() {
 write_common_config_targets() {
  for t in ${all_targets}; do
    if enabled ${t}; then
-      if enabled universal || enabled child; then
+      if enabled child; then
        fwrite config.mk "ALL_TARGETS += ${t}-${toolchain}"
      else
        fwrite config.mk "ALL_TARGETS += ${t}"
@@ -640,12 +640,6 @@ process_common_toolchain() {
      *i[3456]86*)
        tgt_isa=x86
        ;;
-      *powerpc64*)
-        tgt_isa=ppc64
-        ;;
-      *powerpc*)
-        tgt_isa=ppc32
-        ;;
      *sparc*)
        tgt_isa=sparc
        ;;
@@ -653,14 +647,6 @@ process_common_toolchain() {

    # detect tgt_os
    case "$gcctarget" in
-      *darwin8*)
-        tgt_isa=universal
-        tgt_os=darwin8
-        ;;
-      *darwin9*)
-        tgt_isa=universal
-        tgt_os=darwin9
-        ;;
      *darwin10*)
        tgt_isa=x86_64
        tgt_os=darwin10
@@ -742,6 +728,13 @@ process_common_toolchain() {
  # Handle darwin variants. Newer SDKs allow targeting older
  # platforms, so use the newest one available.
  case ${toolchain} in
+    arm*-darwin*)
+      ios_sdk_dir="$(show_darwin_sdk_path iphoneos)"
+      if [ -d "${ios_sdk_dir}" ]; then
+        add_cflags  "-isysroot ${ios_sdk_dir}"
+        add_ldflags "-isysroot ${ios_sdk_dir}"
+      fi
+      ;;
    *-darwin*)
      osx_sdk_dir="$(show_darwin_sdk_path macosx)"
      if [ -d "${osx_sdk_dir}" ]; then
@@ -795,7 +788,6 @@ process_common_toolchain() {
  case ${toolchain} in
    sparc-solaris-*)
      add_extralibs -lposix4
-      disable_feature fast_unaligned
      ;;
    *-solaris-*)
      add_extralibs -lposix4
@@ -818,12 +810,17 @@ process_common_toolchain() {
          if disabled neon && enabled neon_asm; then
            die "Disabling neon while keeping neon-asm is not supported"
          fi
-          soft_enable media
-          soft_enable fast_unaligned
+          case ${toolchain} in
+            *-darwin*)
+              # Neon is guaranteed on iOS 6+ devices, while old media extensions
+              # no longer assemble with iOS 9 SDK
+              ;;
+            *)
+              soft_enable media
+          esac
          ;;
        armv6)
          soft_enable media
-          soft_enable fast_unaligned
          ;;
      esac

@@ -1039,30 +1036,32 @@ EOF
      tune_cflags="-mtune="
      if enabled dspr2; then
        check_add_cflags -mips32r2 -mdspr2
-        disable_feature fast_unaligned
+      fi
+
+      if enabled runtime_cpu_detect; then
+        disable_feature runtime_cpu_detect
      fi

      if [ -n "${tune_cpu}" ]; then
        case ${tune_cpu} in
          p5600)
-            add_cflags -mips32r5 -funroll-loops -mload-store-pairs
-            add_cflags -msched-weight -mhard-float
-            add_asflags -mips32r5 -mhard-float
+            check_add_cflags -mips32r5 -funroll-loops -mload-store-pairs
+            check_add_cflags -msched-weight -mhard-float -mfp64
+            check_add_asflags -mips32r5 -mhard-float -mfp64
+            check_add_ldflags -mfp64
            ;;
          i6400)
-            add_cflags -mips64r6 -mabi=64 -funroll-loops -mload-store-pairs
-            add_cflags -msched-weight -mhard-float
-            add_asflags -mips64r6 -mabi=64 -mhard-float
-            add_ldflags -mips64r6 -mabi=64
+            check_add_cflags -mips64r6 -mabi=64 -funroll-loops -msched-weight 
+            check_add_cflags  -mload-store-pairs -mhard-float -mfp64
+            check_add_asflags -mips64r6 -mabi=64 -mhard-float -mfp64
+            check_add_ldflags -mips64r6 -mabi=64 -mfp64
            ;;
        esac

        if enabled msa; then
-          add_cflags -mmsa -mfp64 -flax-vector-conversions
-          add_asflags -mmsa -mfp64 -flax-vector-conversions
-          add_ldflags -mmsa -mfp64 -flax-vector-conversions
-
-          disable_feature fast_unaligned
+          add_cflags -mmsa
+          add_asflags -mmsa
+          add_ldflags -mmsa
        fi
      fi

@@ -1070,29 +1069,6 @@ EOF
      check_add_asflags -march=${tgt_isa}
      check_add_asflags -KPIC
      ;;
-    ppc*)
-      enable_feature ppc
-      bits=${tgt_isa##ppc}
-      link_with_cc=gcc
-      setup_gnu_toolchain
-      add_asflags -force_cpusubtype_ALL -I"\$(dir \$<)darwin"
-      soft_enable altivec
-      enabled altivec && add_cflags -maltivec
-
-      case "$tgt_os" in
-        linux*)
-          add_asflags -maltivec -mregnames -I"\$(dir \$<)linux"
-          ;;
-        darwin*)
-          darwin_arch="-arch ppc"
-          enabled ppc64 && darwin_arch="${darwin_arch}64"
-          add_cflags  ${darwin_arch} -m${bits} -fasm-blocks
-          add_asflags ${darwin_arch} -force_cpusubtype_ALL -I"\$(dir \$<)darwin"
-          add_ldflags ${darwin_arch} -m${bits}
-          enabled altivec && add_cflags -faltivec
-          ;;
-      esac
-      ;;
    x86*)
      case  ${tgt_os} in
        win*)
@@ -1105,7 +1081,9 @@ EOF
          CROSS=${CROSS:-g}
          ;;
        os2)
+          disable_feature pic
          AS=${AS:-nasm}
+          add_ldflags -Zhigh-mem
          ;;
      esac

@@ -1245,7 +1223,7 @@ EOF
          ;;
      esac
      ;;
-    universal*|*-gcc|generic-gnu)
+    *-gcc|generic-gnu)
      link_with_cc=gcc
      enable_feature gcc
      setup_gnu_toolchain
@@ -1329,11 +1307,15 @@ EOF
  # only for MIPS platforms
  case ${toolchain} in
    mips*)
-      if enabled dspr2; then
-        if enabled big_endian; then
+      if enabled big_endian; then
+        if enabled dspr2; then
          echo "dspr2 optimizations are available only for little endian platforms"
          disable_feature dspr2
        fi
+        if enabled msa; then
+          echo "msa optimizations are available only for little endian platforms"
+          disable_feature msa
+        fi
      fi
      ;;
  esac
@@ -1343,12 +1325,6 @@ EOF
    add_cflags -D_LARGEFILE_SOURCE
    add_cflags -D_FILE_OFFSET_BITS=64
  fi
-
-  # append any user defined extra cflags
-  if [ -n "${extra_cflags}" ] ; then
-    check_add_cflags ${extra_cflags} || \
-    die "Requested extra CFLAGS '${extra_cflags}' not supported by compiler"
-  fi
 }

 process_toolchain() {
--- a/build/make/gen_msvs_proj.sh
+++ b/build/make/gen_msvs_proj.sh
@@ -73,6 +73,10 @@ generate_filter() {
                open_tag File RelativePath="$f"

                if [ "$pat" == "asm" ] && $asm_use_custom_step; then
+                    # Avoid object file name collisions, i.e. vpx_config.c and
+                    # vpx_config.asm produce the same object file without
+                    # this additional suffix.
+                    objf=${objf%.obj}_asm.obj
                    for plat in "${platforms[@]}"; do
                        for cfg in Debug Release; do
                            open_tag FileConfiguration \
--- a/build/make/gen_msvs_sln.sh
+++ b/build/make/gen_msvs_sln.sh
@@ -19,13 +19,13 @@ show_help() {
    cat <<EOF
 Usage: ${self_basename} [options] file1 [file2 ...]

-This script generates a Visual Studio 2005 solution file from a list of project
+This script generates a Visual Studio solution file from a list of project
 files.

 Options:
    --help                      Print this message
    --out=outfile               Redirect output to a file
-    --ver=version               Version (7,8,9,10,11) of visual studio to generate for
+    --ver=version               Version (7,8,9,10,11,12,14) of visual studio to generate for
    --target=isa-os-cc          Target specifier
 EOF
    exit 1
@@ -255,7 +255,7 @@ for opt in "$@"; do
    ;;
    --ver=*) vs_ver="$optval"
             case $optval in
-             [789]|10|11|12)
+             [789]|10|11|12|14)
             ;;
             *) die Unrecognized Visual Studio Version in $opt
             ;;
@@ -300,12 +300,15 @@ case "${vs_ver:-8}" in
    12) sln_vers="12.00"
       sln_vers_str="Visual Studio 2013"
    ;;
+    14) sln_vers="14.00"
+       sln_vers_str="Visual Studio 2015"
+    ;;
 esac
 case "${vs_ver:-8}" in
    [789])
    sfx=vcproj
    ;;
-    10|11|12)
+    10|11|12|14)
    sfx=vcxproj
    ;;
 esac
--- a/build/make/gen_msvs_vcxproj.sh
+++ b/build/make/gen_msvs_vcxproj.sh
@@ -34,7 +34,7 @@ Options:
    --name=project_name         Name of the project (required)
    --proj-guid=GUID            GUID to use for the project
    --module-def=filename       File containing export definitions (for DLLs)
-    --ver=version               Version (10,11,12) of visual studio to generate for
+    --ver=version               Version (10,11,12,14) of visual studio to generate for
    --src-path-bare=dir         Path to root of source tree
    -Ipath/to/include           Additional include directories
    -DFLAG[=value]              Preprocessor macros to define
@@ -168,7 +168,7 @@ for opt in "$@"; do
        --ver=*)
            vs_ver="$optval"
            case "$optval" in
-                10|11|12)
+                10|11|12|14)
                ;;
                *) die Unrecognized Visual Studio Version in $opt
                ;;
@@ -218,7 +218,7 @@ guid=${guid:-`generate_uuid`}
 asm_use_custom_step=false
 uses_asm=${uses_asm:-false}
 case "${vs_ver:-11}" in
-    10|11|12)
+    10|11|12|14)
       asm_use_custom_step=$uses_asm
    ;;
 esac
@@ -263,8 +263,8 @@ case "$target" in
    ;;
    arm*)
        platforms[0]="ARM"
-        asm_Debug_cmdline="armasm -nologo &quot;%(FullPath)&quot;"
-        asm_Release_cmdline="armasm -nologo &quot;%(FullPath)&quot;"
+        asm_Debug_cmdline="armasm -nologo -oldit &quot;%(FullPath)&quot;"
+        asm_Release_cmdline="armasm -nologo -oldit &quot;%(FullPath)&quot;"
    ;;
    *) die "Unsupported target $target!"
    ;;
@@ -344,6 +344,9 @@ generate_vcxproj() {
                # has to enable AppContainerApplication as well.
                tag_content PlatformToolset v120
            fi
+            if [ "$vs_ver" = "14" ]; then
+                tag_content PlatformToolset v140
+            fi
            tag_content CharacterSet Unicode
            if [ "$config" = "Release" ]; then
                tag_content WholeProgramOptimization true
--- a/build/make/rtcd.pl
+++ b/build/make/rtcd.pl
@@ -319,13 +319,14 @@ EOF

  print <<EOF;
 #if HAVE_DSPR2
+void vpx_dsputil_static_init();
 #if CONFIG_VP8
 void dsputil_static_init();
-dsputil_static_init();
 #endif
-#if CONFIG_VP9
-void vp9_dsputil_static_init();
-vp9_dsputil_static_init();
+
+vpx_dsputil_static_init();
+#if CONFIG_VP8
+dsputil_static_init();
 #endif
 #endif
 }
--- a/codereview.settings
+++ b/codereview.settings
@@ -0,0 +1,4 @@
+# This file is used by gcl to get repository specific information.
+GERRIT_HOST: chromium-review.googlesource.com
+GERRIT_PORT: 29418
+CODE_REVIEW_SERVER: chromium-review.googlesource.com
--- a/134
+++ b/134
@@ -31,16 +31,14 @@ Advanced options:
  --size-limit=WxH                max size to allow in the decoder
  --as={yasm|nasm|auto}           use specified assembler [auto, yasm preferred]
  --sdk-path=PATH                 path to root of sdk (android builds only)
-  ${toggle_fast_unaligned}        don't use unaligned accesses, even when
-                                  supported by hardware [auto]
  ${toggle_codec_srcs}            in/exclude codec library source code
  ${toggle_debug_libs}            in/exclude debug version of libraries
  ${toggle_static_msvcrt}         use static MSVCRT (VS builds only)
  ${toggle_vp9_highbitdepth}      use VP9 high bit depth (10/12) profiles
  ${toggle_vp8}                   VP8 codec support
  ${toggle_vp9}                   VP9 codec support
+  ${toggle_vp10}                  VP10 codec support
  ${toggle_internal_stats}        output of encoder internal stats for debug, if supported (encoders)
-  ${toggle_mem_tracker}           track memory usage
  ${toggle_postproc}              postprocessing
  ${toggle_vp9_postproc}          vp9 specific postprocessing
  ${toggle_multithread}           multithreaded encoding and decoding
@@ -109,15 +107,10 @@ all_platforms="${all_platforms} armv7-linux-gcc"     #neon Cortex-A8
 all_platforms="${all_platforms} armv7-none-rvct"     #neon Cortex-A8
 all_platforms="${all_platforms} armv7-win32-vs11"
 all_platforms="${all_platforms} armv7-win32-vs12"
+all_platforms="${all_platforms} armv7-win32-vs14"
 all_platforms="${all_platforms} armv7s-darwin-gcc"
 all_platforms="${all_platforms} mips32-linux-gcc"
 all_platforms="${all_platforms} mips64-linux-gcc"
-all_platforms="${all_platforms} ppc32-darwin8-gcc"
-all_platforms="${all_platforms} ppc32-darwin9-gcc"
-all_platforms="${all_platforms} ppc32-linux-gcc"
-all_platforms="${all_platforms} ppc64-darwin8-gcc"
-all_platforms="${all_platforms} ppc64-darwin9-gcc"
-all_platforms="${all_platforms} ppc64-linux-gcc"
 all_platforms="${all_platforms} sparc-solaris-gcc"
 all_platforms="${all_platforms} x86-android-gcc"
 all_platforms="${all_platforms} x86-darwin8-gcc"
@@ -141,6 +134,8 @@ all_platforms="${all_platforms} x86-win32-vs9"
 all_platforms="${all_platforms} x86-win32-vs10"
 all_platforms="${all_platforms} x86-win32-vs11"
 all_platforms="${all_platforms} x86-win32-vs12"
+all_platforms="${all_platforms} x86-win32-vs14"
+all_platforms="${all_platforms} x86_64-android-gcc"
 all_platforms="${all_platforms} x86_64-darwin9-gcc"
 all_platforms="${all_platforms} x86_64-darwin10-gcc"
 all_platforms="${all_platforms} x86_64-darwin11-gcc"
@@ -157,13 +152,7 @@ all_platforms="${all_platforms} x86_64-win64-vs9"
 all_platforms="${all_platforms} x86_64-win64-vs10"
 all_platforms="${all_platforms} x86_64-win64-vs11"
 all_platforms="${all_platforms} x86_64-win64-vs12"
-all_platforms="${all_platforms} universal-darwin8-gcc"
-all_platforms="${all_platforms} universal-darwin9-gcc"
-all_platforms="${all_platforms} universal-darwin10-gcc"
-all_platforms="${all_platforms} universal-darwin11-gcc"
-all_platforms="${all_platforms} universal-darwin12-gcc"
-all_platforms="${all_platforms} universal-darwin13-gcc"
-all_platforms="${all_platforms} universal-darwin14-gcc"
+all_platforms="${all_platforms} x86_64-win64-vs14"
 all_platforms="${all_platforms} generic-gnu"

 # all_targets is a list of all targets that can be configured
@@ -200,6 +189,14 @@ if [ ${doxy_major:-0} -ge 1 ]; then
    [ $doxy_minor -eq 5 ] && [ $doxy_patch -ge 3 ] && enable_feature doxygen
 fi

+# disable codecs when their source directory does not exist
+[ -d "${source_path}/vp8" ] || disable_feature vp8
+[ -d "${source_path}/vp9" ] || disable_feature vp9
+[ -d "${source_path}/vp10" ] || disable_feature vp10
+
+# disable vp10 codec by default
+disable_feature vp10
+
 # install everything except the sources, by default. sources will have
 # to be enabled when doing dist builds, since that's no longer a common
 # case.
@@ -210,45 +207,30 @@ enable_feature install_libs
 enable_feature static
 enable_feature optimizations
 enable_feature dependency_tracking
-enable_feature fast_unaligned #allow unaligned accesses, if supported by hw
 enable_feature spatial_resampling
 enable_feature multithread
 enable_feature os_support
 enable_feature temporal_denoising

-[ -d "${source_path}/../include" ] && enable_feature alt_tree_layout
-for d in vp8 vp9; do
-    [ -d "${source_path}/${d}" ] && disable_feature alt_tree_layout;
-done
-
-if ! enabled alt_tree_layout; then
-# development environment
-[ -d "${source_path}/vp8" ] && CODECS="${CODECS} vp8_encoder vp8_decoder"
-[ -d "${source_path}/vp9" ] && CODECS="${CODECS} vp9_encoder vp9_decoder"
-else
-# customer environment
-[ -f "${source_path}/../include/vpx/vp8cx.h" ] && CODECS="${CODECS} vp8_encoder"
-[ -f "${source_path}/../include/vpx/vp8dx.h" ] && CODECS="${CODECS} vp8_decoder"
-[ -f "${source_path}/../include/vpx/vp9cx.h" ] && CODECS="${CODECS} vp9_encoder"
-[ -f "${source_path}/../include/vpx/vp9dx.h" ] && CODECS="${CODECS} vp9_decoder"
-[ -f "${source_path}/../include/vpx/vp8cx.h" ] || disable_feature vp8_encoder
-[ -f "${source_path}/../include/vpx/vp8dx.h" ] || disable_feature vp8_decoder
-[ -f "${source_path}/../include/vpx/vp9cx.h" ] || disable_feature vp9_encoder
-[ -f "${source_path}/../include/vpx/vp9dx.h" ] || disable_feature vp9_decoder
-
-[ -f "${source_path}/../lib/*/*mt.lib" ] && soft_enable static_msvcrt
-fi
-
-CODECS="$(echo ${CODECS} | tr ' ' '\n')"
-CODEC_FAMILIES="$(for c in ${CODECS}; do echo ${c%_*}; done | sort | uniq)"
+CODECS="
+    vp8_encoder
+    vp8_decoder
+    vp9_encoder
+    vp9_decoder
+    vp10_encoder
+    vp10_decoder
+"
+CODEC_FAMILIES="
+    vp8
+    vp9
+    vp10
+"

 ARCH_LIST="
    arm
    mips
    x86
    x86_64
-    ppc32
-    ppc64
 "
 ARCH_EXT_LIST="
    edsp
@@ -269,14 +251,11 @@ ARCH_EXT_LIST="
    sse4_1
    avx
    avx2
-
-    altivec
 "
 HAVE_LIST="
    ${ARCH_EXT_LIST}
    vpx_ports
    stdint_h
-    alt_tree_layout
    pthread_h
    sys_mman_h
    unistd_h
@@ -305,10 +284,6 @@ CONFIG_LIST="

    codec_srcs
    debug_libs
-    fast_unaligned
-    mem_manager
-    mem_tracker
-    mem_checks

    dequant_tokens
    dc_recon
@@ -370,7 +345,6 @@ CMDLINE_SELECT="
    libc
    as
    size_limit
-    fast_unaligned
    codec_srcs
    debug_libs

@@ -383,7 +357,6 @@ CMDLINE_SELECT="
    ${CODECS}
    ${CODEC_FAMILIES}
    static_msvcrt
-    mem_tracker
    spatial_resampling
    realtime_only
    onthefly_bitpacking
@@ -456,22 +429,8 @@ post_process_cmdline() {

 process_targets() {
    enabled child || write_common_config_banner
-    enabled universal || write_common_target_config_h  ${BUILD_PFX}vpx_config.h
-
-    # For fat binaries, call configure recursively to configure for each
-    # binary architecture to be included.
-    if enabled universal; then
-        # Call configure (ourselves) for each subarchitecture
-        for arch in $fat_bin_archs; do
-            BUILD_PFX=${arch}/ toolchain=${arch} $self --child $cmdline_args || exit $?
-        done
-    fi
-
-    # The write_common_config (config.mk) logic is deferred until after the
-    # recursive calls to configure complete, because we want our universal
-    # targets to be executed last.
+    write_common_target_config_h ${BUILD_PFX}vpx_config.h
    write_common_config_targets
-    enabled universal && echo "FAT_ARCHS=${fat_bin_archs}" >> config.mk

    # Calculate the default distribution name, based on the enabled features
    cf=""
@@ -547,11 +506,11 @@ process_detect() {
        # Can only build shared libs on a subset of platforms. Doing this check
        # here rather than at option parse time because the target auto-detect
        # magic happens after the command line has been parsed.
-        if ! enabled linux; then
+        if ! enabled linux && ! enabled os2; then
            if enabled gnu; then
                echo "--enable-shared is only supported on ELF; assuming this is OK"
            else
-                die "--enable-shared only supported on ELF for now"
+                die "--enable-shared only supported on ELF and OS/2 for now"
            fi
        fi
    fi
@@ -616,30 +575,6 @@ EOF
 process_toolchain() {
    process_common_toolchain

-    # Handle universal binaries for this architecture
-    case $toolchain in
-        universal-darwin*)
-            darwin_ver=${tgt_os##darwin}
-
-            # Snow Leopard (10.6/darwin10) dropped support for PPC
-            # Include PPC support for all prior versions
-            if [ $darwin_ver -lt 10 ]; then
-                fat_bin_archs="$fat_bin_archs ppc32-${tgt_os}-gcc"
-            fi
-
-            # Tiger (10.4/darwin8) brought support for x86
-            if [ $darwin_ver -ge 8 ]; then
-                fat_bin_archs="$fat_bin_archs x86-${tgt_os}-${tgt_cc}"
-            fi
-
-            # Leopard (10.5/darwin9) brought 64 bit support
-            if [ $darwin_ver -ge 9 ]; then
-                fat_bin_archs="$fat_bin_archs x86_64-${tgt_os}-${tgt_cc}"
-            fi
-            ;;
-    esac
-
-
    # Enable some useful compiler flags
    if enabled gcc; then
        enabled werror && check_add_cflags -Werror
@@ -715,7 +650,7 @@ process_toolchain() {
                 VCPROJ_SFX=vcproj
                 gen_vcproj_cmd=${source_path}/build/make/gen_msvs_proj.sh
                 ;;
-             10|11|12)
+             10|11|12|14)
                 VCPROJ_SFX=vcxproj
                 gen_vcproj_cmd=${source_path}/build/make/gen_msvs_vcxproj.sh
                 enabled werror && gen_vcproj_cmd="${gen_vcproj_cmd} --enable-werror"
@@ -727,7 +662,7 @@ process_toolchain() {
    esac

    # Other toolchain specific defaults
-    case $toolchain in x86*|ppc*|universal*) soft_enable postproc;; esac
+    case $toolchain in x86*) soft_enable postproc;; esac

    if enabled postproc_visualizer; then
        enabled postproc || die "postproc_visualizer requires postproc to be enabled"
@@ -781,6 +716,12 @@ EOF
    esac
    # libwebm needs to be linked with C++ standard library
    enabled webm_io && LD=${CXX}
+
+    # append any user defined extra cflags
+    if [ -n "${extra_cflags}" ] ; then
+        check_add_cflags ${extra_cflags} || \
+        die "Requested extra CFLAGS '${extra_cflags}' not supported by compiler"
+    fi
 }


@@ -791,6 +732,7 @@ CONFIGURE_ARGS="$@"
 process "$@"
 print_webm_license ${BUILD_PFX}vpx_config.c "/*" " */"
 cat <<EOF >> ${BUILD_PFX}vpx_config.c
+#include "vpx/vpx_codec.h"
 static const char* const cfg = "$CONFIGURE_ARGS";
 const char *vpx_codec_build_config(void) {return cfg;}
 EOF
--- a/examples.mk
+++ b/examples.mk
@@ -22,17 +22,18 @@ LIBYUV_SRCS +=  third_party/libyuv/include/libyuv/basic_types.h  \
                third_party/libyuv/source/planar_functions.cc \
                third_party/libyuv/source/row_any.cc \
                third_party/libyuv/source/row_common.cc \
+                third_party/libyuv/source/row_gcc.cc \
                third_party/libyuv/source/row_mips.cc \
                third_party/libyuv/source/row_neon.cc \
                third_party/libyuv/source/row_neon64.cc \
-                third_party/libyuv/source/row_posix.cc \
                third_party/libyuv/source/row_win.cc \
                third_party/libyuv/source/scale.cc \
+                third_party/libyuv/source/scale_any.cc \
                third_party/libyuv/source/scale_common.cc \
+                third_party/libyuv/source/scale_gcc.cc \
                third_party/libyuv/source/scale_mips.cc \
                third_party/libyuv/source/scale_neon.cc \
                third_party/libyuv/source/scale_neon64.cc \
-                third_party/libyuv/source/scale_posix.cc \
                third_party/libyuv/source/scale_win.cc \

 LIBWEBM_MUXER_SRCS += third_party/libwebm/mkvmuxer.cpp \
@@ -56,6 +57,7 @@ UTILS-$(CONFIG_DECODERS)    += vpxdec.c
 vpxdec.SRCS                 += md5_utils.c md5_utils.h
 vpxdec.SRCS                 += vpx_ports/mem_ops.h
 vpxdec.SRCS                 += vpx_ports/mem_ops_aligned.h
+vpxdec.SRCS                 += vpx_ports/msvc.h
 vpxdec.SRCS                 += vpx_ports/vpx_timer.h
 vpxdec.SRCS                 += vpx/vpx_integer.h
 vpxdec.SRCS                 += args.c args.h
@@ -80,6 +82,7 @@ vpxenc.SRCS                 += tools_common.c tools_common.h
 vpxenc.SRCS                 += warnings.c warnings.h
 vpxenc.SRCS                 += vpx_ports/mem_ops.h
 vpxenc.SRCS                 += vpx_ports/mem_ops_aligned.h
+vpxenc.SRCS                 += vpx_ports/msvc.h
 vpxenc.SRCS                 += vpx_ports/vpx_timer.h
 vpxenc.SRCS                 += vpxstats.c vpxstats.h
 ifeq ($(CONFIG_LIBYUV),yes)
@@ -98,6 +101,7 @@ ifeq ($(CONFIG_SPATIAL_SVC),yes)
  vp9_spatial_svc_encoder.SRCS        += tools_common.c tools_common.h
  vp9_spatial_svc_encoder.SRCS        += video_common.h
  vp9_spatial_svc_encoder.SRCS        += video_writer.h video_writer.c
+  vp9_spatial_svc_encoder.SRCS        += vpx_ports/msvc.h
  vp9_spatial_svc_encoder.SRCS        += vpxstats.c vpxstats.h
  vp9_spatial_svc_encoder.GUID        = 4A38598D-627D-4505-9C7B-D4020C84100D
  vp9_spatial_svc_encoder.DESCRIPTION = VP9 Spatial SVC Encoder
@@ -112,6 +116,7 @@ vpx_temporal_svc_encoder.SRCS        += ivfenc.c ivfenc.h
 vpx_temporal_svc_encoder.SRCS        += tools_common.c tools_common.h
 vpx_temporal_svc_encoder.SRCS        += video_common.h
 vpx_temporal_svc_encoder.SRCS        += video_writer.h video_writer.c
+vpx_temporal_svc_encoder.SRCS        += vpx_ports/msvc.h
 vpx_temporal_svc_encoder.GUID        = B18C08F2-A439-4502-A78E-849BE3D60947
 vpx_temporal_svc_encoder.DESCRIPTION = Temporal SVC Encoder
 EXAMPLES-$(CONFIG_DECODERS)        += simple_decoder.c
@@ -122,6 +127,7 @@ simple_decoder.SRCS                += video_common.h
 simple_decoder.SRCS                += video_reader.h video_reader.c
 simple_decoder.SRCS                += vpx_ports/mem_ops.h
 simple_decoder.SRCS                += vpx_ports/mem_ops_aligned.h
+simple_decoder.SRCS                += vpx_ports/msvc.h
 simple_decoder.DESCRIPTION          = Simplified decoder loop
 EXAMPLES-$(CONFIG_DECODERS)        += postproc.c
 postproc.SRCS                      += ivfdec.h ivfdec.c
@@ -130,6 +136,7 @@ postproc.SRCS                      += video_common.h
 postproc.SRCS                      += video_reader.h video_reader.c
 postproc.SRCS                      += vpx_ports/mem_ops.h
 postproc.SRCS                      += vpx_ports/mem_ops_aligned.h
+postproc.SRCS                      += vpx_ports/msvc.h
 postproc.GUID                       = 65E33355-F35E-4088-884D-3FD4905881D7
 postproc.DESCRIPTION                = Decoder postprocessor control
 EXAMPLES-$(CONFIG_DECODERS)        += decode_to_md5.c
@@ -140,6 +147,7 @@ decode_to_md5.SRCS                 += video_common.h
 decode_to_md5.SRCS                 += video_reader.h video_reader.c
 decode_to_md5.SRCS                 += vpx_ports/mem_ops.h
 decode_to_md5.SRCS                 += vpx_ports/mem_ops_aligned.h
+decode_to_md5.SRCS                 += vpx_ports/msvc.h
 decode_to_md5.GUID                  = 59120B9B-2735-4BFE-B022-146CA340FE42
 decode_to_md5.DESCRIPTION           = Frame by frame MD5 checksum
 EXAMPLES-$(CONFIG_ENCODERS)     += simple_encoder.c
@@ -147,6 +155,7 @@ simple_encoder.SRCS             += ivfenc.h ivfenc.c
 simple_encoder.SRCS             += tools_common.h tools_common.c
 simple_encoder.SRCS             += video_common.h
 simple_encoder.SRCS             += video_writer.h video_writer.c
+simple_encoder.SRCS             += vpx_ports/msvc.h
 simple_encoder.GUID              = 4607D299-8A71-4D2C-9B1D-071899B6FBFD
 simple_encoder.DESCRIPTION       = Simplified encoder loop
 EXAMPLES-$(CONFIG_VP9_ENCODER)  += vp9_lossless_encoder.c
@@ -154,6 +163,7 @@ vp9_lossless_encoder.SRCS       += ivfenc.h ivfenc.c
 vp9_lossless_encoder.SRCS       += tools_common.h tools_common.c
 vp9_lossless_encoder.SRCS       += video_common.h
 vp9_lossless_encoder.SRCS       += video_writer.h video_writer.c
+vp9_lossless_encoder.SRCS       += vpx_ports/msvc.h
 vp9_lossless_encoder.GUID        = B63C7C88-5348-46DC-A5A6-CC151EF93366
 vp9_lossless_encoder.DESCRIPTION = Simplified lossless VP9 encoder
 EXAMPLES-$(CONFIG_ENCODERS)     += twopass_encoder.c
@@ -161,6 +171,7 @@ twopass_encoder.SRCS            += ivfenc.h ivfenc.c
 twopass_encoder.SRCS            += tools_common.h tools_common.c
 twopass_encoder.SRCS            += video_common.h
 twopass_encoder.SRCS            += video_writer.h video_writer.c
+twopass_encoder.SRCS            += vpx_ports/msvc.h
 twopass_encoder.GUID             = 73494FA6-4AF9-4763-8FBB-265C92402FD8
 twopass_encoder.DESCRIPTION      = Two-pass encoder loop
 EXAMPLES-$(CONFIG_DECODERS)     += decode_with_drops.c
@@ -170,6 +181,7 @@ decode_with_drops.SRCS          += video_common.h
 decode_with_drops.SRCS          += video_reader.h video_reader.c
 decode_with_drops.SRCS          += vpx_ports/mem_ops.h
 decode_with_drops.SRCS          += vpx_ports/mem_ops_aligned.h
+decode_with_drops.SRCS          += vpx_ports/msvc.h
 decode_with_drops.GUID           = CE5C53C4-8DDA-438A-86ED-0DDD3CDB8D26
 decode_with_drops.DESCRIPTION    = Drops frames while decoding
 EXAMPLES-$(CONFIG_ENCODERS)        += set_maps.c
@@ -177,6 +189,7 @@ set_maps.SRCS                      += ivfenc.h ivfenc.c
 set_maps.SRCS                      += tools_common.h tools_common.c
 set_maps.SRCS                      += video_common.h
 set_maps.SRCS                      += video_writer.h video_writer.c
+set_maps.SRCS                      += vpx_ports/msvc.h
 set_maps.GUID                       = ECB2D24D-98B8-4015-A465-A4AF3DCC145F
 set_maps.DESCRIPTION                = Set active and ROI maps
 EXAMPLES-$(CONFIG_VP8_ENCODER)     += vp8cx_set_ref.c
@@ -184,6 +197,7 @@ vp8cx_set_ref.SRCS                 += ivfenc.h ivfenc.c
 vp8cx_set_ref.SRCS                 += tools_common.h tools_common.c
 vp8cx_set_ref.SRCS                 += video_common.h
 vp8cx_set_ref.SRCS                 += video_writer.h video_writer.c
+vp8cx_set_ref.SRCS                 += vpx_ports/msvc.h
 vp8cx_set_ref.GUID                  = C5E31F7F-96F6-48BD-BD3E-10EBF6E8057A
 vp8cx_set_ref.DESCRIPTION           = VP8 set encoder reference frame

@@ -194,6 +208,7 @@ EXAMPLES-$(CONFIG_VP8_ENCODER)          += vp8_multi_resolution_encoder.c
 vp8_multi_resolution_encoder.SRCS       += ivfenc.h ivfenc.c
 vp8_multi_resolution_encoder.SRCS       += tools_common.h tools_common.c
 vp8_multi_resolution_encoder.SRCS       += video_writer.h video_writer.c
+vp8_multi_resolution_encoder.SRCS       += vpx_ports/msvc.h
 vp8_multi_resolution_encoder.SRCS       += $(LIBYUV_SRCS)
 vp8_multi_resolution_encoder.GUID        = 04f8738e-63c8-423b-90fa-7c2703a374de
 vp8_multi_resolution_encoder.DESCRIPTION = VP8 Multiple-resolution Encoding
@@ -254,14 +269,6 @@ CODEC_EXTRA_LIBS=$(sort $(call enabled,CODEC_EXTRA_LIBS))
 $(foreach ex,$(ALL_EXAMPLES),$(eval $(notdir $(ex:.c=)).SRCS += $(ex) examples.mk))


-# If this is a universal (fat) binary, then all the subarchitectures have
-# already been built and our job is to stitch them together. The
-# BUILD_OBJS variable indicates whether we should be building
-# (compiling, linking) the library. The LIPO_OBJS variable indicates
-# that we're stitching.
-$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_OBJS,BUILD_OBJS):=yes)
-
-
 # Create build/install dependencies for all examples. The common case
 # is handled here. The MSVS case is handled below.
 NOT_MSVS = $(if $(CONFIG_MSVS),,yes)
@@ -269,24 +276,28 @@ DIST-BINS-$(NOT_MSVS)      += $(addprefix bin/,$(ALL_EXAMPLES:.c=$(EXE_SFX)))
 INSTALL-BINS-$(NOT_MSVS)   += $(addprefix bin/,$(UTILS:.c=$(EXE_SFX)))
 DIST-SRCS-yes              += $(ALL_SRCS)
 INSTALL-SRCS-yes           += $(UTIL_SRCS)
-OBJS-$(NOT_MSVS)           += $(if $(BUILD_OBJS),$(call objs,$(ALL_SRCS)))
+OBJS-$(NOT_MSVS)           += $(call objs,$(ALL_SRCS))
 BINS-$(NOT_MSVS)           += $(addprefix $(BUILD_PFX),$(ALL_EXAMPLES:.c=$(EXE_SFX)))


 # Instantiate linker template for all examples.
 CODEC_LIB=$(if $(CONFIG_DEBUG_LIBS),vpx_g,vpx)
-SHARED_LIB_SUF=$(if $(filter darwin%,$(TGT_OS)),.dylib,.so)
+ifneq ($(filter darwin%,$(TGT_OS)),)
+SHARED_LIB_SUF=.dylib
+else
+ifneq ($(filter os2%,$(TGT_OS)),)
+SHARED_LIB_SUF=_dll.a
+else
+SHARED_LIB_SUF=.so
+endif
+endif
 CODEC_LIB_SUF=$(if $(CONFIG_SHARED),$(SHARED_LIB_SUF),.a)
 $(foreach bin,$(BINS-yes),\
-    $(if $(BUILD_OBJS),$(eval $(bin):\
-        $(LIB_PATH)/lib$(CODEC_LIB)$(CODEC_LIB_SUF)))\
-    $(if $(BUILD_OBJS),$(eval $(call linker_template,$(bin),\
+    $(eval $(bin):$(LIB_PATH)/lib$(CODEC_LIB)$(CODEC_LIB_SUF))\
+    $(eval $(call linker_template,$(bin),\
        $(call objs,$($(notdir $(bin:$(EXE_SFX)=)).SRCS)) \
        -l$(CODEC_LIB) $(addprefix -l,$(CODEC_EXTRA_LIBS))\
-        )))\
-    $(if $(LIPO_OBJS),$(eval $(call lipo_bin_template,$(bin))))\
-    )
-
+        )))

 # The following pairs define a mapping of locations in the distribution
 # tree to locations in the source/build trees.
@@ -314,8 +325,8 @@ endif
 # the makefiles). We may want to revisit this.
 define vcproj_template
 $(1): $($(1:.$(VCPROJ_SFX)=).SRCS) vpx.$(VCPROJ_SFX)
-	@echo "    [vcproj] $$@"
-	$$(GEN_VCPROJ)\
+	$(if $(quiet),@echo "    [vcproj] $$@")
+	$(qexec)$$(GEN_VCPROJ)\
            --exe\
            --target=$$(TOOLCHAIN)\
            --name=$$(@:.$(VCPROJ_SFX)=)\
--- a/examples/decode_to_md5.c
+++ b/examples/decode_to_md5.c
@@ -71,7 +71,7 @@ static void print_md5(FILE *stream, unsigned char digest[16]) {

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <infile> <outfile>\n", exec_name);
  exit(EXIT_FAILURE);
 }
--- a/examples/decode_with_drops.c
+++ b/examples/decode_with_drops.c
@@ -65,7 +65,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <infile> <outfile> <N-M|N/M>\n", exec_name);
  exit(EXIT_FAILURE);
 }
--- a/examples/postproc.c
+++ b/examples/postproc.c
@@ -52,7 +52,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <infile> <outfile>\n", exec_name);
  exit(EXIT_FAILURE);
 }
--- a/examples/resize_util.c
+++ b/examples/resize_util.c
@@ -15,6 +15,7 @@
 #include <stdlib.h>
 #include <string.h>

+#include "../tools_common.h"
 #include "../vp9/encoder/vp9_resize.h"

 static const char *exec_name = NULL;
@@ -26,7 +27,7 @@ static void usage() {
  printf("<output_yuv> [<frames>]\n");
 }

-void usage_exit() {
+void usage_exit(void) {
  usage();
  exit(EXIT_FAILURE);
 }
--- a/examples/set_maps.c
+++ b/examples/set_maps.c
@@ -55,7 +55,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <codec> <width> <height> <infile> <outfile>\n",
          exec_name);
  exit(EXIT_FAILURE);
--- a/examples/simple_decoder.c
+++ b/examples/simple_decoder.c
@@ -88,7 +88,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <infile> <outfile>\n", exec_name);
  exit(EXIT_FAILURE);
 }
--- a/examples/simple_encoder.c
+++ b/examples/simple_encoder.c
@@ -106,7 +106,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr,
          "Usage: %s <codec> <width> <height> <infile> <outfile> "
              "<keyframe-interval> [<error-resilient>]\nSee comments in "
--- a/examples/twopass_encoder.c
+++ b/examples/twopass_encoder.c
@@ -58,7 +58,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <codec> <width> <height> <infile> <outfile>\n",
          exec_name);
  exit(EXIT_FAILURE);
--- a/examples/vp8_multi_resolution_encoder.c
+++ b/examples/vp8_multi_resolution_encoder.c
@@ -37,15 +37,14 @@
 #include <unistd.h>
 #endif
 #include "vpx_ports/vpx_timer.h"
-#define VPX_CODEC_DISABLE_COMPAT 1
 #include "vpx/vpx_encoder.h"
 #include "vpx/vp8cx.h"
 #include "vpx_ports/mem_ops.h"
-#include "./tools_common.h"
+#include "../tools_common.h"
 #define interface (vpx_codec_vp8_cx())
 #define fourcc    0x30385056

-void usage_exit() {
+void usage_exit(void) {
  exit(EXIT_FAILURE);
 }

--- a/examples/vp8cx_set_ref.c
+++ b/examples/vp8cx_set_ref.c
@@ -58,7 +58,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <width> <height> <infile> <outfile> <frame>\n",
          exec_name);
  exit(EXIT_FAILURE);
--- a/examples/vp9_lossless_encoder.c
+++ b/examples/vp9_lossless_encoder.c
@@ -20,7 +20,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "vp9_lossless_encoder: Example demonstrating VP9 lossless "
                  "encoding feature. Supports raw input only.\n");
  fprintf(stderr, "Usage: %s <width> <height> <infile> <outfile>\n", exec_name);
--- a/examples/vp9_spatial_svc_encoder.c
+++ b/examples/vp9_spatial_svc_encoder.c
@@ -14,24 +14,34 @@
 * that benefit from a scalable bitstream.
 */

+#include <math.h>
 #include <stdarg.h>
 #include <stdlib.h>
 #include <string.h>
 #include <time.h>

+
 #include "../args.h"
 #include "../tools_common.h"
 #include "../video_writer.h"

+#include "../vpx_ports/vpx_timer.h"
 #include "vpx/svc_context.h"
 #include "vpx/vp8cx.h"
 #include "vpx/vpx_encoder.h"
 #include "../vpxstats.h"
+#define OUTPUT_RC_STATS 1

 static const arg_def_t skip_frames_arg =
    ARG_DEF("s", "skip-frames", 1, "input frames to skip");
 static const arg_def_t frames_arg =
    ARG_DEF("f", "frames", 1, "number of frames to encode");
+static const arg_def_t threads_arg =
+    ARG_DEF("th", "threads", 1, "number of threads to use");
+#if OUTPUT_RC_STATS
+static const arg_def_t output_rc_stats_arg =
+    ARG_DEF("rcstat", "output_rc_stats", 1, "output rc stats");
+#endif
 static const arg_def_t width_arg = ARG_DEF("w", "width", 1, "source width");
 static const arg_def_t height_arg = ARG_DEF("h", "height", 1, "source height");
 static const arg_def_t timebase_arg =
@@ -42,6 +52,9 @@ static const arg_def_t spatial_layers_arg =
    ARG_DEF("sl", "spatial-layers", 1, "number of spatial SVC layers");
 static const arg_def_t temporal_layers_arg =
    ARG_DEF("tl", "temporal-layers", 1, "number of temporal SVC layers");
+static const arg_def_t temporal_layering_mode_arg =
+    ARG_DEF("tlm", "temporal-layering-mode", 1, "temporal layering scheme."
+        "VP9E_TEMPORAL_LAYERING_MODE");
 static const arg_def_t kf_dist_arg =
    ARG_DEF("k", "kf-dist", 1, "number of frames between keyframes");
 static const arg_def_t scale_factors_arg =
@@ -65,6 +78,8 @@ static const arg_def_t lag_in_frame_arg =
        "generating any outputs");
 static const arg_def_t rc_end_usage_arg =
    ARG_DEF(NULL, "rc-end-usage", 1, "0 - 3: VBR, CBR, CQ, Q");
+static const arg_def_t speed_arg =
+    ARG_DEF("sp", "speed", 1, "speed configuration");

 #if CONFIG_VP9_HIGHBITDEPTH
 static const struct arg_enum_list bitdepth_enum[] = {
@@ -85,10 +100,16 @@ static const arg_def_t *svc_args[] = {
  &timebase_arg,      &bitrate_arg,       &skip_frames_arg, &spatial_layers_arg,
  &kf_dist_arg,       &scale_factors_arg, &passes_arg,      &pass_arg,
  &fpf_name_arg,      &min_q_arg,         &max_q_arg,       &min_bitrate_arg,
-  &max_bitrate_arg,   &temporal_layers_arg,                 &lag_in_frame_arg,
+  &max_bitrate_arg,   &temporal_layers_arg, &temporal_layering_mode_arg,
+  &lag_in_frame_arg,  &threads_arg,
+#if OUTPUT_RC_STATS
+  &output_rc_stats_arg,
+#endif
+
 #if CONFIG_VP9_HIGHBITDEPTH
  &bitdepth_arg,
 #endif
+  &speed_arg,
  &rc_end_usage_arg,  NULL
 };

@@ -102,6 +123,10 @@ static const uint32_t default_bitrate = 1000;
 static const uint32_t default_spatial_layers = 5;
 static const uint32_t default_temporal_layers = 1;
 static const uint32_t default_kf_dist = 100;
+static const uint32_t default_temporal_layering_mode = 0;
+static const uint32_t default_output_rc_stats = 0;
+static const int32_t default_speed = -1;  // -1 means use library default.
+static const uint32_t default_threads = 0;  // zero means use library default.

 typedef struct {
  const char *input_filename;
@@ -116,7 +141,7 @@ typedef struct {

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  fprintf(stderr, "Usage: %s <options> input_filename output_filename\n",
          exec_name);
  fprintf(stderr, "Options:\n");
@@ -143,6 +168,12 @@ static void parse_command_line(int argc, const char **argv_,
  svc_ctx->log_level = SVC_LOG_DEBUG;
  svc_ctx->spatial_layers = default_spatial_layers;
  svc_ctx->temporal_layers = default_temporal_layers;
+  svc_ctx->temporal_layering_mode = default_temporal_layering_mode;
+#if OUTPUT_RC_STATS
+  svc_ctx->output_rc_stat = default_output_rc_stats;
+#endif
+  svc_ctx->speed = default_speed;
+  svc_ctx->threads = default_threads;

  // start with default encoder configuration
  res = vpx_codec_enc_config_default(vpx_codec_vp9_cx(), enc_cfg, 0);
@@ -184,6 +215,20 @@ static void parse_command_line(int argc, const char **argv_,
      svc_ctx->spatial_layers = arg_parse_uint(&arg);
    } else if (arg_match(&arg, &temporal_layers_arg, argi)) {
      svc_ctx->temporal_layers = arg_parse_uint(&arg);
+#if OUTPUT_RC_STATS
+    } else if (arg_match(&arg, &output_rc_stats_arg, argi)) {
+      svc_ctx->output_rc_stat = arg_parse_uint(&arg);
+#endif
+    } else if (arg_match(&arg, &speed_arg, argi)) {
+      svc_ctx->speed = arg_parse_uint(&arg);
+    } else if (arg_match(&arg, &threads_arg, argi)) {
+      svc_ctx->threads = arg_parse_uint(&arg);
+    } else if (arg_match(&arg, &temporal_layering_mode_arg, argi)) {
+      svc_ctx->temporal_layering_mode =
+          enc_cfg->temporal_layering_mode = arg_parse_int(&arg);
+      if (svc_ctx->temporal_layering_mode) {
+        enc_cfg->g_error_resilient = 1;
+      }
    } else if (arg_match(&arg, &kf_dist_arg, argi)) {
      enc_cfg->kf_min_dist = arg_parse_uint(&arg);
      enc_cfg->kf_max_dist = enc_cfg->kf_min_dist;
@@ -316,6 +361,185 @@ static void parse_command_line(int argc, const char **argv_,
      enc_cfg->rc_target_bitrate, enc_cfg->kf_max_dist);
 }

+#if OUTPUT_RC_STATS
+// For rate control encoding stats.
+struct RateControlStats {
+  // Number of input frames per layer.
+  int layer_input_frames[VPX_MAX_LAYERS];
+  // Total (cumulative) number of encoded frames per layer.
+  int layer_tot_enc_frames[VPX_MAX_LAYERS];
+  // Number of encoded non-key frames per layer.
+  int layer_enc_frames[VPX_MAX_LAYERS];
+  // Framerate per layer (cumulative).
+  double layer_framerate[VPX_MAX_LAYERS];
+  // Target average frame size per layer (per-frame-bandwidth per layer).
+  double layer_pfb[VPX_MAX_LAYERS];
+  // Actual average frame size per layer.
+  double layer_avg_frame_size[VPX_MAX_LAYERS];
+  // Average rate mismatch per layer (|target - actual| / target).
+  double layer_avg_rate_mismatch[VPX_MAX_LAYERS];
+  // Actual encoding bitrate per layer (cumulative).
+  double layer_encoding_bitrate[VPX_MAX_LAYERS];
+  // Average of the short-time encoder actual bitrate.
+  // TODO(marpan): Should we add these short-time stats for each layer?
+  double avg_st_encoding_bitrate;
+  // Variance of the short-time encoder actual bitrate.
+  double variance_st_encoding_bitrate;
+  // Window (number of frames) for computing short-time encoding bitrate.
+  int window_size;
+  // Number of window measurements.
+  int window_count;
+};
+
+// Note: these rate control stats assume only 1 key frame in the
+// sequence (i.e., first frame only).
+static void set_rate_control_stats(struct RateControlStats *rc,
+                                     vpx_codec_enc_cfg_t *cfg) {
+  unsigned int sl, tl;
+  // Set the layer (cumulative) framerate and the target layer (non-cumulative)
+  // per-frame-bandwidth, for the rate control encoding stats below.
+  const double framerate = cfg->g_timebase.den / cfg->g_timebase.num;
+
+  for (sl = 0; sl < cfg->ss_number_layers; ++sl) {
+    for (tl = 0; tl < cfg->ts_number_layers; ++tl) {
+      const int layer = sl * cfg->ts_number_layers + tl;
+      const int tlayer0 = sl * cfg->ts_number_layers;
+      rc->layer_framerate[layer] =
+          framerate / cfg->ts_rate_decimator[tl];
+      if (tl > 0) {
+        rc->layer_pfb[layer] = 1000.0 *
+            (cfg->layer_target_bitrate[layer] -
+                cfg->layer_target_bitrate[layer - 1]) /
+            (rc->layer_framerate[layer] -
+                rc->layer_framerate[layer - 1]);
+      } else {
+        rc->layer_pfb[tlayer0] = 1000.0 *
+            cfg->layer_target_bitrate[tlayer0] /
+            rc->layer_framerate[tlayer0];
+      }
+      rc->layer_input_frames[layer] = 0;
+      rc->layer_enc_frames[layer] = 0;
+      rc->layer_tot_enc_frames[layer] = 0;
+      rc->layer_encoding_bitrate[layer] = 0.0;
+      rc->layer_avg_frame_size[layer] = 0.0;
+      rc->layer_avg_rate_mismatch[layer] = 0.0;
+    }
+  }
+  rc->window_count = 0;
+  rc->window_size = 15;
+  rc->avg_st_encoding_bitrate = 0.0;
+  rc->variance_st_encoding_bitrate = 0.0;
+}
+
+static void printout_rate_control_summary(struct RateControlStats *rc,
+                                          vpx_codec_enc_cfg_t *cfg,
+                                          int frame_cnt) {
+  unsigned int sl, tl;
+  int tot_num_frames = 0;
+  double perc_fluctuation = 0.0;
+  printf("Total number of processed frames: %d\n\n", frame_cnt - 1);
+  printf("Rate control layer stats for sl%d tl%d layer(s):\n\n",
+      cfg->ss_number_layers, cfg->ts_number_layers);
+  for (sl = 0; sl < cfg->ss_number_layers; ++sl) {
+    for (tl = 0; tl < cfg->ts_number_layers; ++tl) {
+      const int layer = sl * cfg->ts_number_layers + tl;
+      const int num_dropped = (tl > 0) ?
+          (rc->layer_input_frames[layer] - rc->layer_enc_frames[layer]) :
+          (rc->layer_input_frames[layer] - rc->layer_enc_frames[layer] - 1);
+      if (!sl)
+        tot_num_frames += rc->layer_input_frames[layer];
+      rc->layer_encoding_bitrate[layer] = 0.001 * rc->layer_framerate[layer] *
+          rc->layer_encoding_bitrate[layer] / tot_num_frames;
+      rc->layer_avg_frame_size[layer] = rc->layer_avg_frame_size[layer] /
+          rc->layer_enc_frames[layer];
+      rc->layer_avg_rate_mismatch[layer] =
+          100.0 * rc->layer_avg_rate_mismatch[layer] /
+          rc->layer_enc_frames[layer];
+      printf("For layer#: sl%d tl%d \n", sl, tl);
+      printf("Bitrate (target vs actual): %d %f.0 kbps\n",
+             cfg->layer_target_bitrate[layer],
+             rc->layer_encoding_bitrate[layer]);
+      printf("Average frame size (target vs actual): %f %f bits\n",
+             rc->layer_pfb[layer], rc->layer_avg_frame_size[layer]);
+      printf("Average rate_mismatch: %f\n",
+             rc->layer_avg_rate_mismatch[layer]);
+      printf("Number of input frames, encoded (non-key) frames, "
+          "and percent dropped frames: %d %d %f.0 \n",
+          rc->layer_input_frames[layer], rc->layer_enc_frames[layer],
+          100.0 * num_dropped / rc->layer_input_frames[layer]);
+      printf("\n");
+    }
+  }
+  rc->avg_st_encoding_bitrate = rc->avg_st_encoding_bitrate / rc->window_count;
+  rc->variance_st_encoding_bitrate =
+      rc->variance_st_encoding_bitrate / rc->window_count -
+      (rc->avg_st_encoding_bitrate * rc->avg_st_encoding_bitrate);
+  perc_fluctuation = 100.0 * sqrt(rc->variance_st_encoding_bitrate) /
+      rc->avg_st_encoding_bitrate;
+  printf("Short-time stats, for window of %d frames: \n", rc->window_size);
+  printf("Average, rms-variance, and percent-fluct: %f %f %f \n",
+         rc->avg_st_encoding_bitrate,
+         sqrt(rc->variance_st_encoding_bitrate),
+         perc_fluctuation);
+  if (frame_cnt != tot_num_frames)
+    die("Error: Number of input frames not equal to output encoded frames != "
+        "%d tot_num_frames = %d\n", frame_cnt, tot_num_frames);
+}
+
+vpx_codec_err_t parse_superframe_index(const uint8_t *data,
+                                       size_t data_sz,
+                                       uint32_t sizes[8], int *count) {
+  // A chunk ending with a byte matching 0xc0 is an invalid chunk unless
+  // it is a super frame index. If the last byte of real video compression
+  // data is 0xc0 the encoder must add a 0 byte. If we have the marker but
+  // not the associated matching marker byte at the front of the index we have
+  // an invalid bitstream and need to return an error.
+
+  uint8_t marker;
+
+  marker = *(data + data_sz - 1);
+  *count = 0;
+
+
+  if ((marker & 0xe0) == 0xc0) {
+    const uint32_t frames = (marker & 0x7) + 1;
+    const uint32_t mag = ((marker >> 3) & 0x3) + 1;
+    const size_t index_sz = 2 + mag * frames;
+
+    // This chunk is marked as having a superframe index but doesn't have
+    // enough data for it, thus it's an invalid superframe index.
+    if (data_sz < index_sz)
+      return VPX_CODEC_CORRUPT_FRAME;
+
+    {
+      const uint8_t marker2 = *(data + data_sz - index_sz);
+
+      // This chunk is marked as having a superframe index but doesn't have
+      // the matching marker byte at the front of the index therefore it's an
+      // invalid chunk.
+      if (marker != marker2)
+        return VPX_CODEC_CORRUPT_FRAME;
+    }
+
+    {
+      // Found a valid superframe index.
+      uint32_t i, j;
+      const uint8_t *x = &data[data_sz - index_sz + 1];
+
+      for (i = 0; i < frames; ++i) {
+        uint32_t this_sz = 0;
+
+        for (j = 0; j < mag; ++j)
+          this_sz |= (*x++) << (j * 8);
+        sizes[i] = this_sz;
+      }
+      *count = frames;
+    }
+  }
+  return VPX_CODEC_OK;
+}
+#endif
+
 int main(int argc, const char **argv) {
  AppInput app_input = {0};
  VpxVideoWriter *writer = NULL;
@@ -332,7 +556,17 @@ int main(int argc, const char **argv) {
  FILE *infile = NULL;
  int end_of_stream = 0;
  int frames_received = 0;
-
+#if OUTPUT_RC_STATS
+  VpxVideoWriter *outfile[VPX_TS_MAX_LAYERS] = {NULL};
+  struct RateControlStats rc;
+  vpx_svc_layer_id_t layer_id;
+  int sl, tl;
+  double sum_bitrate = 0.0;
+  double sum_bitrate2 = 0.0;
+  double framerate  = 30.0;
+#endif
+  struct vpx_usec_timer timer;
+  int64_t cx_time = 0;
  memset(&svc_ctx, 0, sizeof(svc_ctx));
  svc_ctx.log_print = 1;
  exec_name = argv[0];
@@ -359,6 +593,13 @@ int main(int argc, const char **argv) {
      VPX_CODEC_OK)
    die("Failed to initialize encoder\n");

+#if OUTPUT_RC_STATS
+  if (svc_ctx.output_rc_stat) {
+    set_rate_control_stats(&rc, &enc_cfg);
+    framerate = enc_cfg.g_timebase.den / enc_cfg.g_timebase.num;
+  }
+#endif
+
  info.codec_fourcc = VP9_FOURCC;
  info.time_base.numerator = enc_cfg.g_timebase.num;
  info.time_base.denominator = enc_cfg.g_timebase.den;
@@ -370,11 +611,34 @@ int main(int argc, const char **argv) {
    if (!writer)
      die("Failed to open %s for writing\n", app_input.output_filename);
  }
+#if OUTPUT_RC_STATS
+  // For now, just write temporal layer streams.
+  // TODO(wonkap): do spatial by re-writing superframe.
+  if (svc_ctx.output_rc_stat) {
+    for (tl = 0; tl < enc_cfg.ts_number_layers; ++tl) {
+      char file_name[PATH_MAX];
+
+      snprintf(file_name, sizeof(file_name), "%s_t%d.ivf",
+               app_input.output_filename, tl);
+      outfile[tl] = vpx_video_writer_open(file_name, kContainerIVF, &info);
+      if (!outfile[tl])
+        die("Failed to open %s for writing", file_name);
+    }
+  }
+#endif

  // skip initial frames
  for (i = 0; i < app_input.frames_to_skip; ++i)
    vpx_img_read(&raw, infile);

+  if (svc_ctx.speed != -1)
+    vpx_codec_control(&codec, VP8E_SET_CPUUSED, svc_ctx.speed);
+  if (svc_ctx.threads)
+    vpx_codec_control(&codec, VP9E_SET_TILE_COLUMNS, (svc_ctx.threads >> 1));
+  if (svc_ctx.speed >= 5)
+    vpx_codec_control(&codec, VP9E_SET_AQ_MODE, 3);
+
+
  // Encode frames
  while (!end_of_stream) {
    vpx_codec_iter_t iter = NULL;
@@ -385,8 +649,13 @@ int main(int argc, const char **argv) {
      end_of_stream = 1;
    }

+    vpx_usec_timer_start(&timer);
    res = vpx_svc_encode(&svc_ctx, &codec, (end_of_stream ? NULL : &raw),
-                         pts, frame_duration, VPX_DL_GOOD_QUALITY);
+                         pts, frame_duration, svc_ctx.speed >= 5 ?
+                         VPX_DL_REALTIME : VPX_DL_GOOD_QUALITY);
+    vpx_usec_timer_mark(&timer);
+    cx_time += vpx_usec_timer_elapsed(&timer);
+
    printf("%s", vpx_svc_get_message(&svc_ctx));
    if (res != VPX_CODEC_OK) {
      die_codec(&codec, "Failed to encode frame");
@@ -395,11 +664,90 @@ int main(int argc, const char **argv) {
    while ((cx_pkt = vpx_codec_get_cx_data(&codec, &iter)) != NULL) {
      switch (cx_pkt->kind) {
        case VPX_CODEC_CX_FRAME_PKT: {
-          if (cx_pkt->data.frame.sz > 0)
+          if (cx_pkt->data.frame.sz > 0) {
+#if OUTPUT_RC_STATS
+            uint32_t sizes[8];
+            int count = 0;
+#endif
            vpx_video_writer_write_frame(writer,
                                         cx_pkt->data.frame.buf,
                                         cx_pkt->data.frame.sz,
                                         cx_pkt->data.frame.pts);
+#if OUTPUT_RC_STATS
+            // TODO(marpan/wonkap): Put this (to line728) in separate function.
+            if (svc_ctx.output_rc_stat) {
+              vpx_codec_control(&codec, VP9E_GET_SVC_LAYER_ID, &layer_id);
+              parse_superframe_index(cx_pkt->data.frame.buf,
+                                     cx_pkt->data.frame.sz, sizes, &count);
+              for (sl = 0; sl < enc_cfg.ss_number_layers; ++sl) {
+                ++rc.layer_input_frames[sl * enc_cfg.ts_number_layers +
+                                        layer_id.temporal_layer_id];
+              }
+              for (tl = layer_id.temporal_layer_id;
+                  tl < enc_cfg.ts_number_layers; ++tl) {
+                vpx_video_writer_write_frame(outfile[tl],
+                                             cx_pkt->data.frame.buf,
+                                             cx_pkt->data.frame.sz,
+                                             cx_pkt->data.frame.pts);
+              }
+
+              for (sl = 0; sl < enc_cfg.ss_number_layers; ++sl) {
+                for (tl = layer_id.temporal_layer_id;
+                    tl < enc_cfg.ts_number_layers; ++tl) {
+                  const int layer = sl * enc_cfg.ts_number_layers + tl;
+                  ++rc.layer_tot_enc_frames[layer];
+                  rc.layer_encoding_bitrate[layer] += 8.0 * sizes[sl];
+                  // Keep count of rate control stats per layer, for non-key
+                  // frames.
+                  if (tl == layer_id.temporal_layer_id &&
+                      !(cx_pkt->data.frame.flags & VPX_FRAME_IS_KEY)) {
+                    rc.layer_avg_frame_size[layer] += 8.0 * sizes[sl];
+                    rc.layer_avg_rate_mismatch[layer] +=
+                        fabs(8.0 * sizes[sl] - rc.layer_pfb[layer]) /
+                        rc.layer_pfb[layer];
+                    ++rc.layer_enc_frames[layer];
+                  }
+                }
+              }
+
+              // Update for short-time encoding bitrate states, for moving
+              // window of size rc->window, shifted by rc->window / 2.
+              // Ignore first window segment, due to key frame.
+              if (frame_cnt > rc.window_size) {
+                tl = layer_id.temporal_layer_id;
+                for (sl = 0; sl < enc_cfg.ss_number_layers; ++sl) {
+                  sum_bitrate += 0.001 * 8.0 * sizes[sl] * framerate;
+                }
+                if (frame_cnt % rc.window_size == 0) {
+                  rc.window_count += 1;
+                  rc.avg_st_encoding_bitrate += sum_bitrate / rc.window_size;
+                  rc.variance_st_encoding_bitrate +=
+                      (sum_bitrate / rc.window_size) *
+                      (sum_bitrate / rc.window_size);
+                  sum_bitrate = 0.0;
+                }
+              }
+
+              // Second shifted window.
+              if (frame_cnt > rc.window_size + rc.window_size / 2) {
+               tl = layer_id.temporal_layer_id;
+               for (sl = 0; sl < enc_cfg.ss_number_layers; ++sl) {
+                 sum_bitrate2 += 0.001 * 8.0 * sizes[sl] * framerate;
+               }
+
+               if (frame_cnt > 2 * rc.window_size &&
+                  frame_cnt % rc.window_size == 0) {
+                 rc.window_count += 1;
+                 rc.avg_st_encoding_bitrate += sum_bitrate2 / rc.window_size;
+                 rc.variance_st_encoding_bitrate +=
+                    (sum_bitrate2 / rc.window_size) *
+                    (sum_bitrate2 / rc.window_size);
+                 sum_bitrate2 = 0.0;
+               }
+              }
+            }
+#endif
+          }

          printf("SVC frame: %d, kf: %d, size: %d, pts: %d\n", frames_received,
                 !!(cx_pkt->data.frame.flags & VPX_FRAME_IS_KEY),
@@ -424,25 +772,34 @@ int main(int argc, const char **argv) {
      pts += frame_duration;
    }
  }
-
  printf("Processed %d frames\n", frame_cnt);
-
  fclose(infile);
+#if OUTPUT_RC_STATS
+  if (svc_ctx.output_rc_stat) {
+    printout_rate_control_summary(&rc, &enc_cfg, frame_cnt);
+    printf("\n");
+  }
+#endif
  if (vpx_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
-
  if (app_input.passes == 2)
    stats_close(&app_input.rc_stats, 1);
-
  if (writer) {
    vpx_video_writer_close(writer);
  }
-
+#if OUTPUT_RC_STATS
+  if (svc_ctx.output_rc_stat) {
+    for (tl = 0; tl < enc_cfg.ts_number_layers; ++tl) {
+      vpx_video_writer_close(outfile[tl]);
+    }
+  }
+#endif
+  printf("Frame cnt and encoding time/FPS stats for encoding: %d %f %f \n",
+         frame_cnt,
+         1000 * (float)cx_time / (double)(frame_cnt * 1000000),
+         1000000 * (double)frame_cnt / (double)cx_time);
  vpx_img_free(&raw);
-
  // display average size, psnr
  printf("%s", vpx_svc_dump_statistics(&svc_ctx));
-
  vpx_svc_release(&svc_ctx);
-
  return EXIT_SUCCESS;
 }
--- a/examples/vpx_temporal_svc_encoder.c
+++ b/examples/vpx_temporal_svc_encoder.c
@@ -28,7 +28,7 @@

 static const char *exec_name;

-void usage_exit() {
+void usage_exit(void) {
  exit(EXIT_FAILURE);
 }

@@ -70,6 +70,7 @@ struct RateControlMetrics {
  int window_size;
  // Number of window measurements.
  int window_count;
+  int layer_target_bitrate[VPX_MAX_LAYERS];
 };

 // Note: these rate control metrics assume only 1 key frame in the
@@ -85,13 +86,13 @@ static void set_rate_control_metrics(struct RateControlMetrics *rc,
  // per-frame-bandwidth, for the rate control encoding stats below.
  const double framerate = cfg->g_timebase.den / cfg->g_timebase.num;
  rc->layer_framerate[0] = framerate / cfg->ts_rate_decimator[0];
-  rc->layer_pfb[0] = 1000.0 * cfg->ts_target_bitrate[0] /
+  rc->layer_pfb[0] = 1000.0 * rc->layer_target_bitrate[0] /
      rc->layer_framerate[0];
  for (i = 0; i < cfg->ts_number_layers; ++i) {
    if (i > 0) {
      rc->layer_framerate[i] = framerate / cfg->ts_rate_decimator[i];
      rc->layer_pfb[i] = 1000.0 *
-          (cfg->ts_target_bitrate[i] - cfg->ts_target_bitrate[i - 1]) /
+          (rc->layer_target_bitrate[i] - rc->layer_target_bitrate[i - 1]) /
          (rc->layer_framerate[i] - rc->layer_framerate[i - 1]);
    }
    rc->layer_input_frames[i] = 0;
@@ -128,7 +129,7 @@ static void printout_rate_control_summary(struct RateControlMetrics *rc,
    rc->layer_avg_rate_mismatch[i] = 100.0 * rc->layer_avg_rate_mismatch[i] /
        rc->layer_enc_frames[i];
    printf("For layer#: %d \n", i);
-    printf("Bitrate (target vs actual): %d %f \n", cfg->ts_target_bitrate[i],
+    printf("Bitrate (target vs actual): %d %f \n", rc->layer_target_bitrate[i],
           rc->layer_encoding_bitrate[i]);
    printf("Average frame size (target vs actual): %f %f \n", rc->layer_pfb[i],
           rc->layer_avg_frame_size[i]);
@@ -597,13 +598,16 @@ int main(int argc, char **argv) {
  for (i = min_args_base;
       (int)i < min_args_base + mode_to_num_layers[layering_mode];
       ++i) {
-    cfg.ts_target_bitrate[i - 11] = strtol(argv[i], NULL, 0);
+    rc.layer_target_bitrate[i - 11] = strtol(argv[i], NULL, 0);
+    if (strncmp(encoder->name, "vp8", 3) == 0)
+      cfg.ts_target_bitrate[i - 11] = rc.layer_target_bitrate[i - 11];
+    else if (strncmp(encoder->name, "vp9", 3) == 0)
+      cfg.layer_target_bitrate[i - 11] = rc.layer_target_bitrate[i - 11];
  }

  // Real time parameters.
  cfg.rc_dropframe_thresh = strtol(argv[9], NULL, 0);
  cfg.rc_end_usage = VPX_CBR;
-  cfg.rc_resize_allowed = 0;
  cfg.rc_min_quantizer = 2;
  cfg.rc_max_quantizer = 56;
  if (strncmp(encoder->name, "vp9", 3) == 0)
@@ -614,6 +618,9 @@ int main(int argc, char **argv) {
  cfg.rc_buf_optimal_sz = 600;
  cfg.rc_buf_sz = 1000;

+  // Disable dynamic resizing by default.
+  cfg.rc_resize_allowed = 0;
+
  // Use 1 thread as default.
  cfg.g_threads = 1;

@@ -625,6 +632,8 @@ int main(int argc, char **argv) {
  // Disable automatic keyframe placement.
  cfg.kf_min_dist = cfg.kf_max_dist = 3000;

+  cfg.temporal_layering_mode = VP9E_TEMPORAL_LAYERING_MODE_BYPASS;
+
  set_temporal_layer_pattern(layering_mode,
                             &cfg,
                             layer_flags,
@@ -633,8 +642,8 @@ int main(int argc, char **argv) {
  set_rate_control_metrics(&rc, &cfg);

  // Target bandwidth for the whole stream.
-  // Set to ts_target_bitrate for highest layer (total bitrate).
-  cfg.rc_target_bitrate = cfg.ts_target_bitrate[cfg.ts_number_layers - 1];
+  // Set to layer_target_bitrate for highest layer (total bitrate).
+  cfg.rc_target_bitrate = rc.layer_target_bitrate[cfg.ts_number_layers - 1];

  // Open input file.
  if (!(infile = fopen(argv[1], "rb"))) {
@@ -674,18 +683,26 @@ int main(int argc, char **argv) {

  if (strncmp(encoder->name, "vp8", 3) == 0) {
    vpx_codec_control(&codec, VP8E_SET_CPUUSED, -speed);
-    vpx_codec_control(&codec, VP8E_SET_NOISE_SENSITIVITY, kDenoiserOnYOnly);
-    vpx_codec_control(&codec, VP8E_SET_STATIC_THRESHOLD, 1);
+    vpx_codec_control(&codec, VP8E_SET_NOISE_SENSITIVITY, kDenoiserOff);
+    vpx_codec_control(&codec, VP8E_SET_STATIC_THRESHOLD, 0);
  } else if (strncmp(encoder->name, "vp9", 3) == 0) {
-      vpx_codec_control(&codec, VP8E_SET_CPUUSED, speed);
-      vpx_codec_control(&codec, VP9E_SET_AQ_MODE, 3);
-      vpx_codec_control(&codec, VP9E_SET_FRAME_PERIODIC_BOOST, 0);
-      vpx_codec_control(&codec, VP9E_SET_NOISE_SENSITIVITY, 0);
-      vpx_codec_control(&codec, VP8E_SET_STATIC_THRESHOLD, 1);
-      vpx_codec_control(&codec, VP9E_SET_TILE_COLUMNS, (cfg.g_threads >> 1));
-      if (vpx_codec_control(&codec, VP9E_SET_SVC, layering_mode > 0 ? 1: 0)) {
-        die_codec(&codec, "Failed to set SVC");
+    vpx_svc_extra_cfg_t svc_params;
+    vpx_codec_control(&codec, VP8E_SET_CPUUSED, speed);
+    vpx_codec_control(&codec, VP9E_SET_AQ_MODE, 3);
+    vpx_codec_control(&codec, VP9E_SET_FRAME_PERIODIC_BOOST, 0);
+    vpx_codec_control(&codec, VP9E_SET_NOISE_SENSITIVITY, 0);
+    vpx_codec_control(&codec, VP8E_SET_STATIC_THRESHOLD, 0);
+    vpx_codec_control(&codec, VP9E_SET_TUNE_CONTENT, 0);
+    vpx_codec_control(&codec, VP9E_SET_TILE_COLUMNS, (cfg.g_threads >> 1));
+    if (vpx_codec_control(&codec, VP9E_SET_SVC, layering_mode > 0 ? 1: 0))
+      die_codec(&codec, "Failed to set SVC");
+    for (i = 0; i < cfg.ts_number_layers; ++i) {
+      svc_params.max_quantizers[i] = cfg.rc_max_quantizer;
+      svc_params.min_quantizers[i] = cfg.rc_min_quantizer;
    }
+    svc_params.scaling_factor_num[0] = cfg.g_h;
+    svc_params.scaling_factor_den[0] = cfg.g_h;
+    vpx_codec_control(&codec, VP9E_SET_SVC_PARAMETERS, &svc_params);
  }
  if (strncmp(encoder->name, "vp8", 3) == 0) {
    vpx_codec_control(&codec, VP8E_SET_SCREEN_CONTENT_MODE, 0);
--- a/libs.mk
+++ b/libs.mk
@@ -25,7 +25,7 @@ $$(BUILD_PFX)$(1).h: $$(SRC_PATH_BARE)/$(2)
 	@echo "    [CREATE] $$@"
 	$$(qexec)$$(SRC_PATH_BARE)/build/make/rtcd.pl --arch=$$(TGT_ISA) \
          --sym=$(1) \
-          --config=$$(CONFIG_DIR)$$(target)$$(if $$(FAT_ARCHS),,-$$(TOOLCHAIN)).mk \
+          --config=$$(CONFIG_DIR)$$(target)-$$(TOOLCHAIN).mk \
          $$(RTCD_OPTIONS) $$^ > $$@
 CLEAN-OBJS += $$(BUILD_PFX)$(1).h
 RTCD += $$(BUILD_PFX)$(1).h
@@ -34,13 +34,6 @@ endef
 CODEC_SRCS-yes += CHANGELOG
 CODEC_SRCS-yes += libs.mk

-# If this is a universal (fat) binary, then all the subarchitectures have
-# already been built and our job is to stitch them together. The
-# BUILD_LIBVPX variable indicates whether we should be building
-# (compiling, linking) the library. The LIPO_LIBVPX variable indicates
-# that we're stitching.
-$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_LIBVPX,BUILD_LIBVPX):=yes)
-
 include $(SRC_PATH_BARE)/vpx/vpx_codec.mk
 CODEC_SRCS-yes += $(addprefix vpx/,$(call enabled,API_SRCS))
 CODEC_DOC_SRCS += $(addprefix vpx/,$(call enabled,API_DOC_SRCS))
@@ -54,7 +47,13 @@ CODEC_SRCS-yes += $(addprefix vpx_scale/,$(call enabled,SCALE_SRCS))
 include $(SRC_PATH_BARE)/vpx_ports/vpx_ports.mk
 CODEC_SRCS-yes += $(addprefix vpx_ports/,$(call enabled,PORTS_SRCS))

-ifneq ($(CONFIG_VP8_ENCODER)$(CONFIG_VP8_DECODER),)
+include $(SRC_PATH_BARE)/vpx_dsp/vpx_dsp.mk
+CODEC_SRCS-yes += $(addprefix vpx_dsp/,$(call enabled,DSP_SRCS))
+
+include $(SRC_PATH_BARE)/vpx_util/vpx_util.mk
+CODEC_SRCS-yes += $(addprefix vpx_util/,$(call enabled,UTIL_SRCS))
+
+ifeq ($(CONFIG_VP8),yes)
  VP8_PREFIX=vp8/
  include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk
 endif
@@ -77,7 +76,7 @@ ifeq ($(CONFIG_VP8_DECODER),yes)
  CODEC_DOC_SECTIONS += vp8 vp8_decoder
 endif

-ifneq ($(CONFIG_VP9_ENCODER)$(CONFIG_VP9_DECODER),)
+ifeq ($(CONFIG_VP9),yes)
  VP9_PREFIX=vp9/
  include $(SRC_PATH_BARE)/$(VP9_PREFIX)vp9_common.mk
 endif
@@ -110,6 +109,40 @@ endif
 VP9_PREFIX=vp9/
 $(BUILD_PFX)$(VP9_PREFIX)%.c.o: CFLAGS += -Wextra

+#  VP10 make file
+ifeq ($(CONFIG_VP10),yes)
+  VP10_PREFIX=vp10/
+  include $(SRC_PATH_BARE)/$(VP10_PREFIX)vp10_common.mk
+endif
+
+ifeq ($(CONFIG_VP10_ENCODER),yes)
+  VP10_PREFIX=vp10/
+  include $(SRC_PATH_BARE)/$(VP10_PREFIX)vp10cx.mk
+  CODEC_SRCS-yes += $(addprefix $(VP10_PREFIX),$(call enabled,VP10_CX_SRCS))
+  CODEC_EXPORTS-yes += $(addprefix $(VP10_PREFIX),$(VP10_CX_EXPORTS))
+  CODEC_SRCS-yes += $(VP10_PREFIX)vp10cx.mk vpx/vp8.h vpx/vp8cx.h
+  INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8cx.h
+  INSTALL-LIBS-$(CONFIG_SPATIAL_SVC) += include/vpx/svc_context.h
+  INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP10_PREFIX)/%
+  CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8cx.h
+  CODEC_DOC_SECTIONS += vp9 vp9_encoder
+endif
+
+ifeq ($(CONFIG_VP10_DECODER),yes)
+  VP10_PREFIX=vp10/
+  include $(SRC_PATH_BARE)/$(VP10_PREFIX)vp10dx.mk
+  CODEC_SRCS-yes += $(addprefix $(VP10_PREFIX),$(call enabled,VP10_DX_SRCS))
+  CODEC_EXPORTS-yes += $(addprefix $(VP10_PREFIX),$(VP10_DX_EXPORTS))
+  CODEC_SRCS-yes += $(VP10_PREFIX)vp10dx.mk vpx/vp8.h vpx/vp8dx.h
+  INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8dx.h
+  INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP10_PREFIX)/%
+  CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8dx.h
+  CODEC_DOC_SECTIONS += vp9 vp9_decoder
+endif
+
+VP10_PREFIX=vp10/
+$(BUILD_PFX)$(VP10_PREFIX)%.c.o: CFLAGS += -Wextra
+
 ifeq ($(CONFIG_ENCODERS),yes)
  CODEC_DOC_SECTIONS += encoder
 endif
@@ -137,18 +170,18 @@ INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/%  $(p)/Release/%)
 INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/%  $(p)/Debug/%)
 endif

-CODEC_SRCS-$(BUILD_LIBVPX) += build/make/version.sh
-CODEC_SRCS-$(BUILD_LIBVPX) += build/make/rtcd.pl
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/emmintrin_compat.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/mem_ops.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/mem_ops_aligned.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/vpx_once.h
-CODEC_SRCS-$(BUILD_LIBVPX) += $(BUILD_PFX)vpx_config.c
+CODEC_SRCS-yes += build/make/version.sh
+CODEC_SRCS-yes += build/make/rtcd.pl
+CODEC_SRCS-yes += vpx_ports/emmintrin_compat.h
+CODEC_SRCS-yes += vpx_ports/mem_ops.h
+CODEC_SRCS-yes += vpx_ports/mem_ops_aligned.h
+CODEC_SRCS-yes += vpx_ports/vpx_once.h
+CODEC_SRCS-yes += $(BUILD_PFX)vpx_config.c
 INSTALL-SRCS-no += $(BUILD_PFX)vpx_config.c
 ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
 INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += third_party/x86inc/x86inc.asm
 endif
-CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx/exports_com
+CODEC_EXPORTS-yes += vpx/exports_com
 CODEC_EXPORTS-$(CONFIG_ENCODERS) += vpx/exports_enc
 CODEC_EXPORTS-$(CONFIG_DECODERS) += vpx/exports_dec

@@ -215,7 +248,7 @@ vpx.$(VCPROJ_SFX): $(CODEC_SRCS) vpx.def
            $(filter-out $(addprefix %, $(ASM_INCLUDES)), $^) \
            --src-path-bare="$(SRC_PATH_BARE)" \

-PROJECTS-$(BUILD_LIBVPX) += vpx.$(VCPROJ_SFX)
+PROJECTS-yes += vpx.$(VCPROJ_SFX)

 vpx.$(VCPROJ_SFX): vpx_config.asm
 vpx.$(VCPROJ_SFX): $(RTCD)
@@ -223,32 +256,42 @@ vpx.$(VCPROJ_SFX): $(RTCD)
 endif
 else
 LIBVPX_OBJS=$(call objs,$(CODEC_SRCS))
-OBJS-$(BUILD_LIBVPX) += $(LIBVPX_OBJS)
-LIBS-$(if $(BUILD_LIBVPX),$(CONFIG_STATIC)) += $(BUILD_PFX)libvpx.a $(BUILD_PFX)libvpx_g.a
+OBJS-yes += $(LIBVPX_OBJS)
+LIBS-$(if yes,$(CONFIG_STATIC)) += $(BUILD_PFX)libvpx.a $(BUILD_PFX)libvpx_g.a
 $(BUILD_PFX)libvpx_g.a: $(LIBVPX_OBJS)

-
-BUILD_LIBVPX_SO         := $(if $(BUILD_LIBVPX),$(CONFIG_SHARED))
-
+SO_VERSION_MAJOR := 2
+SO_VERSION_MINOR := 0
+SO_VERSION_PATCH := 0
 ifeq ($(filter darwin%,$(TGT_OS)),$(TGT_OS))
-LIBVPX_SO               := libvpx.$(VERSION_MAJOR).dylib
+LIBVPX_SO               := libvpx.$(SO_VERSION_MAJOR).dylib
+SHARED_LIB_SUF          := .dylib
 EXPORT_FILE             := libvpx.syms
 LIBVPX_SO_SYMLINKS      := $(addprefix $(LIBSUBDIR)/, \
                             libvpx.dylib  )
 else
-LIBVPX_SO               := libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH)
+ifeq ($(filter os2%,$(TGT_OS)),$(TGT_OS))
+LIBVPX_SO               := libvpx$(SO_VERSION_MAJOR).dll
+SHARED_LIB_SUF          := _dll.a
+EXPORT_FILE             := libvpx.def
+LIBVPX_SO_SYMLINKS      :=
+LIBVPX_SO_IMPLIB        := libvpx_dll.a
+else
+LIBVPX_SO               := libvpx.so.$(SO_VERSION_MAJOR).$(SO_VERSION_MINOR).$(SO_VERSION_PATCH)
+SHARED_LIB_SUF          := .so
 EXPORT_FILE             := libvpx.ver
-SYM_LINK                := libvpx.so
 LIBVPX_SO_SYMLINKS      := $(addprefix $(LIBSUBDIR)/, \
-                             libvpx.so libvpx.so.$(VERSION_MAJOR) \
-                             libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR))
+                             libvpx.so libvpx.so.$(SO_VERSION_MAJOR) \
+                             libvpx.so.$(SO_VERSION_MAJOR).$(SO_VERSION_MINOR))
+endif
 endif

-LIBS-$(BUILD_LIBVPX_SO) += $(BUILD_PFX)$(LIBVPX_SO)\
-                           $(notdir $(LIBVPX_SO_SYMLINKS))
+LIBS-$(CONFIG_SHARED) += $(BUILD_PFX)$(LIBVPX_SO)\
+                           $(notdir $(LIBVPX_SO_SYMLINKS)) \
+                           $(if $(LIBVPX_SO_IMPLIB), $(BUILD_PFX)$(LIBVPX_SO_IMPLIB))
 $(BUILD_PFX)$(LIBVPX_SO): $(LIBVPX_OBJS) $(EXPORT_FILE)
 $(BUILD_PFX)$(LIBVPX_SO): extralibs += -lm
-$(BUILD_PFX)$(LIBVPX_SO): SONAME = libvpx.so.$(VERSION_MAJOR)
+$(BUILD_PFX)$(LIBVPX_SO): SONAME = libvpx.so.$(SO_VERSION_MAJOR)
 $(BUILD_PFX)$(LIBVPX_SO): EXPORTS_FILE = $(EXPORT_FILE)

 libvpx.ver: $(call enabled,CODEC_EXPORTS)
@@ -263,6 +306,19 @@ libvpx.syms: $(call enabled,CODEC_EXPORTS)
 	$(qexec)awk '{print "_"$$2}' $^ >$@
 CLEAN-OBJS += libvpx.syms

+libvpx.def: $(call enabled,CODEC_EXPORTS)
+	@echo "    [CREATE] $@"
+	$(qexec)echo LIBRARY $(LIBVPX_SO:.dll=) INITINSTANCE TERMINSTANCE > $@
+	$(qexec)echo "DATA MULTIPLE NONSHARED" >> $@
+	$(qexec)echo "EXPORTS" >> $@
+	$(qexec)awk '!/vpx_svc_*/ {print "_"$$2}' $^ >>$@
+CLEAN-OBJS += libvpx.def
+
+libvpx_dll.a: $(LIBVPX_SO)
+	@echo "    [IMPLIB] $@"
+	$(qexec)emximp -o $@ $<
+CLEAN-OBJS += libvpx_dll.a
+
 define libvpx_symlink_template
 $(1): $(2)
 	@echo "    [LN]     $(2) $$@"
@@ -278,11 +334,12 @@ $(eval $(call libvpx_symlink_template,\
    $(LIBVPX_SO)))


-INSTALL-LIBS-$(BUILD_LIBVPX_SO) += $(LIBVPX_SO_SYMLINKS)
-INSTALL-LIBS-$(BUILD_LIBVPX_SO) += $(LIBSUBDIR)/$(LIBVPX_SO)
+INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBVPX_SO_SYMLINKS)
+INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBSUBDIR)/$(LIBVPX_SO)
+INSTALL-LIBS-$(CONFIG_SHARED) += $(if $(LIBVPX_SO_IMPLIB),$(LIBSUBDIR)/$(LIBVPX_SO_IMPLIB))


-LIBS-$(BUILD_LIBVPX) += vpx.pc
+LIBS-yes += vpx.pc
 vpx.pc: config.mk libs.mk
 	@echo "    [CREATE] $@"
 	$(qexec)echo '# pkg-config file from libvpx $(VERSION_STRING)' > $@
@@ -308,9 +365,6 @@ INSTALL_MAPS += $(LIBSUBDIR)/pkgconfig/%.pc %.pc
 CLEAN-OBJS += vpx.pc
 endif

-LIBS-$(LIPO_LIBVPX) += libvpx.a
-$(eval $(if $(LIPO_LIBVPX),$(call lipo_lib_template,libvpx.a)))
-
 #
 # Rule to make assembler configuration file from C configuration file
 #
@@ -349,11 +403,15 @@ LIBVPX_TEST_DATA_PATH ?= .

 include $(SRC_PATH_BARE)/test/test.mk
 LIBVPX_TEST_SRCS=$(addprefix test/,$(call enabled,LIBVPX_TEST_SRCS))
-LIBVPX_TEST_BINS=./test_libvpx$(EXE_SFX)
+LIBVPX_TEST_BIN=./test_libvpx$(EXE_SFX)
 LIBVPX_TEST_DATA=$(addprefix $(LIBVPX_TEST_DATA_PATH)/,\
                     $(call enabled,LIBVPX_TEST_DATA))
 libvpx_test_data_url=http://downloads.webmproject.org/test_data/libvpx/$(1)

+TEST_INTRA_PRED_SPEED_BIN=./test_intra_pred_speed$(EXE_SFX)
+TEST_INTRA_PRED_SPEED_SRCS=$(addprefix test/,$(call enabled,TEST_INTRA_PRED_SPEED_SRCS))
+TEST_INTRA_PRED_SPEED_OBJS := $(sort $(call objs,$(TEST_INTRA_PRED_SPEED_SRCS)))
+
 libvpx_test_srcs.txt:
 	@echo "    [CREATE] $@"
 	@echo $(LIBVPX_TEST_SRCS) | xargs -n1 echo | LC_ALL=C sort -u > $@
@@ -417,7 +475,25 @@ test_libvpx.$(VCPROJ_SFX): $(LIBVPX_TEST_SRCS) vpx.$(VCPROJ_SFX) gtest.$(VCPROJ_

 PROJECTS-$(CONFIG_MSVS) += test_libvpx.$(VCPROJ_SFX)

-LIBVPX_TEST_BINS := $(addprefix $(TGT_OS:win64=x64)/Release/,$(notdir $(LIBVPX_TEST_BINS)))
+LIBVPX_TEST_BIN := $(addprefix $(TGT_OS:win64=x64)/Release/,$(notdir $(LIBVPX_TEST_BIN)))
+
+ifneq ($(strip $(TEST_INTRA_PRED_SPEED_OBJS)),)
+PROJECTS-$(CONFIG_MSVS) += test_intra_pred_speed.$(VCPROJ_SFX)
+test_intra_pred_speed.$(VCPROJ_SFX): $(TEST_INTRA_PRED_SPEED_SRCS) vpx.$(VCPROJ_SFX) gtest.$(VCPROJ_SFX)
+	@echo "    [CREATE] $@"
+	$(qexec)$(GEN_VCPROJ) \
+            --exe \
+            --target=$(TOOLCHAIN) \
+            --name=test_intra_pred_speed \
+            -D_VARIADIC_MAX=10 \
+            --proj-guid=CD837F5F-52D8-4314-A370-895D614166A7 \
+            --ver=$(CONFIG_VS_VERSION) \
+            --src-path-bare="$(SRC_PATH_BARE)" \
+            $(if $(CONFIG_STATIC_MSVCRT),--static-crt) \
+            --out=$@ $(INTERNAL_CFLAGS) $(CFLAGS) \
+            -I. -I"$(SRC_PATH_BARE)/third_party/googletest/src/include" \
+            -L. -l$(CODEC_LIB) -l$(GTEST_LIB) $^
+endif  # TEST_INTRA_PRED_SPEED
 endif
 else

@@ -428,45 +504,54 @@ ifeq ($(filter win%,$(TGT_OS)),$(TGT_OS))
 # Disabling pthreads globally will cause issues on darwin and possibly elsewhere
 $(GTEST_OBJS) $(GTEST_OBJS:.o=.d): CXXFLAGS += -DGTEST_HAS_PTHREAD=0
 endif
-$(GTEST_OBJS) $(GTEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src
-$(GTEST_OBJS) $(GTEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src/include
-OBJS-$(BUILD_LIBVPX) += $(GTEST_OBJS)
-LIBS-$(BUILD_LIBVPX) += $(BUILD_PFX)libgtest.a $(BUILD_PFX)libgtest_g.a
+GTEST_INCLUDES := -I$(SRC_PATH_BARE)/third_party/googletest/src
+GTEST_INCLUDES += -I$(SRC_PATH_BARE)/third_party/googletest/src/include
+$(GTEST_OBJS) $(GTEST_OBJS:.o=.d): CXXFLAGS += $(GTEST_INCLUDES)
+OBJS-yes += $(GTEST_OBJS)
+LIBS-yes += $(BUILD_PFX)libgtest.a $(BUILD_PFX)libgtest_g.a
 $(BUILD_PFX)libgtest_g.a: $(GTEST_OBJS)

 LIBVPX_TEST_OBJS=$(sort $(call objs,$(LIBVPX_TEST_SRCS)))
-$(LIBVPX_TEST_OBJS) $(LIBVPX_TEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src
-$(LIBVPX_TEST_OBJS) $(LIBVPX_TEST_OBJS:.o=.d): CXXFLAGS += -I$(SRC_PATH_BARE)/third_party/googletest/src/include
-OBJS-$(BUILD_LIBVPX) += $(LIBVPX_TEST_OBJS)
-BINS-$(BUILD_LIBVPX) += $(LIBVPX_TEST_BINS)
+$(LIBVPX_TEST_OBJS) $(LIBVPX_TEST_OBJS:.o=.d): CXXFLAGS += $(GTEST_INCLUDES)
+OBJS-yes += $(LIBVPX_TEST_OBJS)
+BINS-yes += $(LIBVPX_TEST_BIN)

 CODEC_LIB=$(if $(CONFIG_DEBUG_LIBS),vpx_g,vpx)
-CODEC_LIB_SUF=$(if $(CONFIG_SHARED),.so,.a)
-$(foreach bin,$(LIBVPX_TEST_BINS),\
-    $(if $(BUILD_LIBVPX),$(eval $(bin): \
-        lib$(CODEC_LIB)$(CODEC_LIB_SUF) libgtest.a ))\
-    $(if $(BUILD_LIBVPX),$(eval $(call linkerxx_template,$(bin),\
-        $(LIBVPX_TEST_OBJS) \
-        -L. -lvpx -lgtest $(extralibs) -lm)\
-        )))\
-    $(if $(LIPO_LIBS),$(eval $(call lipo_bin_template,$(bin))))\
+CODEC_LIB_SUF=$(if $(CONFIG_SHARED),$(SHARED_LIB_SUF),.a)
+TEST_LIBS := lib$(CODEC_LIB)$(CODEC_LIB_SUF) libgtest.a
+$(LIBVPX_TEST_BIN): $(TEST_LIBS)
+$(eval $(call linkerxx_template,$(LIBVPX_TEST_BIN), \
+              $(LIBVPX_TEST_OBJS) \
+              -L. -lvpx -lgtest $(extralibs) -lm))

-endif
+ifneq ($(strip $(TEST_INTRA_PRED_SPEED_OBJS)),)
+$(TEST_INTRA_PRED_SPEED_OBJS) $(TEST_INTRA_PRED_SPEED_OBJS:.o=.d): CXXFLAGS += $(GTEST_INCLUDES)
+OBJS-yes += $(TEST_INTRA_PRED_SPEED_OBJS)
+BINS-yes += $(TEST_INTRA_PRED_SPEED_BIN)
+
+$(TEST_INTRA_PRED_SPEED_BIN): $(TEST_LIBS)
+$(eval $(call linkerxx_template,$(TEST_INTRA_PRED_SPEED_BIN), \
+              $(TEST_INTRA_PRED_SPEED_OBJS) \
+              -L. -lvpx -lgtest $(extralibs) -lm))
+endif  # TEST_INTRA_PRED_SPEED
+
+endif  # CONFIG_UNIT_TESTS

 # Install test sources only if codec source is included
 INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(patsubst $(SRC_PATH_BARE)/%,%,\
    $(shell find $(SRC_PATH_BARE)/third_party/googletest -type f))
 INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(LIBVPX_TEST_SRCS)
+INSTALL-SRCS-$(CONFIG_CODEC_SRCS) += $(TEST_INTRA_PRED_SPEED_SRCS)

 define test_shard_template
 test:: test_shard.$(1)
-test_shard.$(1): $(LIBVPX_TEST_BINS) testdata
+test-no-data-check:: test_shard_ndc.$(1)
+test_shard.$(1) test_shard_ndc.$(1): $(LIBVPX_TEST_BIN)
 	@set -e; \
-	 for t in $(LIBVPX_TEST_BINS); do \
-	   export GTEST_SHARD_INDEX=$(1); \
-	   export GTEST_TOTAL_SHARDS=$(2); \
-	   $$$$t; \
-	 done
+	 export GTEST_SHARD_INDEX=$(1); \
+	 export GTEST_TOTAL_SHARDS=$(2); \
+	 $(LIBVPX_TEST_BIN)
+test_shard.$(1): testdata
 .PHONY: test_shard.$(1)
 endef

@@ -511,15 +596,16 @@ ifeq ($(CONFIG_MSVS),yes)
 # TODO(tomfinegan): Support running the debug versions of tools?
 TEST_BIN_PATH := $(addsuffix /$(TGT_OS:win64=x64)/Release, $(TEST_BIN_PATH))
 endif
-utiltest: testdata
+utiltest utiltest-no-data-check:
 	$(qexec)$(SRC_PATH_BARE)/test/vpxdec.sh \
 		--test-data-path $(LIBVPX_TEST_DATA_PATH) \
 		--bin-path $(TEST_BIN_PATH)
 	$(qexec)$(SRC_PATH_BARE)/test/vpxenc.sh \
 		--test-data-path $(LIBVPX_TEST_DATA_PATH) \
 		--bin-path $(TEST_BIN_PATH)
+utiltest: testdata
 else
-utiltest:
+utiltest utiltest-no-data-check:
 	@echo Unit tests must be enabled to make the utiltest target.
 endif

@@ -537,11 +623,12 @@ ifeq ($(CONFIG_MSVS),yes)
 # TODO(tomfinegan): Support running the debug versions of tools?
 EXAMPLES_BIN_PATH := $(TGT_OS:win64=x64)/Release
 endif
-exampletest: examples testdata
+exampletest exampletest-no-data-check: examples
 	$(qexec)$(SRC_PATH_BARE)/test/examples.sh \
 		--test-data-path $(LIBVPX_TEST_DATA_PATH) \
 		--bin-path $(EXAMPLES_BIN_PATH)
+exampletest: testdata
 else
-exampletest:
+exampletest exampletest-no-data-check:
 	@echo Unit tests must be enabled to make the exampletest target.
 endif
--- a/md5_utils.c
+++ b/md5_utils.c
@@ -24,7 +24,7 @@

 #include "md5_utils.h"

-void
+static void
 byteSwap(UWORD32 *buf, unsigned words) {
  md5byte *p;

--- a/rate_hist.c
+++ b/rate_hist.c
@@ -88,6 +88,9 @@ void update_rate_histogram(struct rate_hist *hist,
  if (now < cfg->rc_buf_initial_sz)
    return;

+  if (!cfg->rc_target_bitrate)
+    return;
+
  then = now;

  /* Sum the size over the past rc_buf_sz ms */
--- a/test/android/Android.mk
+++ b/test/android/Android.mk
@@ -40,9 +40,17 @@ include $(CLEAR_VARS)
 LOCAL_ARM_MODE := arm
 LOCAL_MODULE := libvpx_test
 LOCAL_STATIC_LIBRARIES := gtest libwebm
-LOCAL_SHARED_LIBRARIES := vpx
+
+ifeq ($(ENABLE_SHARED),1)
+  LOCAL_SHARED_LIBRARIES := vpx
+else
+  LOCAL_STATIC_LIBRARIES += vpx
+endif
+
 include $(LOCAL_PATH)/test/test.mk
 LOCAL_C_INCLUDES := $(BINDINGS_DIR)
 FILTERED_SRC := $(sort $(filter %.cc %.c, $(LIBVPX_TEST_SRCS-yes)))
 LOCAL_SRC_FILES := $(addprefix ./test/, $(FILTERED_SRC))
+# some test files depend on *_rtcd.h, ensure they're generated first.
+$(eval $(call rtcd_dep_template))
 include $(BUILD_EXECUTABLE)
--- a/test/blockiness_test.cc
+++ b/test/blockiness_test.cc
@@ -0,0 +1,229 @@
+/*
+ *  Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <limits.h>
+#include <stdio.h>
+#include <string.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_config.h"
+#if CONFIG_VP9_ENCODER
+#include "./vp9_rtcd.h"
+#endif
+
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
+#include "test/util.h"
+
+#include "vpx_mem/vpx_mem.h"
+
+
+extern "C"
+double vp9_get_blockiness(const unsigned char *img1, int img1_pitch,
+                          const unsigned char *img2, int img2_pitch,
+                          int width, int height);
+
+using libvpx_test::ACMRandom;
+
+namespace {
+class BlockinessTestBase : public ::testing::Test {
+ public:
+  BlockinessTestBase(int width, int height) : width_(width), height_(height) {}
+
+  static void SetUpTestCase() {
+    source_data_ = reinterpret_cast<uint8_t*>(
+        vpx_memalign(kDataAlignment, kDataBufferSize));
+    reference_data_ = reinterpret_cast<uint8_t*>(
+        vpx_memalign(kDataAlignment, kDataBufferSize));
+  }
+
+  static void TearDownTestCase() {
+    vpx_free(source_data_);
+    source_data_ = NULL;
+    vpx_free(reference_data_);
+    reference_data_ = NULL;
+  }
+
+  virtual void TearDown() {
+    libvpx_test::ClearSystemState();
+  }
+
+ protected:
+  // Handle frames up to 640x480
+  static const int kDataAlignment = 16;
+  static const int kDataBufferSize = 640*480;
+
+  virtual void SetUp() {
+    source_stride_ = (width_ + 31) & ~31;
+    reference_stride_ = width_ * 2;
+    rnd_.Reset(ACMRandom::DeterministicSeed());
+  }
+
+  void FillConstant(uint8_t *data, int stride, uint8_t fill_constant,
+                    int width, int height) {
+    for (int h = 0; h < height; ++h) {
+      for (int w = 0; w < width; ++w) {
+        data[h * stride + w] = fill_constant;
+      }
+    }
+  }
+
+  void FillConstant(uint8_t *data, int stride, uint8_t fill_constant) {
+    FillConstant(data, stride, fill_constant, width_, height_);
+  }
+
+  void FillRandom(uint8_t *data, int stride, int width, int height) {
+    for (int h = 0; h < height; ++h) {
+      for (int w = 0; w < width; ++w) {
+        data[h * stride + w] = rnd_.Rand8();
+      }
+    }
+  }
+
+  void FillRandom(uint8_t *data, int stride) {
+    FillRandom(data, stride, width_, height_);
+  }
+
+  void FillRandomBlocky(uint8_t *data, int stride) {
+    for (int h = 0; h < height_; h += 4) {
+      for (int w = 0; w < width_; w += 4) {
+        FillRandom(data + h * stride + w, stride, 4, 4);
+      }
+    }
+  }
+
+  void FillCheckerboard(uint8_t *data, int stride) {
+    for (int h = 0; h < height_; h += 4) {
+      for (int w = 0; w < width_; w += 4) {
+        if (((h/4) ^ (w/4)) & 1)
+          FillConstant(data + h * stride + w, stride, 255, 4, 4);
+        else
+          FillConstant(data + h * stride + w, stride, 0, 4, 4);
+      }
+    }
+  }
+
+  void Blur(uint8_t *data, int stride, int taps) {
+    int sum = 0;
+    int half_taps = taps / 2;
+    for (int h = 0; h < height_; ++h) {
+      for (int w = 0; w < taps; ++w) {
+        sum += data[w + h * stride];
+      }
+      for (int w = taps; w < width_; ++w) {
+        sum += data[w + h * stride] - data[w - taps + h * stride];
+        data[w - half_taps + h * stride] = (sum + half_taps) / taps;
+      }
+    }
+    for (int w = 0; w < width_; ++w) {
+      for (int h = 0; h < taps; ++h) {
+        sum += data[h + w * stride];
+      }
+      for (int h = taps; h < height_; ++h) {
+        sum += data[w + h * stride] - data[(h - taps) * stride + w];
+        data[(h - half_taps) * stride + w] = (sum + half_taps) / taps;
+      }
+    }
+  }
+  int width_, height_;
+  static uint8_t* source_data_;
+  int source_stride_;
+  static uint8_t* reference_data_;
+  int reference_stride_;
+
+  ACMRandom rnd_;
+};
+
+#if CONFIG_VP9_ENCODER
+typedef std::tr1::tuple<int, int> BlockinessParam;
+class BlockinessVP9Test
+    : public BlockinessTestBase,
+      public ::testing::WithParamInterface<BlockinessParam> {
+ public:
+  BlockinessVP9Test() : BlockinessTestBase(GET_PARAM(0), GET_PARAM(1)) {}
+
+ protected:
+  int CheckBlockiness() {
+    return vp9_get_blockiness(source_data_, source_stride_,
+                              reference_data_, reference_stride_,
+                              width_, height_);
+  }
+};
+#endif  // CONFIG_VP9_ENCODER
+
+uint8_t* BlockinessTestBase::source_data_ = NULL;
+uint8_t* BlockinessTestBase::reference_data_ = NULL;
+
+#if CONFIG_VP9_ENCODER
+TEST_P(BlockinessVP9Test, SourceBlockierThanReference) {
+  // Source is blockier than reference.
+  FillRandomBlocky(source_data_, source_stride_);
+  FillConstant(reference_data_, reference_stride_, 128);
+  int super_blocky = CheckBlockiness();
+
+  EXPECT_EQ(0, super_blocky) << "Blocky source should produce 0 blockiness.";
+}
+
+TEST_P(BlockinessVP9Test, ReferenceBlockierThanSource) {
+  // Source is blockier than reference.
+  FillConstant(source_data_, source_stride_, 128);
+  FillRandomBlocky(reference_data_, reference_stride_);
+  int super_blocky = CheckBlockiness();
+
+  EXPECT_GT(super_blocky, 0.0)
+      << "Blocky reference should score high for blockiness.";
+}
+
+TEST_P(BlockinessVP9Test, BlurringDecreasesBlockiness) {
+  // Source is blockier than reference.
+  FillConstant(source_data_, source_stride_, 128);
+  FillRandomBlocky(reference_data_, reference_stride_);
+  int super_blocky = CheckBlockiness();
+
+  Blur(reference_data_, reference_stride_, 4);
+  int less_blocky = CheckBlockiness();
+
+  EXPECT_GT(super_blocky, less_blocky)
+      << "A straight blur should decrease blockiness.";
+}
+
+TEST_P(BlockinessVP9Test, WorstCaseBlockiness) {
+  // Source is blockier than reference.
+  FillConstant(source_data_, source_stride_, 128);
+  FillCheckerboard(reference_data_, reference_stride_);
+
+  int super_blocky = CheckBlockiness();
+
+  Blur(reference_data_, reference_stride_, 4);
+  int less_blocky = CheckBlockiness();
+
+  EXPECT_GT(super_blocky, less_blocky)
+      << "A straight blur should decrease blockiness.";
+}
+#endif  // CONFIG_VP9_ENCODER
+
+
+using std::tr1::make_tuple;
+
+//------------------------------------------------------------------------------
+// C functions
+
+#if CONFIG_VP9_ENCODER
+const BlockinessParam c_vp9_tests[] = {
+  make_tuple(320, 240),
+  make_tuple(318, 242),
+  make_tuple(318, 238),
+};
+INSTANTIATE_TEST_CASE_P(C, BlockinessVP9Test, ::testing::ValuesIn(c_vp9_tests));
+#endif
+
+}  // namespace
--- a/test/borders_test.cc
+++ b/test/borders_test.cc
@@ -80,4 +80,7 @@ TEST_P(BordersTest, TestLowBitrate) {

 VP9_INSTANTIATE_TEST_CASE(BordersTest, ::testing::Values(
    ::libvpx_test::kTwoPassGood));
+
+VP10_INSTANTIATE_TEST_CASE(BordersTest, ::testing::Values(
+    ::libvpx_test::kTwoPassGood));
 }  // namespace
--- a/test/codec_factory.h
+++ b/test/codec_factory.h
@@ -13,10 +13,10 @@
 #include "./vpx_config.h"
 #include "vpx/vpx_decoder.h"
 #include "vpx/vpx_encoder.h"
-#if CONFIG_VP8_ENCODER || CONFIG_VP9_ENCODER
+#if CONFIG_VP8_ENCODER || CONFIG_VP9_ENCODER || CONFIG_VP10_ENCODER
 #include "vpx/vp8cx.h"
 #endif
-#if CONFIG_VP8_DECODER || CONFIG_VP9_DECODER
+#if CONFIG_VP8_DECODER || CONFIG_VP9_DECODER || CONFIG_VP10_DECODER
 #include "vpx/vp8dx.h"
 #endif

@@ -233,6 +233,8 @@ class VP9CodecFactory : public CodecFactory {
                                               int usage) const {
 #if CONFIG_VP9_ENCODER
    return vpx_codec_enc_config_default(&vpx_codec_vp9_cx_algo, cfg, usage);
+#elif CONFIG_VP10_ENCODER
+    return vpx_codec_enc_config_default(&vpx_codec_vp10_cx_algo, cfg, usage);
 #else
    return VPX_CODEC_INCAPABLE;
 #endif
@@ -251,7 +253,96 @@ const libvpx_test::VP9CodecFactory kVP9;
 #define VP9_INSTANTIATE_TEST_CASE(test, ...)
 #endif  // CONFIG_VP9

+/*
+ * VP10 Codec Definitions
+ */
+#if CONFIG_VP10
+class VP10Decoder : public Decoder {
+ public:
+  VP10Decoder(vpx_codec_dec_cfg_t cfg, unsigned long deadline)
+      : Decoder(cfg, deadline) {}
+
+  VP10Decoder(vpx_codec_dec_cfg_t cfg, const vpx_codec_flags_t flag,
+              unsigned long deadline)  // NOLINT
+      : Decoder(cfg, flag, deadline) {}
+
+ protected:
+  virtual vpx_codec_iface_t* CodecInterface() const {
+#if CONFIG_VP10_DECODER
+    return &vpx_codec_vp10_dx_algo;
+#else
+    return NULL;
+#endif
+  }
+};
+
+class VP10Encoder : public Encoder {
+ public:
+  VP10Encoder(vpx_codec_enc_cfg_t cfg, unsigned long deadline,
+              const unsigned long init_flags, TwopassStatsStore *stats)
+      : Encoder(cfg, deadline, init_flags, stats) {}
+
+ protected:
+  virtual vpx_codec_iface_t* CodecInterface() const {
+#if CONFIG_VP10_ENCODER
+    return &vpx_codec_vp10_cx_algo;
+#else
+    return NULL;
+#endif
+  }
+};
+
+class VP10CodecFactory : public CodecFactory {
+ public:
+  VP10CodecFactory() : CodecFactory() {}
+
+  virtual Decoder* CreateDecoder(vpx_codec_dec_cfg_t cfg,
+                                 unsigned long deadline) const {
+    return CreateDecoder(cfg, 0, deadline);
+  }
+
+  virtual Decoder* CreateDecoder(vpx_codec_dec_cfg_t cfg,
+                                 const vpx_codec_flags_t flags,
+                                 unsigned long deadline) const {  // NOLINT
+#if CONFIG_VP10_DECODER
+    return new VP10Decoder(cfg, flags, deadline);
+#else
+    return NULL;
+#endif
+  }
+
+  virtual Encoder* CreateEncoder(vpx_codec_enc_cfg_t cfg,
+                                 unsigned long deadline,
+                                 const unsigned long init_flags,
+                                 TwopassStatsStore *stats) const {
+#if CONFIG_VP10_ENCODER
+    return new VP10Encoder(cfg, deadline, init_flags, stats);
+#else
+    return NULL;
+#endif
+  }
+
+  virtual vpx_codec_err_t DefaultEncoderConfig(vpx_codec_enc_cfg_t *cfg,
+                                               int usage) const {
+#if CONFIG_VP10_ENCODER
+    return vpx_codec_enc_config_default(&vpx_codec_vp10_cx_algo, cfg, usage);
+#else
+    return VPX_CODEC_INCAPABLE;
+#endif
+  }
+};
+
+const libvpx_test::VP10CodecFactory kVP10;
+
+#define VP10_INSTANTIATE_TEST_CASE(test, ...)\
+  INSTANTIATE_TEST_CASE_P(VP10, test, \
+      ::testing::Combine( \
+          ::testing::Values(static_cast<const libvpx_test::CodecFactory*>( \
+               &libvpx_test::kVP10)), \
+          __VA_ARGS__))
+#else
+#define VP10_INSTANTIATE_TEST_CASE(test, ...)
+#endif  // CONFIG_VP10

 }  // namespace libvpx_test
-
 #endif  // TEST_CODEC_FACTORY_H_
--- a/test/consistency_test.cc
+++ b/test/consistency_test.cc
@@ -0,0 +1,224 @@
+/*
+ *  Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <limits.h>
+#include <stdio.h>
+#include <string.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_config.h"
+#if CONFIG_VP9_ENCODER
+#include "./vp9_rtcd.h"
+#endif
+
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
+#include "test/util.h"
+#include "vpx_dsp/ssim.h"
+#include "vpx_mem/vpx_mem.h"
+
+extern "C"
+double vpx_get_ssim_metrics(uint8_t *img1, int img1_pitch,
+                            uint8_t *img2, int img2_pitch,
+                            int width, int height,
+                            Ssimv *sv2, Metrics *m,
+                            int do_inconsistency);
+
+using libvpx_test::ACMRandom;
+
+namespace {
+class ConsistencyTestBase : public ::testing::Test {
+ public:
+  ConsistencyTestBase(int width, int height) : width_(width), height_(height) {}
+
+  static void SetUpTestCase() {
+    source_data_[0] = reinterpret_cast<uint8_t*>(
+        vpx_memalign(kDataAlignment, kDataBufferSize));
+    reference_data_[0] = reinterpret_cast<uint8_t*>(
+        vpx_memalign(kDataAlignment, kDataBufferSize));
+    source_data_[1] = reinterpret_cast<uint8_t*>(
+        vpx_memalign(kDataAlignment, kDataBufferSize));
+    reference_data_[1] = reinterpret_cast<uint8_t*>(
+        vpx_memalign(kDataAlignment, kDataBufferSize));
+    ssim_array_ = new Ssimv[kDataBufferSize / 16];
+  }
+
+  static void ClearSsim() {
+    memset(ssim_array_, 0, kDataBufferSize / 16);
+  }
+  static void TearDownTestCase() {
+    vpx_free(source_data_[0]);
+    source_data_[0] = NULL;
+    vpx_free(reference_data_[0]);
+    reference_data_[0] = NULL;
+    vpx_free(source_data_[1]);
+    source_data_[1] = NULL;
+    vpx_free(reference_data_[1]);
+    reference_data_[1] = NULL;
+
+    delete[] ssim_array_;
+  }
+
+  virtual void TearDown() {
+    libvpx_test::ClearSystemState();
+  }
+
+ protected:
+  // Handle frames up to 640x480
+  static const int kDataAlignment = 16;
+  static const int kDataBufferSize = 640*480;
+
+  virtual void SetUp() {
+    source_stride_ = (width_ + 31) & ~31;
+    reference_stride_ = width_ * 2;
+    rnd_.Reset(ACMRandom::DeterministicSeed());
+  }
+
+  void FillRandom(uint8_t *data, int stride, int width, int height) {
+    for (int h = 0; h < height; ++h) {
+      for (int w = 0; w < width; ++w) {
+        data[h * stride + w] = rnd_.Rand8();
+      }
+    }
+  }
+
+  void FillRandom(uint8_t *data, int stride) {
+    FillRandom(data, stride, width_, height_);
+  }
+
+  void Copy(uint8_t *reference, uint8_t *source) {
+    memcpy(reference, source, kDataBufferSize);
+  }
+
+  void Blur(uint8_t *data, int stride, int taps) {
+    int sum = 0;
+    int half_taps = taps / 2;
+    for (int h = 0; h < height_; ++h) {
+      for (int w = 0; w < taps; ++w) {
+        sum += data[w + h * stride];
+      }
+      for (int w = taps; w < width_; ++w) {
+        sum += data[w + h * stride] - data[w - taps + h * stride];
+        data[w - half_taps + h * stride] = (sum + half_taps) / taps;
+      }
+    }
+    for (int w = 0; w < width_; ++w) {
+      for (int h = 0; h < taps; ++h) {
+        sum += data[h + w * stride];
+      }
+      for (int h = taps; h < height_; ++h) {
+        sum += data[w + h * stride] - data[(h - taps) * stride + w];
+        data[(h - half_taps) * stride + w] = (sum + half_taps) / taps;
+      }
+    }
+  }
+  int width_, height_;
+  static uint8_t* source_data_[2];
+  int source_stride_;
+  static uint8_t* reference_data_[2];
+  int reference_stride_;
+  static Ssimv *ssim_array_;
+  Metrics metrics_;
+
+  ACMRandom rnd_;
+};
+
+#if CONFIG_VP9_ENCODER
+typedef std::tr1::tuple<int, int> ConsistencyParam;
+class ConsistencyVP9Test
+    : public ConsistencyTestBase,
+      public ::testing::WithParamInterface<ConsistencyParam> {
+ public:
+  ConsistencyVP9Test() : ConsistencyTestBase(GET_PARAM(0), GET_PARAM(1)) {}
+
+ protected:
+  double CheckConsistency(int frame) {
+    EXPECT_LT(frame, 2)<< "Frame to check has to be less than 2.";
+    return
+        vpx_get_ssim_metrics(source_data_[frame], source_stride_,
+                             reference_data_[frame], reference_stride_,
+                             width_, height_, ssim_array_, &metrics_, 1);
+  }
+};
+#endif  // CONFIG_VP9_ENCODER
+
+uint8_t* ConsistencyTestBase::source_data_[2] = {NULL, NULL};
+uint8_t* ConsistencyTestBase::reference_data_[2] = {NULL, NULL};
+Ssimv* ConsistencyTestBase::ssim_array_ = NULL;
+
+#if CONFIG_VP9_ENCODER
+TEST_P(ConsistencyVP9Test, ConsistencyIsZero) {
+  FillRandom(source_data_[0], source_stride_);
+  Copy(source_data_[1], source_data_[0]);
+  Copy(reference_data_[0], source_data_[0]);
+  Blur(reference_data_[0], reference_stride_, 3);
+  Copy(reference_data_[1], source_data_[0]);
+  Blur(reference_data_[1], reference_stride_, 3);
+
+  double inconsistency = CheckConsistency(1);
+  inconsistency = CheckConsistency(0);
+  EXPECT_EQ(inconsistency, 0.0)
+      << "Should have 0 inconsistency if they are exactly the same.";
+
+  // If sources are not consistent reference frames inconsistency should
+  // be less than if the source is consistent.
+  FillRandom(source_data_[0], source_stride_);
+  FillRandom(source_data_[1], source_stride_);
+  FillRandom(reference_data_[0], reference_stride_);
+  FillRandom(reference_data_[1], reference_stride_);
+  CheckConsistency(0);
+  inconsistency = CheckConsistency(1);
+
+  Copy(source_data_[1], source_data_[0]);
+  CheckConsistency(0);
+  double inconsistency2 = CheckConsistency(1);
+  EXPECT_LT(inconsistency, inconsistency2)
+      << "Should have less inconsistency if source itself is inconsistent.";
+
+  // Less of a blur should be less inconsistent than more blur coming off a
+  // a frame with no blur.
+  ClearSsim();
+  FillRandom(source_data_[0], source_stride_);
+  Copy(source_data_[1], source_data_[0]);
+  Copy(reference_data_[0], source_data_[0]);
+  Copy(reference_data_[1], source_data_[0]);
+  Blur(reference_data_[1], reference_stride_, 4);
+  CheckConsistency(0);
+  inconsistency = CheckConsistency(1);
+  ClearSsim();
+  Copy(reference_data_[1], source_data_[0]);
+  Blur(reference_data_[1], reference_stride_, 8);
+  CheckConsistency(0);
+  inconsistency2 = CheckConsistency(1);
+
+  EXPECT_LT(inconsistency, inconsistency2)
+      << "Stronger Blur should produce more inconsistency.";
+}
+#endif  // CONFIG_VP9_ENCODER
+
+
+using std::tr1::make_tuple;
+
+//------------------------------------------------------------------------------
+// C functions
+
+#if CONFIG_VP9_ENCODER
+const ConsistencyParam c_vp9_tests[] = {
+  make_tuple(320, 240),
+  make_tuple(318, 242),
+  make_tuple(318, 238),
+};
+INSTANTIATE_TEST_CASE_P(C, ConsistencyVP9Test,
+                        ::testing::ValuesIn(c_vp9_tests));
+#endif
+
+}  // namespace
--- a/test/convolve_test.cc
+++ b/test/convolve_test.cc
@@ -9,16 +9,20 @@
 */

 #include <string.h>
-#include "test/acm_random.h"
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
-#include "test/util.h"
+
 #include "third_party/googletest/src/include/gtest/gtest.h"

 #include "./vpx_config.h"
 #include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
+#include "test/util.h"
 #include "vp9/common/vp9_common.h"
 #include "vp9/common/vp9_filter.h"
+#include "vpx_dsp/vpx_dsp_common.h"
+#include "vpx_dsp/vpx_filter.h"
 #include "vpx_mem/vpx_mem.h"
 #include "vpx_ports/mem.h"

@@ -37,9 +41,14 @@ struct ConvolveFunctions {
                    ConvolveFunc h8, ConvolveFunc h8_avg,
                    ConvolveFunc v8, ConvolveFunc v8_avg,
                    ConvolveFunc hv8, ConvolveFunc hv8_avg,
+                    ConvolveFunc sh8, ConvolveFunc sh8_avg,
+                    ConvolveFunc sv8, ConvolveFunc sv8_avg,
+                    ConvolveFunc shv8, ConvolveFunc shv8_avg,
                    int bd)
      : copy_(copy), avg_(avg), h8_(h8), v8_(v8), hv8_(hv8), h8_avg_(h8_avg),
-        v8_avg_(v8_avg), hv8_avg_(hv8_avg), use_highbd_(bd) {}
+        v8_avg_(v8_avg), hv8_avg_(hv8_avg), sh8_(sh8), sv8_(sv8), shv8_(shv8),
+        sh8_avg_(sh8_avg), sv8_avg_(sv8_avg), shv8_avg_(shv8_avg),
+        use_highbd_(bd) {}

  ConvolveFunc copy_;
  ConvolveFunc avg_;
@@ -49,6 +58,12 @@ struct ConvolveFunctions {
  ConvolveFunc h8_avg_;
  ConvolveFunc v8_avg_;
  ConvolveFunc hv8_avg_;
+  ConvolveFunc sh8_;        // scaled horiz
+  ConvolveFunc sv8_;        // scaled vert
+  ConvolveFunc shv8_;       // scaled horiz/vert
+  ConvolveFunc sh8_avg_;    // scaled avg horiz
+  ConvolveFunc sv8_avg_;    // scaled avg vert
+  ConvolveFunc shv8_avg_;   // scaled avg horiz/vert
  int use_highbd_;  // 0 if high bitdepth not used, else the actual bit depth.
 };

@@ -398,9 +413,9 @@ class ConvolveTest : public ::testing::TestWithParam<ConvolveParam> {
  }

  void CopyOutputToRef() {
-    vpx_memcpy(output_ref_, output_, kOutputBufferSize);
+    memcpy(output_ref_, output_, kOutputBufferSize);
 #if CONFIG_VP9_HIGHBITDEPTH
-    vpx_memcpy(output16_ref_, output16_, kOutputBufferSize);
+    memcpy(output16_ref_, output16_, kOutputBufferSize);
 #endif
  }

@@ -593,8 +608,8 @@ TEST_P(ConvolveTest, CopyHoriz) {
  DECLARE_ALIGNED(256, const int16_t, filter8[8]) = {0, 0, 0, 128, 0, 0, 0, 0};

  ASM_REGISTER_STATE_CHECK(
-      UUT_->h8_(in, kInputStride, out, kOutputStride, filter8, 16, filter8, 16,
-                Width(), Height()));
+      UUT_->sh8_(in, kInputStride, out, kOutputStride, filter8, 16, filter8, 16,
+                 Width(), Height()));

  CheckGuardBlocks();

@@ -611,8 +626,8 @@ TEST_P(ConvolveTest, CopyVert) {
  DECLARE_ALIGNED(256, const int16_t, filter8[8]) = {0, 0, 0, 128, 0, 0, 0, 0};

  ASM_REGISTER_STATE_CHECK(
-      UUT_->v8_(in, kInputStride, out, kOutputStride, filter8, 16, filter8, 16,
-                Width(), Height()));
+      UUT_->sv8_(in, kInputStride, out, kOutputStride, filter8, 16, filter8, 16,
+                 Width(), Height()));

  CheckGuardBlocks();

@@ -629,8 +644,8 @@ TEST_P(ConvolveTest, Copy2D) {
  DECLARE_ALIGNED(256, const int16_t, filter8[8]) = {0, 0, 0, 128, 0, 0, 0, 0};

  ASM_REGISTER_STATE_CHECK(
-      UUT_->hv8_(in, kInputStride, out, kOutputStride, filter8, 16, filter8, 16,
-                 Width(), Height()));
+      UUT_->shv8_(in, kInputStride, out, kOutputStride, filter8, 16, filter8,
+                  16, Width(), Height()));

  CheckGuardBlocks();

@@ -647,7 +662,7 @@ const int kNumFilters = 16;
 TEST(ConvolveTest, FiltersWontSaturateWhenAddedPairwise) {
  for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
    const InterpKernel *filters =
-        vp9_get_interp_kernel(static_cast<INTERP_FILTER>(filter_bank));
+        vp9_filter_kernels[static_cast<INTERP_FILTER>(filter_bank)];
    for (int i = 0; i < kNumFilters; i++) {
      const int p0 = filters[i][0] + filters[i][1];
      const int p1 = filters[i][2] + filters[i][3];
@@ -685,9 +700,7 @@ TEST_P(ConvolveTest, MatchesReferenceSubpixelFilter) {

  for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
    const InterpKernel *filters =
-        vp9_get_interp_kernel(static_cast<INTERP_FILTER>(filter_bank));
-    const InterpKernel *const eighttap_smooth =
-        vp9_get_interp_kernel(EIGHTTAP_SMOOTH);
+        vp9_filter_kernels[static_cast<INTERP_FILTER>(filter_bank)];

    for (int filter_x = 0; filter_x < kNumFilters; ++filter_x) {
      for (int filter_y = 0; filter_y < kNumFilters; ++filter_y) {
@@ -696,7 +709,7 @@ TEST_P(ConvolveTest, MatchesReferenceSubpixelFilter) {
                                   ref, kOutputStride,
                                   Width(), Height());

-        if (filters == eighttap_smooth || (filter_x && filter_y))
+        if (filter_x && filter_y)
          ASM_REGISTER_STATE_CHECK(
              UUT_->hv8_(in, kInputStride, out, kOutputStride,
                         filters[filter_x], 16, filters[filter_y], 16,
@@ -706,11 +719,16 @@ TEST_P(ConvolveTest, MatchesReferenceSubpixelFilter) {
              UUT_->v8_(in, kInputStride, out, kOutputStride,
                        kInvalidFilter, 16, filters[filter_y], 16,
                        Width(), Height()));
-        else
+        else if (filter_x)
          ASM_REGISTER_STATE_CHECK(
              UUT_->h8_(in, kInputStride, out, kOutputStride,
                        filters[filter_x], 16, kInvalidFilter, 16,
                        Width(), Height()));
+        else
+          ASM_REGISTER_STATE_CHECK(
+              UUT_->copy_(in, kInputStride, out, kOutputStride,
+                          kInvalidFilter, 0, kInvalidFilter, 0,
+                          Width(), Height()));

        CheckGuardBlocks();

@@ -764,9 +782,7 @@ TEST_P(ConvolveTest, MatchesReferenceAveragingSubpixelFilter) {

  for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
    const InterpKernel *filters =
-        vp9_get_interp_kernel(static_cast<INTERP_FILTER>(filter_bank));
-    const InterpKernel *const eighttap_smooth =
-        vp9_get_interp_kernel(EIGHTTAP_SMOOTH);
+        vp9_filter_kernels[static_cast<INTERP_FILTER>(filter_bank)];

    for (int filter_x = 0; filter_x < kNumFilters; ++filter_x) {
      for (int filter_y = 0; filter_y < kNumFilters; ++filter_y) {
@@ -775,7 +791,7 @@ TEST_P(ConvolveTest, MatchesReferenceAveragingSubpixelFilter) {
                                           ref, kOutputStride,
                                           Width(), Height());

-        if (filters == eighttap_smooth || (filter_x && filter_y))
+        if (filter_x && filter_y)
          ASM_REGISTER_STATE_CHECK(
              UUT_->hv8_avg_(in, kInputStride, out, kOutputStride,
                             filters[filter_x], 16, filters[filter_y], 16,
@@ -783,13 +799,18 @@ TEST_P(ConvolveTest, MatchesReferenceAveragingSubpixelFilter) {
        else if (filter_y)
          ASM_REGISTER_STATE_CHECK(
              UUT_->v8_avg_(in, kInputStride, out, kOutputStride,
-                            filters[filter_x], 16, filters[filter_y], 16,
+                            kInvalidFilter, 16, filters[filter_y], 16,
+                            Width(), Height()));
+        else if (filter_x)
+          ASM_REGISTER_STATE_CHECK(
+              UUT_->h8_avg_(in, kInputStride, out, kOutputStride,
+                            filters[filter_x], 16, kInvalidFilter, 16,
                            Width(), Height()));
        else
          ASM_REGISTER_STATE_CHECK(
-              UUT_->h8_avg_(in, kInputStride, out, kOutputStride,
-                            filters[filter_x], 16, filters[filter_y], 16,
-                            Width(), Height()));
+              UUT_->avg_(in, kInputStride, out, kOutputStride,
+                          kInvalidFilter, 0, kInvalidFilter, 0,
+                          Width(), Height()));

        CheckGuardBlocks();

@@ -863,16 +884,14 @@ TEST_P(ConvolveTest, FilterExtremes) {

      for (int filter_bank = 0; filter_bank < kNumFilterBanks; ++filter_bank) {
        const InterpKernel *filters =
-            vp9_get_interp_kernel(static_cast<INTERP_FILTER>(filter_bank));
-        const InterpKernel *const eighttap_smooth =
-            vp9_get_interp_kernel(EIGHTTAP_SMOOTH);
+            vp9_filter_kernels[static_cast<INTERP_FILTER>(filter_bank)];
        for (int filter_x = 0; filter_x < kNumFilters; ++filter_x) {
          for (int filter_y = 0; filter_y < kNumFilters; ++filter_y) {
            wrapper_filter_block2d_8_c(in, kInputStride,
                                       filters[filter_x], filters[filter_y],
                                       ref, kOutputStride,
                                       Width(), Height());
-            if (filters == eighttap_smooth || (filter_x && filter_y))
+            if (filter_x && filter_y)
              ASM_REGISTER_STATE_CHECK(
                  UUT_->hv8_(in, kInputStride, out, kOutputStride,
                             filters[filter_x], 16, filters[filter_y], 16,
@@ -882,11 +901,16 @@ TEST_P(ConvolveTest, FilterExtremes) {
                  UUT_->v8_(in, kInputStride, out, kOutputStride,
                            kInvalidFilter, 16, filters[filter_y], 16,
                            Width(), Height()));
-            else
+            else if (filter_x)
              ASM_REGISTER_STATE_CHECK(
                  UUT_->h8_(in, kInputStride, out, kOutputStride,
                            filters[filter_x], 16, kInvalidFilter, 16,
                            Width(), Height()));
+            else
+              ASM_REGISTER_STATE_CHECK(
+                  UUT_->copy_(in, kInputStride, out, kOutputStride,
+                              kInvalidFilter, 0, kInvalidFilter, 0,
+                              Width(), Height()));

            for (int y = 0; y < Height(); ++y)
              for (int x = 0; x < Width(); ++x)
@@ -902,132 +926,22 @@ TEST_P(ConvolveTest, FilterExtremes) {
  }
 }

-DECLARE_ALIGNED(256, const int16_t, kChangeFilters[16][8]) = {
-    { 0,   0,   0,   0,   0,   0,   0, 128},
-    { 0,   0,   0,   0,   0,   0, 128},
-    { 0,   0,   0,   0,   0, 128},
-    { 0,   0,   0,   0, 128},
-    { 0,   0,   0, 128},
-    { 0,   0, 128},
-    { 0, 128},
-    { 128},
-    { 0,   0,   0,   0,   0,   0,   0, 128},
-    { 0,   0,   0,   0,   0,   0, 128},
-    { 0,   0,   0,   0,   0, 128},
-    { 0,   0,   0,   0, 128},
-    { 0,   0,   0, 128},
-    { 0,   0, 128},
-    { 0, 128},
-    { 128}
-};
-
-/* This test exercises the horizontal and vertical filter functions. */
-TEST_P(ConvolveTest, ChangeFilterWorks) {
-  uint8_t* const in = input();
-  uint8_t* const out = output();
-
-  /* Assume that the first input sample is at the 8/16th position. */
-  const int kInitialSubPelOffset = 8;
-
-  /* Filters are 8-tap, so the first filter tap will be applied to the pixel
-   * at position -3 with respect to the current filtering position. Since
-   * kInitialSubPelOffset is set to 8, we first select sub-pixel filter 8,
-   * which is non-zero only in the last tap. So, applying the filter at the
-   * current input position will result in an output equal to the pixel at
-   * offset +4 (-3 + 7) with respect to the current filtering position.
-   */
-  const int kPixelSelected = 4;
-
-  /* Assume that each output pixel requires us to step on by 17/16th pixels in
-   * the input.
-   */
-  const int kInputPixelStep = 17;
-
-  /* The filters are setup in such a way that the expected output produces
-   * sets of 8 identical output samples. As the filter position moves to the
-   * next 1/16th pixel position the only active (=128) filter tap moves one
-   * position to the left, resulting in the same input pixel being replicated
-   * in to the output for 8 consecutive samples. After each set of 8 positions
-   * the filters select a different input pixel. kFilterPeriodAdjust below
-   * computes which input pixel is written to the output for a specified
-   * x or y position.
-   */
-
-  /* Test the horizontal filter. */
-  ASM_REGISTER_STATE_CHECK(
-      UUT_->h8_(in, kInputStride, out, kOutputStride,
-                kChangeFilters[kInitialSubPelOffset],
-                kInputPixelStep, NULL, 0, Width(), Height()));
-
-  for (int x = 0; x < Width(); ++x) {
-    const int kFilterPeriodAdjust = (x >> 3) << 3;
-    const int ref_x =
-        kPixelSelected + ((kInitialSubPelOffset
-            + kFilterPeriodAdjust * kInputPixelStep)
-                          >> SUBPEL_BITS);
-    ASSERT_EQ(lookup(in, ref_x), lookup(out, x))
-        << "x == " << x << "width = " << Width();
-  }
-
-  /* Test the vertical filter. */
-  ASM_REGISTER_STATE_CHECK(
-      UUT_->v8_(in, kInputStride, out, kOutputStride,
-                NULL, 0, kChangeFilters[kInitialSubPelOffset],
-                kInputPixelStep, Width(), Height()));
-
-  for (int y = 0; y < Height(); ++y) {
-    const int kFilterPeriodAdjust = (y >> 3) << 3;
-    const int ref_y =
-        kPixelSelected + ((kInitialSubPelOffset
-            + kFilterPeriodAdjust * kInputPixelStep)
-                          >> SUBPEL_BITS);
-    ASSERT_EQ(lookup(in, ref_y * kInputStride), lookup(out, y * kInputStride))
-        << "y == " << y;
-  }
-
-  /* Test the horizontal and vertical filters in combination. */
-  ASM_REGISTER_STATE_CHECK(
-      UUT_->hv8_(in, kInputStride, out, kOutputStride,
-                 kChangeFilters[kInitialSubPelOffset], kInputPixelStep,
-                 kChangeFilters[kInitialSubPelOffset], kInputPixelStep,
-                 Width(), Height()));
-
-  for (int y = 0; y < Height(); ++y) {
-    const int kFilterPeriodAdjustY = (y >> 3) << 3;
-    const int ref_y =
-        kPixelSelected + ((kInitialSubPelOffset
-            + kFilterPeriodAdjustY * kInputPixelStep)
-                          >> SUBPEL_BITS);
-    for (int x = 0; x < Width(); ++x) {
-      const int kFilterPeriodAdjustX = (x >> 3) << 3;
-      const int ref_x =
-          kPixelSelected + ((kInitialSubPelOffset
-              + kFilterPeriodAdjustX * kInputPixelStep)
-                            >> SUBPEL_BITS);
-
-      ASSERT_EQ(lookup(in, ref_y * kInputStride + ref_x),
-                lookup(out, y * kOutputStride + x))
-          << "x == " << x << ", y == " << y;
-    }
-  }
-}
-
 /* This test exercises that enough rows and columns are filtered with every
   possible initial fractional positions and scaling steps. */
 TEST_P(ConvolveTest, CheckScalingFiltering) {
  uint8_t* const in = input();
  uint8_t* const out = output();
-  const InterpKernel *const eighttap = vp9_get_interp_kernel(EIGHTTAP);
+  const InterpKernel *const eighttap = vp9_filter_kernels[EIGHTTAP];

  SetConstantInput(127);

  for (int frac = 0; frac < 16; ++frac) {
    for (int step = 1; step <= 32; ++step) {
      /* Test the horizontal and vertical filters in combination. */
-      ASM_REGISTER_STATE_CHECK(UUT_->hv8_(in, kInputStride, out, kOutputStride,
-                                          eighttap[frac], step,
-                                          eighttap[frac], step,
-                                          Width(), Height()));
+      ASM_REGISTER_STATE_CHECK(UUT_->shv8_(in, kInputStride, out, kOutputStride,
+                                           eighttap[frac], step,
+                                           eighttap[frac], step,
+                                           Width(), Height()));

      CheckGuardBlocks();

@@ -1054,7 +968,7 @@ void wrap_convolve8_horiz_sse2_8(const uint8_t *src, ptrdiff_t src_stride,
                                 const int16_t *filter_y,
                                 int filter_y_stride,
                                 int w, int h) {
-  vp9_highbd_convolve8_horiz_sse2(src, src_stride, dst, dst_stride, filter_x,
+  vpx_highbd_convolve8_horiz_sse2(src, src_stride, dst, dst_stride, filter_x,
                                  filter_x_stride, filter_y, filter_y_stride,
                                  w, h, 8);
 }
@@ -1066,7 +980,7 @@ void wrap_convolve8_avg_horiz_sse2_8(const uint8_t *src, ptrdiff_t src_stride,
                                     const int16_t *filter_y,
                                     int filter_y_stride,
                                     int w, int h) {
-  vp9_highbd_convolve8_avg_horiz_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_horiz_sse2(src, src_stride, dst, dst_stride,
                                      filter_x, filter_x_stride,
                                      filter_y, filter_y_stride, w, h, 8);
 }
@@ -1078,7 +992,7 @@ void wrap_convolve8_vert_sse2_8(const uint8_t *src, ptrdiff_t src_stride,
                                const int16_t *filter_y,
                                int filter_y_stride,
                                int w, int h) {
-  vp9_highbd_convolve8_vert_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_vert_sse2(src, src_stride, dst, dst_stride,
                                 filter_x, filter_x_stride,
                                 filter_y, filter_y_stride, w, h, 8);
 }
@@ -1090,7 +1004,7 @@ void wrap_convolve8_avg_vert_sse2_8(const uint8_t *src, ptrdiff_t src_stride,
                                    const int16_t *filter_y,
                                    int filter_y_stride,
                                    int w, int h) {
-  vp9_highbd_convolve8_avg_vert_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_vert_sse2(src, src_stride, dst, dst_stride,
                                     filter_x, filter_x_stride,
                                     filter_y, filter_y_stride, w, h, 8);
 }
@@ -1102,7 +1016,7 @@ void wrap_convolve8_sse2_8(const uint8_t *src, ptrdiff_t src_stride,
                           const int16_t *filter_y,
                           int filter_y_stride,
                           int w, int h) {
-  vp9_highbd_convolve8_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_sse2(src, src_stride, dst, dst_stride,
                            filter_x, filter_x_stride,
                            filter_y, filter_y_stride, w, h, 8);
 }
@@ -1114,7 +1028,7 @@ void wrap_convolve8_avg_sse2_8(const uint8_t *src, ptrdiff_t src_stride,
                               const int16_t *filter_y,
                               int filter_y_stride,
                               int w, int h) {
-  vp9_highbd_convolve8_avg_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_sse2(src, src_stride, dst, dst_stride,
                                filter_x, filter_x_stride,
                                filter_y, filter_y_stride, w, h, 8);
 }
@@ -1126,7 +1040,7 @@ void wrap_convolve8_horiz_sse2_10(const uint8_t *src, ptrdiff_t src_stride,
                                  const int16_t *filter_y,
                                  int filter_y_stride,
                                  int w, int h) {
-  vp9_highbd_convolve8_horiz_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_horiz_sse2(src, src_stride, dst, dst_stride,
                                  filter_x, filter_x_stride,
                                  filter_y, filter_y_stride, w, h, 10);
 }
@@ -1138,7 +1052,7 @@ void wrap_convolve8_avg_horiz_sse2_10(const uint8_t *src, ptrdiff_t src_stride,
                                      const int16_t *filter_y,
                                      int filter_y_stride,
                                      int w, int h) {
-  vp9_highbd_convolve8_avg_horiz_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_horiz_sse2(src, src_stride, dst, dst_stride,
                                      filter_x, filter_x_stride,
                                      filter_y, filter_y_stride, w, h, 10);
 }
@@ -1150,7 +1064,7 @@ void wrap_convolve8_vert_sse2_10(const uint8_t *src, ptrdiff_t src_stride,
                                 const int16_t *filter_y,
                                 int filter_y_stride,
                                 int w, int h) {
-  vp9_highbd_convolve8_vert_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_vert_sse2(src, src_stride, dst, dst_stride,
                                 filter_x, filter_x_stride,
                                 filter_y, filter_y_stride, w, h, 10);
 }
@@ -1162,7 +1076,7 @@ void wrap_convolve8_avg_vert_sse2_10(const uint8_t *src, ptrdiff_t src_stride,
                                     const int16_t *filter_y,
                                     int filter_y_stride,
                                     int w, int h) {
-  vp9_highbd_convolve8_avg_vert_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_vert_sse2(src, src_stride, dst, dst_stride,
                                     filter_x, filter_x_stride,
                                     filter_y, filter_y_stride, w, h, 10);
 }
@@ -1174,7 +1088,7 @@ void wrap_convolve8_sse2_10(const uint8_t *src, ptrdiff_t src_stride,
                            const int16_t *filter_y,
                            int filter_y_stride,
                            int w, int h) {
-  vp9_highbd_convolve8_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_sse2(src, src_stride, dst, dst_stride,
                            filter_x, filter_x_stride,
                            filter_y, filter_y_stride, w, h, 10);
 }
@@ -1186,7 +1100,7 @@ void wrap_convolve8_avg_sse2_10(const uint8_t *src, ptrdiff_t src_stride,
                                const int16_t *filter_y,
                                int filter_y_stride,
                                int w, int h) {
-  vp9_highbd_convolve8_avg_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_sse2(src, src_stride, dst, dst_stride,
                                filter_x, filter_x_stride,
                                filter_y, filter_y_stride, w, h, 10);
 }
@@ -1198,7 +1112,7 @@ void wrap_convolve8_horiz_sse2_12(const uint8_t *src, ptrdiff_t src_stride,
                                  const int16_t *filter_y,
                                  int filter_y_stride,
                                  int w, int h) {
-  vp9_highbd_convolve8_horiz_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_horiz_sse2(src, src_stride, dst, dst_stride,
                                  filter_x, filter_x_stride,
                                  filter_y, filter_y_stride, w, h, 12);
 }
@@ -1210,7 +1124,7 @@ void wrap_convolve8_avg_horiz_sse2_12(const uint8_t *src, ptrdiff_t src_stride,
                                      const int16_t *filter_y,
                                      int filter_y_stride,
                                      int w, int h) {
-  vp9_highbd_convolve8_avg_horiz_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_horiz_sse2(src, src_stride, dst, dst_stride,
                                      filter_x, filter_x_stride,
                                      filter_y, filter_y_stride, w, h, 12);
 }
@@ -1222,7 +1136,7 @@ void wrap_convolve8_vert_sse2_12(const uint8_t *src, ptrdiff_t src_stride,
                                 const int16_t *filter_y,
                                 int filter_y_stride,
                                 int w, int h) {
-  vp9_highbd_convolve8_vert_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_vert_sse2(src, src_stride, dst, dst_stride,
                                 filter_x, filter_x_stride,
                                 filter_y, filter_y_stride, w, h, 12);
 }
@@ -1234,7 +1148,7 @@ void wrap_convolve8_avg_vert_sse2_12(const uint8_t *src, ptrdiff_t src_stride,
                                     const int16_t *filter_y,
                                     int filter_y_stride,
                                     int w, int h) {
-  vp9_highbd_convolve8_avg_vert_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_vert_sse2(src, src_stride, dst, dst_stride,
                                     filter_x, filter_x_stride,
                                     filter_y, filter_y_stride, w, h, 12);
 }
@@ -1246,7 +1160,7 @@ void wrap_convolve8_sse2_12(const uint8_t *src, ptrdiff_t src_stride,
                            const int16_t *filter_y,
                            int filter_y_stride,
                            int w, int h) {
-  vp9_highbd_convolve8_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_sse2(src, src_stride, dst, dst_stride,
                            filter_x, filter_x_stride,
                            filter_y, filter_y_stride, w, h, 12);
 }
@@ -1258,7 +1172,7 @@ void wrap_convolve8_avg_sse2_12(const uint8_t *src, ptrdiff_t src_stride,
                                const int16_t *filter_y,
                                int filter_y_stride,
                                int w, int h) {
-  vp9_highbd_convolve8_avg_sse2(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_sse2(src, src_stride, dst, dst_stride,
                                filter_x, filter_x_stride,
                                filter_y, filter_y_stride, w, h, 12);
 }
@@ -1271,7 +1185,7 @@ void wrap_convolve_copy_c_8(const uint8_t *src, ptrdiff_t src_stride,
                            const int16_t *filter_y,
                            int filter_y_stride,
                            int w, int h) {
-  vp9_highbd_convolve_copy_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve_copy_c(src, src_stride, dst, dst_stride,
                             filter_x, filter_x_stride,
                             filter_y, filter_y_stride, w, h, 8);
 }
@@ -1283,7 +1197,7 @@ void wrap_convolve_avg_c_8(const uint8_t *src, ptrdiff_t src_stride,
                           const int16_t *filter_y,
                           int filter_y_stride,
                           int w, int h) {
-  vp9_highbd_convolve_avg_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve_avg_c(src, src_stride, dst, dst_stride,
                            filter_x, filter_x_stride,
                            filter_y, filter_y_stride, w, h, 8);
 }
@@ -1295,7 +1209,7 @@ void wrap_convolve8_horiz_c_8(const uint8_t *src, ptrdiff_t src_stride,
                              const int16_t *filter_y,
                              int filter_y_stride,
                              int w, int h) {
-  vp9_highbd_convolve8_horiz_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_horiz_c(src, src_stride, dst, dst_stride,
                               filter_x, filter_x_stride,
                               filter_y, filter_y_stride, w, h, 8);
 }
@@ -1307,7 +1221,7 @@ void wrap_convolve8_avg_horiz_c_8(const uint8_t *src, ptrdiff_t src_stride,
                                  const int16_t *filter_y,
                                  int filter_y_stride,
                                  int w, int h) {
-  vp9_highbd_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,
                                   filter_x, filter_x_stride,
                                   filter_y, filter_y_stride, w, h, 8);
 }
@@ -1319,7 +1233,7 @@ void wrap_convolve8_vert_c_8(const uint8_t *src, ptrdiff_t src_stride,
                             const int16_t *filter_y,
                             int filter_y_stride,
                             int w, int h) {
-  vp9_highbd_convolve8_vert_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_vert_c(src, src_stride, dst, dst_stride,
                              filter_x, filter_x_stride,
                              filter_y, filter_y_stride, w, h, 8);
 }
@@ -1331,7 +1245,7 @@ void wrap_convolve8_avg_vert_c_8(const uint8_t *src, ptrdiff_t src_stride,
                                 const int16_t *filter_y,
                                 int filter_y_stride,
                                 int w, int h) {
-  vp9_highbd_convolve8_avg_vert_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_vert_c(src, src_stride, dst, dst_stride,
                                  filter_x, filter_x_stride,
                                  filter_y, filter_y_stride, w, h, 8);
 }
@@ -1343,7 +1257,7 @@ void wrap_convolve8_c_8(const uint8_t *src, ptrdiff_t src_stride,
                        const int16_t *filter_y,
                        int filter_y_stride,
                        int w, int h) {
-  vp9_highbd_convolve8_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_c(src, src_stride, dst, dst_stride,
                         filter_x, filter_x_stride,
                         filter_y, filter_y_stride, w, h, 8);
 }
@@ -1355,7 +1269,7 @@ void wrap_convolve8_avg_c_8(const uint8_t *src, ptrdiff_t src_stride,
                            const int16_t *filter_y,
                            int filter_y_stride,
                            int w, int h) {
-  vp9_highbd_convolve8_avg_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_c(src, src_stride, dst, dst_stride,
                             filter_x, filter_x_stride,
                             filter_y, filter_y_stride, w, h, 8);
 }
@@ -1367,7 +1281,7 @@ void wrap_convolve_copy_c_10(const uint8_t *src, ptrdiff_t src_stride,
                             const int16_t *filter_y,
                             int filter_y_stride,
                             int w, int h) {
-  vp9_highbd_convolve_copy_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve_copy_c(src, src_stride, dst, dst_stride,
                             filter_x, filter_x_stride,
                             filter_y, filter_y_stride, w, h, 10);
 }
@@ -1379,7 +1293,7 @@ void wrap_convolve_avg_c_10(const uint8_t *src, ptrdiff_t src_stride,
                            const int16_t *filter_y,
                            int filter_y_stride,
                            int w, int h) {
-  vp9_highbd_convolve_avg_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve_avg_c(src, src_stride, dst, dst_stride,
                            filter_x, filter_x_stride,
                            filter_y, filter_y_stride, w, h, 10);
 }
@@ -1391,7 +1305,7 @@ void wrap_convolve8_horiz_c_10(const uint8_t *src, ptrdiff_t src_stride,
                               const int16_t *filter_y,
                               int filter_y_stride,
                               int w, int h) {
-  vp9_highbd_convolve8_horiz_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_horiz_c(src, src_stride, dst, dst_stride,
                               filter_x, filter_x_stride,
                               filter_y, filter_y_stride, w, h, 10);
 }
@@ -1403,7 +1317,7 @@ void wrap_convolve8_avg_horiz_c_10(const uint8_t *src, ptrdiff_t src_stride,
                                   const int16_t *filter_y,
                                   int filter_y_stride,
                                   int w, int h) {
-  vp9_highbd_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,
                                   filter_x, filter_x_stride,
                                   filter_y, filter_y_stride, w, h, 10);
 }
@@ -1415,7 +1329,7 @@ void wrap_convolve8_vert_c_10(const uint8_t *src, ptrdiff_t src_stride,
                              const int16_t *filter_y,
                              int filter_y_stride,
                              int w, int h) {
-  vp9_highbd_convolve8_vert_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_vert_c(src, src_stride, dst, dst_stride,
                              filter_x, filter_x_stride,
                              filter_y, filter_y_stride, w, h, 10);
 }
@@ -1427,7 +1341,7 @@ void wrap_convolve8_avg_vert_c_10(const uint8_t *src, ptrdiff_t src_stride,
                                  const int16_t *filter_y,
                                  int filter_y_stride,
                                  int w, int h) {
-  vp9_highbd_convolve8_avg_vert_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_vert_c(src, src_stride, dst, dst_stride,
                                  filter_x, filter_x_stride,
                                  filter_y, filter_y_stride, w, h, 10);
 }
@@ -1439,7 +1353,7 @@ void wrap_convolve8_c_10(const uint8_t *src, ptrdiff_t src_stride,
                         const int16_t *filter_y,
                         int filter_y_stride,
                         int w, int h) {
-  vp9_highbd_convolve8_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_c(src, src_stride, dst, dst_stride,
                         filter_x, filter_x_stride,
                         filter_y, filter_y_stride, w, h, 10);
 }
@@ -1451,7 +1365,7 @@ void wrap_convolve8_avg_c_10(const uint8_t *src, ptrdiff_t src_stride,
                             const int16_t *filter_y,
                             int filter_y_stride,
                             int w, int h) {
-  vp9_highbd_convolve8_avg_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_c(src, src_stride, dst, dst_stride,
                             filter_x, filter_x_stride,
                             filter_y, filter_y_stride, w, h, 10);
 }
@@ -1463,7 +1377,7 @@ void wrap_convolve_copy_c_12(const uint8_t *src, ptrdiff_t src_stride,
                             const int16_t *filter_y,
                             int filter_y_stride,
                             int w, int h) {
-  vp9_highbd_convolve_copy_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve_copy_c(src, src_stride, dst, dst_stride,
                             filter_x, filter_x_stride,
                             filter_y, filter_y_stride, w, h, 12);
 }
@@ -1475,7 +1389,7 @@ void wrap_convolve_avg_c_12(const uint8_t *src, ptrdiff_t src_stride,
                            const int16_t *filter_y,
                            int filter_y_stride,
                            int w, int h) {
-  vp9_highbd_convolve_avg_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve_avg_c(src, src_stride, dst, dst_stride,
                            filter_x, filter_x_stride,
                            filter_y, filter_y_stride, w, h, 12);
 }
@@ -1487,7 +1401,7 @@ void wrap_convolve8_horiz_c_12(const uint8_t *src, ptrdiff_t src_stride,
                               const int16_t *filter_y,
                               int filter_y_stride,
                               int w, int h) {
-  vp9_highbd_convolve8_horiz_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_horiz_c(src, src_stride, dst, dst_stride,
                               filter_x, filter_x_stride,
                               filter_y, filter_y_stride, w, h, 12);
 }
@@ -1499,7 +1413,7 @@ void wrap_convolve8_avg_horiz_c_12(const uint8_t *src, ptrdiff_t src_stride,
                                   const int16_t *filter_y,
                                   int filter_y_stride,
                                   int w, int h) {
-  vp9_highbd_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,
                                   filter_x, filter_x_stride,
                                   filter_y, filter_y_stride, w, h, 12);
 }
@@ -1511,7 +1425,7 @@ void wrap_convolve8_vert_c_12(const uint8_t *src, ptrdiff_t src_stride,
                              const int16_t *filter_y,
                              int filter_y_stride,
                              int w, int h) {
-  vp9_highbd_convolve8_vert_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_vert_c(src, src_stride, dst, dst_stride,
                              filter_x, filter_x_stride,
                              filter_y, filter_y_stride, w, h, 12);
 }
@@ -1523,7 +1437,7 @@ void wrap_convolve8_avg_vert_c_12(const uint8_t *src, ptrdiff_t src_stride,
                                  const int16_t *filter_y,
                                  int filter_y_stride,
                                  int w, int h) {
-  vp9_highbd_convolve8_avg_vert_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_vert_c(src, src_stride, dst, dst_stride,
                                  filter_x, filter_x_stride,
                                  filter_y, filter_y_stride, w, h, 12);
 }
@@ -1535,7 +1449,7 @@ void wrap_convolve8_c_12(const uint8_t *src, ptrdiff_t src_stride,
                         const int16_t *filter_y,
                         int filter_y_stride,
                         int w, int h) {
-  vp9_highbd_convolve8_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_c(src, src_stride, dst, dst_stride,
                         filter_x, filter_x_stride,
                         filter_y, filter_y_stride, w, h, 12);
 }
@@ -1547,7 +1461,7 @@ void wrap_convolve8_avg_c_12(const uint8_t *src, ptrdiff_t src_stride,
                             const int16_t *filter_y,
                             int filter_y_stride,
                             int w, int h) {
-  vp9_highbd_convolve8_avg_c(src, src_stride, dst, dst_stride,
+  vpx_highbd_convolve8_avg_c(src, src_stride, dst, dst_stride,
                             filter_x, filter_x_stride,
                             filter_y, filter_y_stride, w, h, 12);
 }
@@ -1556,6 +1470,9 @@ const ConvolveFunctions convolve8_c(
    wrap_convolve_copy_c_8, wrap_convolve_avg_c_8,
    wrap_convolve8_horiz_c_8, wrap_convolve8_avg_horiz_c_8,
    wrap_convolve8_vert_c_8, wrap_convolve8_avg_vert_c_8,
+    wrap_convolve8_c_8, wrap_convolve8_avg_c_8,
+    wrap_convolve8_horiz_c_8, wrap_convolve8_avg_horiz_c_8,
+    wrap_convolve8_vert_c_8, wrap_convolve8_avg_vert_c_8,
    wrap_convolve8_c_8, wrap_convolve8_avg_c_8, 8);
 INSTANTIATE_TEST_CASE_P(C_8, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_c),
@@ -1575,6 +1492,9 @@ const ConvolveFunctions convolve10_c(
    wrap_convolve_copy_c_10, wrap_convolve_avg_c_10,
    wrap_convolve8_horiz_c_10, wrap_convolve8_avg_horiz_c_10,
    wrap_convolve8_vert_c_10, wrap_convolve8_avg_vert_c_10,
+    wrap_convolve8_c_10, wrap_convolve8_avg_c_10,
+    wrap_convolve8_horiz_c_10, wrap_convolve8_avg_horiz_c_10,
+    wrap_convolve8_vert_c_10, wrap_convolve8_avg_vert_c_10,
    wrap_convolve8_c_10, wrap_convolve8_avg_c_10, 10);
 INSTANTIATE_TEST_CASE_P(C_10, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve10_c),
@@ -1594,6 +1514,9 @@ const ConvolveFunctions convolve12_c(
    wrap_convolve_copy_c_12, wrap_convolve_avg_c_12,
    wrap_convolve8_horiz_c_12, wrap_convolve8_avg_horiz_c_12,
    wrap_convolve8_vert_c_12, wrap_convolve8_avg_vert_c_12,
+    wrap_convolve8_c_12, wrap_convolve8_avg_c_12,
+    wrap_convolve8_horiz_c_12, wrap_convolve8_avg_horiz_c_12,
+    wrap_convolve8_vert_c_12, wrap_convolve8_avg_vert_c_12,
    wrap_convolve8_c_12, wrap_convolve8_avg_c_12, 12);
 INSTANTIATE_TEST_CASE_P(C_12, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve12_c),
@@ -1613,10 +1536,13 @@ INSTANTIATE_TEST_CASE_P(C_12, ConvolveTest, ::testing::Values(
 #else

 const ConvolveFunctions convolve8_c(
-    vp9_convolve_copy_c, vp9_convolve_avg_c,
-    vp9_convolve8_horiz_c, vp9_convolve8_avg_horiz_c,
-    vp9_convolve8_vert_c, vp9_convolve8_avg_vert_c,
-    vp9_convolve8_c, vp9_convolve8_avg_c, 0);
+    vpx_convolve_copy_c, vpx_convolve_avg_c,
+    vpx_convolve8_horiz_c, vpx_convolve8_avg_horiz_c,
+    vpx_convolve8_vert_c, vpx_convolve8_avg_vert_c,
+    vpx_convolve8_c, vpx_convolve8_avg_c,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);

 INSTANTIATE_TEST_CASE_P(C, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_c),
@@ -1640,16 +1566,25 @@ const ConvolveFunctions convolve8_sse2(
    wrap_convolve_copy_c_8, wrap_convolve_avg_c_8,
    wrap_convolve8_horiz_sse2_8, wrap_convolve8_avg_horiz_sse2_8,
    wrap_convolve8_vert_sse2_8, wrap_convolve8_avg_vert_sse2_8,
+    wrap_convolve8_sse2_8, wrap_convolve8_avg_sse2_8,
+    wrap_convolve8_horiz_sse2_8, wrap_convolve8_avg_horiz_sse2_8,
+    wrap_convolve8_vert_sse2_8, wrap_convolve8_avg_vert_sse2_8,
    wrap_convolve8_sse2_8, wrap_convolve8_avg_sse2_8, 8);
 const ConvolveFunctions convolve10_sse2(
    wrap_convolve_copy_c_10, wrap_convolve_avg_c_10,
    wrap_convolve8_horiz_sse2_10, wrap_convolve8_avg_horiz_sse2_10,
    wrap_convolve8_vert_sse2_10, wrap_convolve8_avg_vert_sse2_10,
+    wrap_convolve8_sse2_10, wrap_convolve8_avg_sse2_10,
+    wrap_convolve8_horiz_sse2_10, wrap_convolve8_avg_horiz_sse2_10,
+    wrap_convolve8_vert_sse2_10, wrap_convolve8_avg_vert_sse2_10,
    wrap_convolve8_sse2_10, wrap_convolve8_avg_sse2_10, 10);
 const ConvolveFunctions convolve12_sse2(
    wrap_convolve_copy_c_12, wrap_convolve_avg_c_12,
    wrap_convolve8_horiz_sse2_12, wrap_convolve8_avg_horiz_sse2_12,
    wrap_convolve8_vert_sse2_12, wrap_convolve8_avg_vert_sse2_12,
+    wrap_convolve8_sse2_12, wrap_convolve8_avg_sse2_12,
+    wrap_convolve8_horiz_sse2_12, wrap_convolve8_avg_horiz_sse2_12,
+    wrap_convolve8_vert_sse2_12, wrap_convolve8_avg_vert_sse2_12,
    wrap_convolve8_sse2_12, wrap_convolve8_avg_sse2_12, 12);
 INSTANTIATE_TEST_CASE_P(SSE2, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_sse2),
@@ -1693,10 +1628,17 @@ INSTANTIATE_TEST_CASE_P(SSE2, ConvolveTest, ::testing::Values(
    make_tuple(64, 64, &convolve12_sse2)));
 #else
 const ConvolveFunctions convolve8_sse2(
-    vp9_convolve_copy_sse2, vp9_convolve_avg_sse2,
-    vp9_convolve8_horiz_sse2, vp9_convolve8_avg_horiz_sse2,
-    vp9_convolve8_vert_sse2, vp9_convolve8_avg_vert_sse2,
-    vp9_convolve8_sse2, vp9_convolve8_avg_sse2, 0);
+#if CONFIG_USE_X86INC
+    vpx_convolve_copy_sse2, vpx_convolve_avg_sse2,
+#else
+    vpx_convolve_copy_c, vpx_convolve_avg_c,
+#endif  // CONFIG_USE_X86INC
+    vpx_convolve8_horiz_sse2, vpx_convolve8_avg_horiz_sse2,
+    vpx_convolve8_vert_sse2, vpx_convolve8_avg_vert_sse2,
+    vpx_convolve8_sse2, vpx_convolve8_avg_sse2,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);

 INSTANTIATE_TEST_CASE_P(SSE2, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_sse2),
@@ -1717,10 +1659,13 @@ INSTANTIATE_TEST_CASE_P(SSE2, ConvolveTest, ::testing::Values(

 #if HAVE_SSSE3
 const ConvolveFunctions convolve8_ssse3(
-    vp9_convolve_copy_c, vp9_convolve_avg_c,
-    vp9_convolve8_horiz_ssse3, vp9_convolve8_avg_horiz_ssse3,
-    vp9_convolve8_vert_ssse3, vp9_convolve8_avg_vert_ssse3,
-    vp9_convolve8_ssse3, vp9_convolve8_avg_ssse3, 0);
+    vpx_convolve_copy_c, vpx_convolve_avg_c,
+    vpx_convolve8_horiz_ssse3, vpx_convolve8_avg_horiz_ssse3,
+    vpx_convolve8_vert_ssse3, vpx_convolve8_avg_vert_ssse3,
+    vpx_convolve8_ssse3, vpx_convolve8_avg_ssse3,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);

 INSTANTIATE_TEST_CASE_P(SSSE3, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_ssse3),
@@ -1740,10 +1685,13 @@ INSTANTIATE_TEST_CASE_P(SSSE3, ConvolveTest, ::testing::Values(

 #if HAVE_AVX2 && HAVE_SSSE3
 const ConvolveFunctions convolve8_avx2(
-    vp9_convolve_copy_c, vp9_convolve_avg_c,
-    vp9_convolve8_horiz_avx2, vp9_convolve8_avg_horiz_ssse3,
-    vp9_convolve8_vert_avx2, vp9_convolve8_avg_vert_ssse3,
-    vp9_convolve8_avx2, vp9_convolve8_avg_ssse3, 0);
+    vpx_convolve_copy_c, vpx_convolve_avg_c,
+    vpx_convolve8_horiz_avx2, vpx_convolve8_avg_horiz_ssse3,
+    vpx_convolve8_vert_avx2, vpx_convolve8_avg_vert_ssse3,
+    vpx_convolve8_avx2, vpx_convolve8_avg_ssse3,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);

 INSTANTIATE_TEST_CASE_P(AVX2, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_avx2),
@@ -1764,16 +1712,22 @@ INSTANTIATE_TEST_CASE_P(AVX2, ConvolveTest, ::testing::Values(
 #if HAVE_NEON
 #if HAVE_NEON_ASM
 const ConvolveFunctions convolve8_neon(
-    vp9_convolve_copy_neon, vp9_convolve_avg_neon,
-    vp9_convolve8_horiz_neon, vp9_convolve8_avg_horiz_neon,
-    vp9_convolve8_vert_neon, vp9_convolve8_avg_vert_neon,
-    vp9_convolve8_neon, vp9_convolve8_avg_neon, 0);
+    vpx_convolve_copy_neon, vpx_convolve_avg_neon,
+    vpx_convolve8_horiz_neon, vpx_convolve8_avg_horiz_neon,
+    vpx_convolve8_vert_neon, vpx_convolve8_avg_vert_neon,
+    vpx_convolve8_neon, vpx_convolve8_avg_neon,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);
 #else  // HAVE_NEON
 const ConvolveFunctions convolve8_neon(
-    vp9_convolve_copy_neon, vp9_convolve_avg_neon,
-    vp9_convolve8_horiz_neon, vp9_convolve8_avg_horiz_neon,
-    vp9_convolve8_vert_neon, vp9_convolve8_avg_vert_neon,
-    vp9_convolve8_neon, vp9_convolve8_avg_neon, 0);
+    vpx_convolve_copy_neon, vpx_convolve_avg_neon,
+    vpx_convolve8_horiz_neon, vpx_convolve8_avg_horiz_neon,
+    vpx_convolve8_vert_neon, vpx_convolve8_avg_vert_neon,
+    vpx_convolve8_neon, vpx_convolve8_avg_neon,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);
 #endif  // HAVE_NEON_ASM

 INSTANTIATE_TEST_CASE_P(NEON, ConvolveTest, ::testing::Values(
@@ -1794,10 +1748,13 @@ INSTANTIATE_TEST_CASE_P(NEON, ConvolveTest, ::testing::Values(

 #if HAVE_DSPR2
 const ConvolveFunctions convolve8_dspr2(
-    vp9_convolve_copy_dspr2, vp9_convolve_avg_dspr2,
-    vp9_convolve8_horiz_dspr2, vp9_convolve8_avg_horiz_dspr2,
-    vp9_convolve8_vert_dspr2, vp9_convolve8_avg_vert_dspr2,
-    vp9_convolve8_dspr2, vp9_convolve8_avg_dspr2, 0);
+    vpx_convolve_copy_dspr2, vpx_convolve_avg_dspr2,
+    vpx_convolve8_horiz_dspr2, vpx_convolve8_avg_horiz_dspr2,
+    vpx_convolve8_vert_dspr2, vpx_convolve8_avg_vert_dspr2,
+    vpx_convolve8_dspr2, vpx_convolve8_avg_dspr2,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);

 INSTANTIATE_TEST_CASE_P(DSPR2, ConvolveTest, ::testing::Values(
    make_tuple(4, 4, &convolve8_dspr2),
@@ -1814,4 +1771,30 @@ INSTANTIATE_TEST_CASE_P(DSPR2, ConvolveTest, ::testing::Values(
    make_tuple(32, 64, &convolve8_dspr2),
    make_tuple(64, 64, &convolve8_dspr2)));
 #endif
+
+#if HAVE_MSA
+const ConvolveFunctions convolve8_msa(
+    vpx_convolve_copy_msa, vpx_convolve_avg_msa,
+    vpx_convolve8_horiz_msa, vpx_convolve8_avg_horiz_msa,
+    vpx_convolve8_vert_msa, vpx_convolve8_avg_vert_msa,
+    vpx_convolve8_msa, vpx_convolve8_avg_msa,
+    vpx_scaled_horiz_c, vpx_scaled_avg_horiz_c,
+    vpx_scaled_vert_c, vpx_scaled_avg_vert_c,
+    vpx_scaled_2d_c, vpx_scaled_avg_2d_c, 0);
+
+INSTANTIATE_TEST_CASE_P(MSA, ConvolveTest, ::testing::Values(
+    make_tuple(4, 4, &convolve8_msa),
+    make_tuple(8, 4, &convolve8_msa),
+    make_tuple(4, 8, &convolve8_msa),
+    make_tuple(8, 8, &convolve8_msa),
+    make_tuple(16, 8, &convolve8_msa),
+    make_tuple(8, 16, &convolve8_msa),
+    make_tuple(16, 16, &convolve8_msa),
+    make_tuple(32, 16, &convolve8_msa),
+    make_tuple(16, 32, &convolve8_msa),
+    make_tuple(32, 32, &convolve8_msa),
+    make_tuple(64, 32, &convolve8_msa),
+    make_tuple(32, 64, &convolve8_msa),
+    make_tuple(64, 64, &convolve8_msa)));
+#endif  // HAVE_MSA
 }  // namespace
--- a/test/cpu_speed_test.cc
+++ b/test/cpu_speed_test.cc
@@ -140,4 +140,9 @@ VP9_INSTANTIATE_TEST_CASE(
    ::testing::Values(::libvpx_test::kTwoPassGood, ::libvpx_test::kOnePassGood,
                      ::libvpx_test::kRealTime),
    ::testing::Range(0, 9));
+
+VP10_INSTANTIATE_TEST_CASE(
+    CpuSpeedTest,
+    ::testing::Values(::libvpx_test::kTwoPassGood, ::libvpx_test::kOnePassGood),
+    ::testing::Range(0, 3));
 }  // namespace
--- a/test/datarate_test.cc
+++ b/test/datarate_test.cc
@@ -14,6 +14,7 @@
 #include "test/i420_video_source.h"
 #include "test/util.h"
 #include "test/y4m_video_source.h"
+#include "vpx/vpx_codec.h"

 namespace {

@@ -371,9 +372,7 @@ class DatarateTestVP9Large : public ::libvpx_test::EncoderTest,
        encoder->Control(VP9E_SET_SVC, 1);
      }
      vpx_svc_layer_id_t layer_id;
-#if VPX_ENCODER_ABI_VERSION > (4 + VPX_CODEC_ABI_VERSION)
      layer_id.spatial_layer_id = 0;
-#endif
      frame_flags_ = SetFrameFlags(video->frame(), cfg_.ts_number_layers);
      layer_id.temporal_layer_id = SetLayerId(video->frame(),
                                              cfg_.ts_number_layers);
@@ -565,6 +564,8 @@ TEST_P(DatarateTestVP9Large, BasicRateTargeting2TemporalLayers) {
  cfg_.ts_rate_decimator[0] = 2;
  cfg_.ts_rate_decimator[1] = 1;

+  cfg_.temporal_layering_mode = VP9E_TEMPORAL_LAYERING_MODE_BYPASS;
+
  if (deadline_ == VPX_DL_REALTIME)
    cfg_.g_error_resilient = 1;

@@ -574,14 +575,14 @@ TEST_P(DatarateTestVP9Large, BasicRateTargeting2TemporalLayers) {
    cfg_.rc_target_bitrate = i;
    ResetModel();
    // 60-40 bitrate allocation for 2 temporal layers.
-    cfg_.ts_target_bitrate[0] = 60 * cfg_.rc_target_bitrate / 100;
-    cfg_.ts_target_bitrate[1] = cfg_.rc_target_bitrate;
+    cfg_.layer_target_bitrate[0] = 60 * cfg_.rc_target_bitrate / 100;
+    cfg_.layer_target_bitrate[1] = cfg_.rc_target_bitrate;
    ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
    for (int j = 0; j < static_cast<int>(cfg_.ts_number_layers); ++j) {
-      ASSERT_GE(effective_datarate_[j], cfg_.ts_target_bitrate[j] * 0.85)
+      ASSERT_GE(effective_datarate_[j], cfg_.layer_target_bitrate[j] * 0.85)
          << " The datarate for the file is lower than target by too much, "
              "for layer: " << j;
-      ASSERT_LE(effective_datarate_[j], cfg_.ts_target_bitrate[j] * 1.15)
+      ASSERT_LE(effective_datarate_[j], cfg_.layer_target_bitrate[j] * 1.15)
          << " The datarate for the file is greater than target by too much, "
              "for layer: " << j;
    }
@@ -606,25 +607,27 @@ TEST_P(DatarateTestVP9Large, BasicRateTargeting3TemporalLayers) {
  cfg_.ts_rate_decimator[1] = 2;
  cfg_.ts_rate_decimator[2] = 1;

+  cfg_.temporal_layering_mode = VP9E_TEMPORAL_LAYERING_MODE_BYPASS;
+
  ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
                                       30, 1, 0, 200);
  for (int i = 200; i <= 800; i += 200) {
    cfg_.rc_target_bitrate = i;
    ResetModel();
    // 40-20-40 bitrate allocation for 3 temporal layers.
-    cfg_.ts_target_bitrate[0] = 40 * cfg_.rc_target_bitrate / 100;
-    cfg_.ts_target_bitrate[1] = 60 * cfg_.rc_target_bitrate / 100;
-    cfg_.ts_target_bitrate[2] = cfg_.rc_target_bitrate;
+    cfg_.layer_target_bitrate[0] = 40 * cfg_.rc_target_bitrate / 100;
+    cfg_.layer_target_bitrate[1] = 60 * cfg_.rc_target_bitrate / 100;
+    cfg_.layer_target_bitrate[2] = cfg_.rc_target_bitrate;
    ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
    for (int j = 0; j < static_cast<int>(cfg_.ts_number_layers); ++j) {
      // TODO(yaowu): Work out more stable rc control strategy and
      //              Adjust the thresholds to be tighter than .75.
-      ASSERT_GE(effective_datarate_[j], cfg_.ts_target_bitrate[j] * 0.75)
+      ASSERT_GE(effective_datarate_[j], cfg_.layer_target_bitrate[j] * 0.75)
          << " The datarate for the file is lower than target by too much, "
              "for layer: " << j;
      // TODO(yaowu): Work out more stable rc control strategy and
      //              Adjust the thresholds to be tighter than 1.25.
-      ASSERT_LE(effective_datarate_[j], cfg_.ts_target_bitrate[j] * 1.25)
+      ASSERT_LE(effective_datarate_[j], cfg_.layer_target_bitrate[j] * 1.25)
          << " The datarate for the file is greater than target by too much, "
              "for layer: " << j;
    }
@@ -652,20 +655,22 @@ TEST_P(DatarateTestVP9Large, BasicRateTargeting3TemporalLayersFrameDropping) {
  cfg_.ts_rate_decimator[1] = 2;
  cfg_.ts_rate_decimator[2] = 1;

+  cfg_.temporal_layering_mode = VP9E_TEMPORAL_LAYERING_MODE_BYPASS;
+
  ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
                                       30, 1, 0, 200);
  cfg_.rc_target_bitrate = 200;
  ResetModel();
  // 40-20-40 bitrate allocation for 3 temporal layers.
-  cfg_.ts_target_bitrate[0] = 40 * cfg_.rc_target_bitrate / 100;
-  cfg_.ts_target_bitrate[1] = 60 * cfg_.rc_target_bitrate / 100;
-  cfg_.ts_target_bitrate[2] = cfg_.rc_target_bitrate;
+  cfg_.layer_target_bitrate[0] = 40 * cfg_.rc_target_bitrate / 100;
+  cfg_.layer_target_bitrate[1] = 60 * cfg_.rc_target_bitrate / 100;
+  cfg_.layer_target_bitrate[2] = cfg_.rc_target_bitrate;
  ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
  for (int j = 0; j < static_cast<int>(cfg_.ts_number_layers); ++j) {
-    ASSERT_GE(effective_datarate_[j], cfg_.ts_target_bitrate[j] * 0.85)
+    ASSERT_GE(effective_datarate_[j], cfg_.layer_target_bitrate[j] * 0.85)
        << " The datarate for the file is lower than target by too much, "
            "for layer: " << j;
-    ASSERT_LE(effective_datarate_[j], cfg_.ts_target_bitrate[j] * 1.15)
+    ASSERT_LE(effective_datarate_[j], cfg_.layer_target_bitrate[j] * 1.15)
        << " The datarate for the file is greater than target by too much, "
            "for layer: " << j;
    // Expect some frame drops in this test: for this 200 frames test,
@@ -737,9 +742,237 @@ TEST_P(DatarateTestVP9Large, DenoiserOffOn) {
 }
 #endif  // CONFIG_VP9_TEMPORAL_DENOISING

+class DatarateOnePassCbrSvc : public ::libvpx_test::EncoderTest,
+    public ::libvpx_test::CodecTestWith2Params<libvpx_test::TestMode, int> {
+ public:
+  DatarateOnePassCbrSvc() : EncoderTest(GET_PARAM(0)) {}
+  virtual ~DatarateOnePassCbrSvc() {}
+ protected:
+  virtual void SetUp() {
+    InitializeConfig();
+    SetMode(GET_PARAM(1));
+    speed_setting_ = GET_PARAM(2);
+    ResetModel();
+  }
+  virtual void ResetModel() {
+    last_pts_ = 0;
+    bits_in_buffer_model_ = cfg_.rc_target_bitrate * cfg_.rc_buf_initial_sz;
+    frame_number_ = 0;
+    first_drop_ = 0;
+    bits_total_ = 0;
+    duration_ = 0.0;
+    mismatch_psnr_ = 0.0;
+    mismatch_nframes_ = 0;
+  }
+  virtual void BeginPassHook(unsigned int /*pass*/) {
+  }
+  virtual void PreEncodeFrameHook(::libvpx_test::VideoSource *video,
+                                  ::libvpx_test::Encoder *encoder) {
+    if (video->frame() == 0) {
+      int i;
+      for (i = 0; i < 2; ++i) {
+        svc_params_.max_quantizers[i] = 63;
+        svc_params_.min_quantizers[i] = 0;
+      }
+      svc_params_.scaling_factor_num[0] = 144;
+      svc_params_.scaling_factor_den[0] = 288;
+      svc_params_.scaling_factor_num[1] = 288;
+      svc_params_.scaling_factor_den[1] = 288;
+      encoder->Control(VP9E_SET_SVC, 1);
+      encoder->Control(VP9E_SET_SVC_PARAMETERS, &svc_params_);
+      encoder->Control(VP8E_SET_CPUUSED, speed_setting_);
+      encoder->Control(VP9E_SET_TILE_COLUMNS, 0);
+      encoder->Control(VP8E_SET_MAX_INTRA_BITRATE_PCT, 300);
+      encoder->Control(VP9E_SET_TILE_COLUMNS, (cfg_.g_threads >> 1));
+    }
+    const vpx_rational_t tb = video->timebase();
+    timebase_ = static_cast<double>(tb.num) / tb.den;
+    duration_ = 0;
+  }
+  virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+    vpx_codec_pts_t duration = pkt->data.frame.pts - last_pts_;
+    if (last_pts_ == 0)
+      duration = 1;
+    bits_in_buffer_model_ += static_cast<int64_t>(
+        duration * timebase_ * cfg_.rc_target_bitrate * 1000);
+    const bool key_frame = (pkt->data.frame.flags & VPX_FRAME_IS_KEY)
+                         ? true: false;
+    if (!key_frame) {
+      ASSERT_GE(bits_in_buffer_model_, 0) << "Buffer Underrun at frame "
+          << pkt->data.frame.pts;
+    }
+    const size_t frame_size_in_bits = pkt->data.frame.sz * 8;
+    bits_in_buffer_model_ -= frame_size_in_bits;
+    bits_total_ += frame_size_in_bits;
+    if (!first_drop_ && duration > 1)
+      first_drop_ = last_pts_ + 1;
+    last_pts_ = pkt->data.frame.pts;
+    bits_in_last_frame_ = frame_size_in_bits;
+    ++frame_number_;
+  }
+  virtual void EndPassHook(void) {
+    if (bits_total_) {
+      const double file_size_in_kb = bits_total_ / 1000.;  // bits per kilobit
+      duration_ = (last_pts_ + 1) * timebase_;
+      effective_datarate_ = (bits_total_ - bits_in_last_frame_) / 1000.0
+          / (cfg_.rc_buf_initial_sz / 1000.0 + duration_);
+      file_datarate_ = file_size_in_kb / duration_;
+    }
+  }
+
+  virtual void MismatchHook(const vpx_image_t *img1,
+                            const vpx_image_t *img2) {
+    double mismatch_psnr = compute_psnr(img1, img2);
+    mismatch_psnr_ += mismatch_psnr;
+    ++mismatch_nframes_;
+  }
+
+  unsigned int GetMismatchFrames() {
+    return mismatch_nframes_;
+  }
+
+  vpx_codec_pts_t last_pts_;
+  int64_t bits_in_buffer_model_;
+  double timebase_;
+  int frame_number_;
+  vpx_codec_pts_t first_drop_;
+  int64_t bits_total_;
+  double duration_;
+  double file_datarate_;
+  double effective_datarate_;
+  size_t bits_in_last_frame_;
+  vpx_svc_extra_cfg_t svc_params_;
+  int speed_setting_;
+  double mismatch_psnr_;
+  int mismatch_nframes_;
+};
+static void assign_layer_bitrates(vpx_codec_enc_cfg_t *const enc_cfg,
+    const vpx_svc_extra_cfg_t *svc_params,
+    int spatial_layers,
+    int temporal_layers,
+    int temporal_layering_mode,
+    unsigned int total_rate) {
+  int sl, spatial_layer_target;
+  float total = 0;
+  float alloc_ratio[VPX_MAX_LAYERS] = {0};
+  for (sl = 0; sl < spatial_layers; ++sl) {
+    if (svc_params->scaling_factor_den[sl] > 0) {
+      alloc_ratio[sl] = (float)(svc_params->scaling_factor_num[sl] *
+          1.0 / svc_params->scaling_factor_den[sl]);
+      total += alloc_ratio[sl];
+    }
+  }
+  for (sl = 0; sl < spatial_layers; ++sl) {
+    enc_cfg->ss_target_bitrate[sl] = spatial_layer_target =
+        (unsigned int)(enc_cfg->rc_target_bitrate *
+            alloc_ratio[sl] / total);
+    const int index = sl * temporal_layers;
+    if (temporal_layering_mode == 3) {
+      enc_cfg->layer_target_bitrate[index] =
+          spatial_layer_target >> 1;
+      enc_cfg->layer_target_bitrate[index + 1] =
+          (spatial_layer_target >> 1) + (spatial_layer_target >> 2);
+      enc_cfg->layer_target_bitrate[index + 2] =
+          spatial_layer_target;
+    } else if (temporal_layering_mode == 2) {
+      enc_cfg->layer_target_bitrate[index] =
+          spatial_layer_target * 2 / 3;
+      enc_cfg->layer_target_bitrate[index + 1] =
+          spatial_layer_target;
+    }
+  }
+}
+
+// Check basic rate targeting for 1 pass CBR SVC: 2 spatial layers and
+// 3 temporal layers. Run CIF clip with 1 thread.
+TEST_P(DatarateOnePassCbrSvc, OnePassCbrSvc) {
+  cfg_.rc_buf_initial_sz = 500;
+  cfg_.rc_buf_optimal_sz = 500;
+  cfg_.rc_buf_sz = 1000;
+  cfg_.rc_min_quantizer = 0;
+  cfg_.rc_max_quantizer = 63;
+  cfg_.rc_end_usage = VPX_CBR;
+  cfg_.g_lag_in_frames = 0;
+  cfg_.ss_number_layers = 2;
+  cfg_.ts_number_layers = 3;
+  cfg_.ts_rate_decimator[0] = 4;
+  cfg_.ts_rate_decimator[1] = 2;
+  cfg_.ts_rate_decimator[2] = 1;
+  cfg_.g_error_resilient = 1;
+  cfg_.g_threads = 1;
+  cfg_.temporal_layering_mode = 3;
+  svc_params_.scaling_factor_num[0] = 144;
+  svc_params_.scaling_factor_den[0] = 288;
+  svc_params_.scaling_factor_num[1] = 288;
+  svc_params_.scaling_factor_den[1] = 288;
+  // TODO(wonkap/marpan): No frame drop for now, we need to implement correct
+  // frame dropping for SVC.
+  cfg_.rc_dropframe_thresh = 0;
+  ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
+                                       30, 1, 0, 200);
+  // TODO(wonkap/marpan): Check that effective_datarate for each layer hits the
+  // layer target_bitrate. Also check if test can pass at lower bitrate (~200k).
+  for (int i = 400; i <= 800; i += 200) {
+    cfg_.rc_target_bitrate = i;
+    ResetModel();
+    assign_layer_bitrates(&cfg_, &svc_params_, cfg_.ss_number_layers,
+        cfg_.ts_number_layers, cfg_.temporal_layering_mode,
+        cfg_.rc_target_bitrate);
+    ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+    ASSERT_GE(cfg_.rc_target_bitrate, effective_datarate_ * 0.85)
+            << " The datarate for the file exceeds the target by too much!";
+    ASSERT_LE(cfg_.rc_target_bitrate, file_datarate_ * 1.15)
+        << " The datarate for the file is lower than the target by too much!";
+    EXPECT_EQ(GetMismatchFrames(), (unsigned int) 0);
+  }
+}
+
+// Check basic rate targeting for 1 pass CBR SVC: 2 spatial layers and
+// 3 temporal layers. Run HD clip with 4 threads.
+TEST_P(DatarateOnePassCbrSvc, OnePassCbrSvc4threads) {
+  cfg_.rc_buf_initial_sz = 500;
+  cfg_.rc_buf_optimal_sz = 500;
+  cfg_.rc_buf_sz = 1000;
+  cfg_.rc_min_quantizer = 0;
+  cfg_.rc_max_quantizer = 63;
+  cfg_.rc_end_usage = VPX_CBR;
+  cfg_.g_lag_in_frames = 0;
+  cfg_.ss_number_layers = 2;
+  cfg_.ts_number_layers = 3;
+  cfg_.ts_rate_decimator[0] = 4;
+  cfg_.ts_rate_decimator[1] = 2;
+  cfg_.ts_rate_decimator[2] = 1;
+  cfg_.g_error_resilient = 1;
+  cfg_.g_threads = 4;
+  cfg_.temporal_layering_mode = 3;
+  svc_params_.scaling_factor_num[0] = 144;
+  svc_params_.scaling_factor_den[0] = 288;
+  svc_params_.scaling_factor_num[1] = 288;
+  svc_params_.scaling_factor_den[1] = 288;
+  // TODO(wonkap/marpan): No frame drop for now, we need to implement correct
+  // frame dropping for SVC.
+  cfg_.rc_dropframe_thresh = 0;
+  ::libvpx_test::I420VideoSource video("niklas_1280_720_30.y4m", 1280, 720,
+                                       30, 1, 0, 300);
+  cfg_.rc_target_bitrate = 800;
+  ResetModel();
+  assign_layer_bitrates(&cfg_, &svc_params_, cfg_.ss_number_layers,
+      cfg_.ts_number_layers, cfg_.temporal_layering_mode,
+      cfg_.rc_target_bitrate);
+  ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+  ASSERT_GE(cfg_.rc_target_bitrate, effective_datarate_ * 0.85)
+          << " The datarate for the file exceeds the target by too much!";
+  ASSERT_LE(cfg_.rc_target_bitrate, file_datarate_ * 1.15)
+      << " The datarate for the file is lower than the target by too much!";
+  EXPECT_EQ(GetMismatchFrames(), (unsigned int) 0);
+}
+
 VP8_INSTANTIATE_TEST_CASE(DatarateTestLarge, ALL_TEST_MODES);
 VP9_INSTANTIATE_TEST_CASE(DatarateTestVP9Large,
                          ::testing::Values(::libvpx_test::kOnePassGood,
                                            ::libvpx_test::kRealTime),
                          ::testing::Range(2, 7));
+VP9_INSTANTIATE_TEST_CASE(DatarateOnePassCbrSvc,
+                          ::testing::Values(::libvpx_test::kRealTime),
+                          ::testing::Range(5, 8));
 }  // namespace
--- a/test/dct16x16_test.cc
+++ b/test/dct16x16_test.cc
@@ -13,15 +13,18 @@
 #include <string.h>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
+#include "vp9/common/vp9_scan.h"
 #include "vpx/vpx_codec.h"
 #include "vpx/vpx_integer.h"
+#include "vpx_ports/mem.h"

 using libvpx_test::ACMRandom;

@@ -37,30 +40,6 @@ static int round(double x) {
 #endif

 const int kNumCoeffs = 256;
-const double PI = 3.1415926535898;
-void reference2_16x16_idct_2d(double *input, double *output) {
-  double x;
-  for (int l = 0; l < 16; ++l) {
-    for (int k = 0; k < 16; ++k) {
-      double s = 0;
-      for (int i = 0; i < 16; ++i) {
-        for (int j = 0; j < 16; ++j) {
-          x = cos(PI * j * (l + 0.5) / 16.0) *
-              cos(PI * i * (k + 0.5) / 16.0) *
-              input[i * 16 + j] / 256;
-          if (i != 0)
-            x *= sqrt(2.0);
-          if (j != 0)
-            x *= sqrt(2.0);
-          s += x;
-        }
-      }
-      output[k*16+l] = s;
-    }
-  }
-}
-
-
 const double C1 = 0.995184726672197;
 const double C2 = 0.98078528040323;
 const double C3 = 0.956940335732209;
@@ -269,12 +248,12 @@ typedef std::tr1::tuple<IdctFunc, IdctFunc, int, vpx_bit_depth_t>

 void fdct16x16_ref(const int16_t *in, tran_low_t *out, int stride,
                   int /*tx_type*/) {
-  vp9_fdct16x16_c(in, out, stride);
+  vpx_fdct16x16_c(in, out, stride);
 }

 void idct16x16_ref(const tran_low_t *in, uint8_t *dest, int stride,
                   int /*tx_type*/) {
-  vp9_idct16x16_256_add_c(in, dest, stride);
+  vpx_idct16x16_256_add_c(in, dest, stride);
 }

 void fht16x16_ref(const int16_t *in, tran_low_t *out, int stride,
@@ -289,11 +268,11 @@ void iht16x16_ref(const tran_low_t *in, uint8_t *dest, int stride,

 #if CONFIG_VP9_HIGHBITDEPTH
 void idct16x16_10(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_256_add_c(in, out, stride, 10);
+  vpx_highbd_idct16x16_256_add_c(in, out, stride, 10);
 }

 void idct16x16_12(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_256_add_c(in, out, stride, 12);
+  vpx_highbd_idct16x16_256_add_c(in, out, stride, 12);
 }

 void idct16x16_10_ref(const tran_low_t *in, uint8_t *out, int stride,
@@ -315,28 +294,28 @@ void iht16x16_12(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
 }

 void idct16x16_10_add_10_c(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_10_add_c(in, out, stride, 10);
+  vpx_highbd_idct16x16_10_add_c(in, out, stride, 10);
 }

 void idct16x16_10_add_12_c(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_10_add_c(in, out, stride, 12);
+  vpx_highbd_idct16x16_10_add_c(in, out, stride, 12);
 }

 #if HAVE_SSE2
 void idct16x16_256_add_10_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_256_add_sse2(in, out, stride, 10);
+  vpx_highbd_idct16x16_256_add_sse2(in, out, stride, 10);
 }

 void idct16x16_256_add_12_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_256_add_sse2(in, out, stride, 12);
+  vpx_highbd_idct16x16_256_add_sse2(in, out, stride, 12);
 }

 void idct16x16_10_add_10_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_10_add_sse2(in, out, stride, 10);
+  vpx_highbd_idct16x16_10_add_sse2(in, out, stride, 10);
 }

 void idct16x16_10_add_12_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct16x16_10_add_sse2(in, out, stride, 12);
+  vpx_highbd_idct16x16_10_add_sse2(in, out, stride, 12);
 }
 #endif  // HAVE_SSE2
 #endif  // CONFIG_VP9_HIGHBITDEPTH
@@ -356,13 +335,13 @@ class Trans16x16TestBase {
    int64_t total_error = 0;
    const int count_test_block = 10000;
    for (int i = 0; i < count_test_block; ++i) {
-      DECLARE_ALIGNED_ARRAY(16, int16_t, test_input_block, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_temp_block, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+      DECLARE_ALIGNED(16, int16_t, test_input_block[kNumCoeffs]);
+      DECLARE_ALIGNED(16, tran_low_t, test_temp_block[kNumCoeffs]);
+      DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+      DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-      DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
+      DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+      DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
 #endif

      // Initialize a test block with input range [-mask_, mask_].
@@ -416,9 +395,9 @@ class Trans16x16TestBase {
  void RunCoeffCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 1000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_block, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, input_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_block[kNumCoeffs]);

    for (int i = 0; i < count_test_block; ++i) {
      // Initialize a test block with input range [-mask_, mask_].
@@ -437,15 +416,13 @@ class Trans16x16TestBase {
  void RunMemCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 1000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_extreme_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_block, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, input_extreme_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_block[kNumCoeffs]);

    for (int i = 0; i < count_test_block; ++i) {
      // Initialize a test block with input range [-mask_, mask_].
      for (int j = 0; j < kNumCoeffs; ++j) {
-        input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);
        input_extreme_block[j] = rnd.Rand8() % 2 ? mask_ : -mask_;
      }
      if (i == 0) {
@@ -472,24 +449,19 @@ class Trans16x16TestBase {
  void RunQuantCheck(int dc_thred, int ac_thred) {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 100000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_extreme_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, input_extreme_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);

-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, ref, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, ref[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, ref16, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint16_t, ref16[kNumCoeffs]);
 #endif

    for (int i = 0; i < count_test_block; ++i) {
      // Initialize a test block with input range [-mask_, mask_].
      for (int j = 0; j < kNumCoeffs; ++j) {
-        if (bit_depth_ == VPX_BITS_8)
-          input_block[j] = rnd.Rand8() - rnd.Rand8();
-        else
-          input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);
        input_extreme_block[j] = rnd.Rand8() % 2 ? mask_ : -mask_;
      }
      if (i == 0)
@@ -502,11 +474,11 @@ class Trans16x16TestBase {
      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, tx_type_);

      // clear reconstructed pixel buffers
-      vpx_memset(dst, 0, kNumCoeffs * sizeof(uint8_t));
-      vpx_memset(ref, 0, kNumCoeffs * sizeof(uint8_t));
+      memset(dst, 0, kNumCoeffs * sizeof(uint8_t));
+      memset(ref, 0, kNumCoeffs * sizeof(uint8_t));
 #if CONFIG_VP9_HIGHBITDEPTH
-      vpx_memset(dst16, 0, kNumCoeffs * sizeof(uint16_t));
-      vpx_memset(ref16, 0, kNumCoeffs * sizeof(uint16_t));
+      memset(dst16, 0, kNumCoeffs * sizeof(uint16_t));
+      memset(ref16, 0, kNumCoeffs * sizeof(uint16_t));
 #endif

      // quantization with maximum allowed step sizes
@@ -539,13 +511,13 @@ class Trans16x16TestBase {
  void RunInvAccuracyCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 1000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, in[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
 #endif  // CONFIG_VP9_HIGHBITDEPTH

    for (int i = 0; i < count_test_block; ++i) {
@@ -599,12 +571,12 @@ class Trans16x16TestBase {
    const int count_test_block = 10000;
    const int eob = 10;
    const int16_t *scan = vp9_default_scan_orders[TX_16X16].scan;
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, ref, kNumCoeffs);
+    DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, ref[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, ref16, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint16_t, ref16[kNumCoeffs]);
 #endif  // CONFIG_VP9_HIGHBITDEPTH

    for (int i = 0; i < count_test_block; ++i) {
@@ -826,14 +798,14 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    C, Trans16x16DCT,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fdct16x16_c, &idct16x16_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct16x16_c, &idct16x16_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_fdct16x16_c, &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vpx_highbd_fdct16x16_c, &idct16x16_10, 0, VPX_BITS_10),
+        make_tuple(&vpx_highbd_fdct16x16_c, &idct16x16_12, 0, VPX_BITS_12),
+        make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c, 0, VPX_BITS_8)));
 #else
 INSTANTIATE_TEST_CASE_P(
    C, Trans16x16DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct16x16_c, &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c, 0, VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH

 #if CONFIG_VP9_HIGHBITDEPTH
@@ -866,16 +838,16 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    NEON, Trans16x16DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_neon, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_neon, 0, VPX_BITS_8)));
 #endif

 #if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans16x16DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct16x16_sse2,
-                   &vp9_idct16x16_256_add_sse2, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct16x16_sse2,
+                   &vpx_idct16x16_256_add_sse2, 0, VPX_BITS_8)));
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans16x16HT,
    ::testing::Values(
@@ -893,27 +865,19 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans16x16DCT,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fdct16x16_sse2,
+        make_tuple(&vpx_highbd_fdct16x16_sse2,
                   &idct16x16_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct16x16_c,
+        make_tuple(&vpx_highbd_fdct16x16_c,
                   &idct16x16_256_add_10_sse2, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct16x16_sse2,
+        make_tuple(&vpx_highbd_fdct16x16_sse2,
                   &idct16x16_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fdct16x16_c,
+        make_tuple(&vpx_highbd_fdct16x16_c,
                   &idct16x16_256_add_12_sse2, 0, VPX_BITS_12),
-        make_tuple(&vp9_fdct16x16_sse2,
-                   &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct16x16_sse2,
+                   &vpx_idct16x16_256_add_c, 0, VPX_BITS_8)));
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans16x16HT,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_10, 1, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_10, 2, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_10, 3, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_12, 1, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_12, 2, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fht16x16_sse2, &iht16x16_12, 3, VPX_BITS_12),
        make_tuple(&vp9_fht16x16_sse2, &vp9_iht16x16_256_add_c, 0, VPX_BITS_8),
        make_tuple(&vp9_fht16x16_sse2, &vp9_iht16x16_256_add_c, 1, VPX_BITS_8),
        make_tuple(&vp9_fht16x16_sse2, &vp9_iht16x16_256_add_c, 2, VPX_BITS_8),
@@ -934,11 +898,19 @@ INSTANTIATE_TEST_CASE_P(
                   &idct16x16_256_add_12_sse2, 3167, VPX_BITS_12)));
 #endif  // HAVE_SSE2 && CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

-#if HAVE_SSSE3 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#if HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
-    SSSE3, Trans16x16DCT,
+    MSA, Trans16x16DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct16x16_c, &vp9_idct16x16_256_add_ssse3, 0,
+        make_tuple(&vpx_fdct16x16_msa,
+                   &vpx_idct16x16_256_add_msa, 0, VPX_BITS_8)));
+INSTANTIATE_TEST_CASE_P(
+    MSA, Trans16x16HT,
+    ::testing::Values(
+        make_tuple(&vp9_fht16x16_msa, &vp9_iht16x16_256_add_msa, 0, VPX_BITS_8),
+        make_tuple(&vp9_fht16x16_msa, &vp9_iht16x16_256_add_msa, 1, VPX_BITS_8),
+        make_tuple(&vp9_fht16x16_msa, &vp9_iht16x16_256_add_msa, 2, VPX_BITS_8),
+        make_tuple(&vp9_fht16x16_msa, &vp9_iht16x16_256_add_msa, 3,
                   VPX_BITS_8)));
-#endif  // HAVE_SSSE3 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#endif  // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 }  // namespace
--- a/test/dct32x32_test.cc
+++ b/test/dct32x32_test.cc
@@ -13,16 +13,18 @@
 #include <string.h>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp9_rtcd.h"
+#include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vpx_config.h"
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
 #include "vpx/vpx_codec.h"
 #include "vpx/vpx_integer.h"
+#include "vpx_ports/mem.h"

 using libvpx_test::ACMRandom;

@@ -80,15 +82,15 @@ typedef std::tr1::tuple<FwdTxfmFunc, InvTxfmFunc, int, vpx_bit_depth_t>

 #if CONFIG_VP9_HIGHBITDEPTH
 void idct32x32_8(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct32x32_1024_add_c(in, out, stride, 8);
+  vpx_highbd_idct32x32_1024_add_c(in, out, stride, 8);
 }

 void idct32x32_10(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct32x32_1024_add_c(in, out, stride, 10);
+  vpx_highbd_idct32x32_1024_add_c(in, out, stride, 10);
 }

 void idct32x32_12(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct32x32_1024_add_c(in, out, stride, 12);
+  vpx_highbd_idct32x32_1024_add_c(in, out, stride, 12);
 }
 #endif  // CONFIG_VP9_HIGHBITDEPTH

@@ -119,13 +121,13 @@ TEST_P(Trans32x32Test, AccuracyCheck) {
  uint32_t max_error = 0;
  int64_t total_error = 0;
  const int count_test_block = 10000;
-  DECLARE_ALIGNED_ARRAY(16, int16_t, test_input_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_temp_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+  DECLARE_ALIGNED(16, int16_t, test_input_block[kNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, test_temp_block[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
+  DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
 #endif

  for (int i = 0; i < count_test_block; ++i) {
@@ -184,16 +186,16 @@ TEST_P(Trans32x32Test, CoeffCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
  const int count_test_block = 1000;

-  DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_block, kNumCoeffs);
+  DECLARE_ALIGNED(16, int16_t, input_block[kNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, output_block[kNumCoeffs]);

  for (int i = 0; i < count_test_block; ++i) {
    for (int j = 0; j < kNumCoeffs; ++j)
      input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);

    const int stride = 32;
-    vp9_fdct32x32_c(input_block, output_ref_block, stride);
+    vpx_fdct32x32_c(input_block, output_ref_block, stride);
    ASM_REGISTER_STATE_CHECK(fwd_txfm_(input_block, output_block, stride));

    if (version_ == 0) {
@@ -212,15 +214,13 @@ TEST_P(Trans32x32Test, MemCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
  const int count_test_block = 2000;

-  DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, input_extreme_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_block, kNumCoeffs);
+  DECLARE_ALIGNED(16, int16_t, input_extreme_block[kNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, output_block[kNumCoeffs]);

  for (int i = 0; i < count_test_block; ++i) {
    // Initialize a test block with input range [-mask_, mask_].
    for (int j = 0; j < kNumCoeffs; ++j) {
-      input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);
      input_extreme_block[j] = rnd.Rand8() & 1 ? mask_ : -mask_;
    }
    if (i == 0) {
@@ -232,7 +232,7 @@ TEST_P(Trans32x32Test, MemCheck) {
    }

    const int stride = 32;
-    vp9_fdct32x32_c(input_extreme_block, output_ref_block, stride);
+    vpx_fdct32x32_c(input_extreme_block, output_ref_block, stride);
    ASM_REGISTER_STATE_CHECK(
        fwd_txfm_(input_extreme_block, output_block, stride));

@@ -257,13 +257,13 @@ TEST_P(Trans32x32Test, MemCheck) {
 TEST_P(Trans32x32Test, InverseAccuracy) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
  const int count_test_block = 1000;
-  DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+  DECLARE_ALIGNED(16, int16_t, in[kNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
+  DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
 #endif

  for (int i = 0; i < count_test_block; ++i) {
@@ -315,61 +315,61 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    C, Trans32x32Test,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fdct32x32_c,
+        make_tuple(&vpx_highbd_fdct32x32_c,
                   &idct32x32_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct32x32_rd_c,
+        make_tuple(&vpx_highbd_fdct32x32_rd_c,
                   &idct32x32_10, 1, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct32x32_c,
+        make_tuple(&vpx_highbd_fdct32x32_c,
                   &idct32x32_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fdct32x32_rd_c,
+        make_tuple(&vpx_highbd_fdct32x32_rd_c,
                   &idct32x32_12, 1, VPX_BITS_12),
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c, 0, VPX_BITS_8),
-        make_tuple(&vp9_fdct32x32_rd_c,
-                   &vp9_idct32x32_1024_add_c, 1, VPX_BITS_8)));
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c, 0, VPX_BITS_8),
+        make_tuple(&vpx_fdct32x32_rd_c,
+                   &vpx_idct32x32_1024_add_c, 1, VPX_BITS_8)));
 #else
 INSTANTIATE_TEST_CASE_P(
    C, Trans32x32Test,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c, 0, VPX_BITS_8),
-        make_tuple(&vp9_fdct32x32_rd_c,
-                   &vp9_idct32x32_1024_add_c, 1, VPX_BITS_8)));
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c, 0, VPX_BITS_8),
+        make_tuple(&vpx_fdct32x32_rd_c,
+                   &vpx_idct32x32_1024_add_c, 1, VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH

 #if HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    NEON, Trans32x32Test,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_neon, 0, VPX_BITS_8),
-        make_tuple(&vp9_fdct32x32_rd_c,
-                   &vp9_idct32x32_1024_add_neon, 1, VPX_BITS_8)));
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_neon, 0, VPX_BITS_8),
+        make_tuple(&vpx_fdct32x32_rd_c,
+                   &vpx_idct32x32_1024_add_neon, 1, VPX_BITS_8)));
 #endif  // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

 #if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans32x32Test,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_sse2,
-                   &vp9_idct32x32_1024_add_sse2, 0, VPX_BITS_8),
-        make_tuple(&vp9_fdct32x32_rd_sse2,
-                   &vp9_idct32x32_1024_add_sse2, 1, VPX_BITS_8)));
+        make_tuple(&vpx_fdct32x32_sse2,
+                   &vpx_idct32x32_1024_add_sse2, 0, VPX_BITS_8),
+        make_tuple(&vpx_fdct32x32_rd_sse2,
+                   &vpx_idct32x32_1024_add_sse2, 1, VPX_BITS_8)));
 #endif  // HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

 #if HAVE_SSE2 && CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans32x32Test,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fdct32x32_sse2, &idct32x32_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct32x32_rd_sse2, &idct32x32_10, 1,
+        make_tuple(&vpx_highbd_fdct32x32_sse2, &idct32x32_10, 0, VPX_BITS_10),
+        make_tuple(&vpx_highbd_fdct32x32_rd_sse2, &idct32x32_10, 1,
                   VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct32x32_sse2, &idct32x32_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fdct32x32_rd_sse2, &idct32x32_12, 1,
+        make_tuple(&vpx_highbd_fdct32x32_sse2, &idct32x32_12, 0, VPX_BITS_12),
+        make_tuple(&vpx_highbd_fdct32x32_rd_sse2, &idct32x32_12, 1,
                   VPX_BITS_12),
-        make_tuple(&vp9_fdct32x32_sse2, &vp9_idct32x32_1024_add_c, 0,
+        make_tuple(&vpx_fdct32x32_sse2, &vpx_idct32x32_1024_add_c, 0,
                   VPX_BITS_8),
-        make_tuple(&vp9_fdct32x32_rd_sse2, &vp9_idct32x32_1024_add_c, 1,
+        make_tuple(&vpx_fdct32x32_rd_sse2, &vpx_idct32x32_1024_add_c, 1,
                   VPX_BITS_8)));
 #endif  // HAVE_SSE2 && CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

@@ -377,9 +377,19 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    AVX2, Trans32x32Test,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_avx2,
-                   &vp9_idct32x32_1024_add_sse2, 0, VPX_BITS_8),
-        make_tuple(&vp9_fdct32x32_rd_avx2,
-                   &vp9_idct32x32_1024_add_sse2, 1, VPX_BITS_8)));
+        make_tuple(&vpx_fdct32x32_avx2,
+                   &vpx_idct32x32_1024_add_sse2, 0, VPX_BITS_8),
+        make_tuple(&vpx_fdct32x32_rd_avx2,
+                   &vpx_idct32x32_1024_add_sse2, 1, VPX_BITS_8)));
 #endif  // HAVE_AVX2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+
+#if HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+INSTANTIATE_TEST_CASE_P(
+    MSA, Trans32x32Test,
+    ::testing::Values(
+        make_tuple(&vpx_fdct32x32_msa,
+                   &vpx_idct32x32_1024_add_msa, 0, VPX_BITS_8),
+        make_tuple(&vpx_fdct32x32_rd_msa,
+                   &vpx_idct32x32_1024_add_msa, 1, VPX_BITS_8)));
+#endif  // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 }  // namespace
--- a/test/decode_api_test.cc
+++ b/test/decode_api_test.cc
@@ -7,10 +7,11 @@
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */
+
 #include "third_party/googletest/src/include/gtest/gtest.h"

-#include "test/ivf_video_source.h"
 #include "./vpx_config.h"
+#include "test/ivf_video_source.h"
 #include "vpx/vp8dx.h"
 #include "vpx/vpx_decoder.h"

@@ -25,6 +26,9 @@ TEST(DecodeAPI, InvalidParams) {
 #endif
 #if CONFIG_VP9_DECODER
    &vpx_codec_vp9_dx_algo,
+#endif
+#if CONFIG_VP10_DECODER
+    &vpx_codec_vp10_dx_algo,
 #endif
  };
  uint8_t buf[1] = {0};
@@ -129,8 +133,13 @@ TEST(DecodeAPI, Vp9InvalidDecode) {
  vpx_codec_ctx_t dec;
  EXPECT_EQ(VPX_CODEC_OK, vpx_codec_dec_init(&dec, codec, NULL, 0));
  const uint32_t frame_size = static_cast<uint32_t>(video.frame_size());
+#if CONFIG_VP9_HIGHBITDEPTH
  EXPECT_EQ(VPX_CODEC_MEM_ERROR,
            vpx_codec_decode(&dec, video.cxdata(), frame_size, NULL, 0));
+#else
+  EXPECT_EQ(VPX_CODEC_UNSUP_BITSTREAM,
+            vpx_codec_decode(&dec, video.cxdata(), frame_size, NULL, 0));
+#endif
  vpx_codec_iter_t iter = NULL;
  EXPECT_EQ(NULL, vpx_codec_get_frame(&dec, &iter));

--- a/test/decode_test_driver.cc
+++ b/test/decode_test_driver.cc
@@ -7,9 +7,11 @@
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
 #include "test/codec_factory.h"
 #include "test/decode_test_driver.h"
-#include "third_party/googletest/src/include/gtest/gtest.h"
 #include "test/register_state_check.h"
 #include "test/video_source.h"

--- a/test/encode_test_driver.cc
+++ b/test/encode_test_driver.cc
@@ -10,13 +10,14 @@

 #include <string>

+#include "third_party/googletest/src/include/gtest/gtest.h"
+
 #include "./vpx_config.h"
 #include "test/codec_factory.h"
-#include "test/encode_test_driver.h"
 #include "test/decode_test_driver.h"
+#include "test/encode_test_driver.h"
 #include "test/register_state_check.h"
 #include "test/video_source.h"
-#include "third_party/googletest/src/include/gtest/gtest.h"

 namespace libvpx_test {
 void Encoder::InitEncoder(VideoSource *video) {
@@ -29,8 +30,6 @@ void Encoder::InitEncoder(VideoSource *video) {
    cfg_.g_timebase = video->timebase();
    cfg_.rc_twopass_stats_in = stats_->buf();

-    // Default to 1 thread.
-    cfg_.g_threads = 1;
    res = vpx_codec_enc_init(&encoder_, CodecInterface(), &cfg_,
                             init_flags_);
    ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
@@ -43,6 +42,15 @@ void Encoder::InitEncoder(VideoSource *video) {
                               log2_tile_columns);
      ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
    } else
+#endif
+#if CONFIG_VP10_ENCODER
+    if (CodecInterface() == &vpx_codec_vp10_cx_algo) {
+      // Default to 1 tile column for VP10.
+      const int log2_tile_columns = 0;
+      res = vpx_codec_control_(&encoder_, VP9E_SET_TILE_COLUMNS,
+                               log2_tile_columns);
+      ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+    } else
 #endif
    {
 #if CONFIG_VP8_ENCODER
@@ -187,6 +195,7 @@ void EncoderTest::RunLoop(VideoSource *video) {

    video->Begin();
    encoder->InitEncoder(video);
+    ASSERT_FALSE(::testing::Test::HasFatalFailure());

    unsigned long dec_init_flags = 0;  // NOLINT
    // Use fragment decoder if encoder outputs partitions.
--- a/test/encode_test_driver.h
+++ b/test/encode_test_driver.h
@@ -13,12 +13,13 @@
 #include <string>
 #include <vector>

-#include "./vpx_config.h"
 #include "third_party/googletest/src/include/gtest/gtest.h"
-#include "vpx/vpx_encoder.h"
-#if CONFIG_VP8_ENCODER || CONFIG_VP9_ENCODER
+
+#include "./vpx_config.h"
+#if CONFIG_VP8_ENCODER || CONFIG_VP9_ENCODER || CONFIG_VP10_ENCODER
 #include "vpx/vp8cx.h"
 #endif
+#include "vpx/vpx_encoder.h"

 namespace libvpx_test {

@@ -133,7 +134,11 @@ class Encoder {
    ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
  }

-#if CONFIG_VP8_ENCODER || CONFIG_VP9_ENCODER
+  void Control(int ctrl_id, struct vpx_svc_parameters *arg) {
+    const vpx_codec_err_t res = vpx_codec_control_(&encoder_, ctrl_id, arg);
+    ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
+  }
+#if CONFIG_VP8_ENCODER || CONFIG_VP9_ENCODER || CONFIG_VP10_ENCODER
  void Control(int ctrl_id, vpx_active_map_t *arg) {
    const vpx_codec_err_t res = vpx_codec_control_(&encoder_, ctrl_id, arg);
    ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
@@ -183,7 +188,10 @@ class EncoderTest {
 protected:
  explicit EncoderTest(const CodecFactory *codec)
      : codec_(codec), abort_(false), init_flags_(0), frame_flags_(0),
-        last_pts_(0) {}
+        last_pts_(0) {
+    // Default to 1 thread.
+    cfg_.g_threads = 1;
+  }

  virtual ~EncoderTest() {}

--- a/test/error_resilience_test.cc
+++ b/test/error_resilience_test.cc
@@ -20,10 +20,11 @@ const int kMaxErrorFrames = 12;
 const int kMaxDroppableFrames = 12;

 class ErrorResilienceTestLarge : public ::libvpx_test::EncoderTest,
-    public ::libvpx_test::CodecTestWithParam<libvpx_test::TestMode> {
+    public ::libvpx_test::CodecTestWith2Params<libvpx_test::TestMode, bool> {
 protected:
  ErrorResilienceTestLarge()
      : EncoderTest(GET_PARAM(0)),
+        svc_support_(GET_PARAM(2)),
        psnr_(0.0),
        nframes_(0),
        mismatch_psnr_(0.0),
@@ -193,6 +194,8 @@ class ErrorResilienceTestLarge : public ::libvpx_test::EncoderTest,
     pattern_switch_ = frame_switch;
   }

+  bool svc_support_;
+
 private:
  double psnr_;
  unsigned int nframes_;
@@ -302,6 +305,10 @@ TEST_P(ErrorResilienceTestLarge, DropFramesWithoutRecovery) {
 // two layer temporal pattern. The base layer does not predict from the top
 // layer, so successful decoding is expected.
 TEST_P(ErrorResilienceTestLarge, 2LayersDropEnhancement) {
+  // This test doesn't run if SVC is not supported.
+  if (!svc_support_)
+    return;
+
  const vpx_rational timebase = { 33333333, 1000000000 };
  cfg_.g_timebase = timebase;
  cfg_.rc_target_bitrate = 500;
@@ -347,6 +354,10 @@ TEST_P(ErrorResilienceTestLarge, 2LayersDropEnhancement) {
 // for a two layer temporal pattern, where at some point in the
 // sequence, the LAST ref is not used anymore.
 TEST_P(ErrorResilienceTestLarge, 2LayersNoRefLast) {
+  // This test doesn't run if SVC is not supported.
+  if (!svc_support_)
+    return;
+
  const vpx_rational timebase = { 33333333, 1000000000 };
  cfg_.g_timebase = timebase;
  cfg_.rc_target_bitrate = 500;
@@ -579,8 +590,13 @@ TEST_P(ErrorResilienceTestLargeCodecControls, CodecControl3TemporalLayers) {
  }
 }

-VP8_INSTANTIATE_TEST_CASE(ErrorResilienceTestLarge, ONE_PASS_TEST_MODES);
+VP8_INSTANTIATE_TEST_CASE(ErrorResilienceTestLarge, ONE_PASS_TEST_MODES,
+                          ::testing::Values(true));
 VP8_INSTANTIATE_TEST_CASE(ErrorResilienceTestLargeCodecControls,
                          ONE_PASS_TEST_MODES);
-VP9_INSTANTIATE_TEST_CASE(ErrorResilienceTestLarge, ONE_PASS_TEST_MODES);
+VP9_INSTANTIATE_TEST_CASE(ErrorResilienceTestLarge, ONE_PASS_TEST_MODES,
+                          ::testing::Values(true));
+// SVC-related tests don't run for VP10 since SVC is not supported.
+VP10_INSTANTIATE_TEST_CASE(ErrorResilienceTestLarge, ONE_PASS_TEST_MODES,
+                           ::testing::Values(false));
 }  // namespace
--- a/test/fdct4x4_test.cc
+++ b/test/fdct4x4_test.cc
@@ -13,15 +13,17 @@
 #include <string.h>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
 #include "vpx/vpx_codec.h"
 #include "vpx/vpx_integer.h"
+#include "vpx_ports/mem.h"

 using libvpx_test::ACMRandom;

@@ -39,7 +41,7 @@ typedef std::tr1::tuple<FhtFunc, IhtFunc, int, vpx_bit_depth_t> Ht4x4Param;

 void fdct4x4_ref(const int16_t *in, tran_low_t *out, int stride,
                 int tx_type) {
-  vp9_fdct4x4_c(in, out, stride);
+  vpx_fdct4x4_c(in, out, stride);
 }

 void fht4x4_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
@@ -53,11 +55,11 @@ void fwht4x4_ref(const int16_t *in, tran_low_t *out, int stride,

 #if CONFIG_VP9_HIGHBITDEPTH
 void idct4x4_10(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct4x4_16_add_c(in, out, stride, 10);
+  vpx_highbd_idct4x4_16_add_c(in, out, stride, 10);
 }

 void idct4x4_12(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct4x4_16_add_c(in, out, stride, 12);
+  vpx_highbd_idct4x4_16_add_c(in, out, stride, 12);
 }

 void iht4x4_10(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
@@ -69,20 +71,20 @@ void iht4x4_12(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
 }

 void iwht4x4_10(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_iwht4x4_16_add_c(in, out, stride, 10);
+  vpx_highbd_iwht4x4_16_add_c(in, out, stride, 10);
 }

 void iwht4x4_12(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_iwht4x4_16_add_c(in, out, stride, 12);
+  vpx_highbd_iwht4x4_16_add_c(in, out, stride, 12);
 }

 #if HAVE_SSE2
 void idct4x4_10_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct4x4_16_add_sse2(in, out, stride, 10);
+  vpx_highbd_idct4x4_16_add_sse2(in, out, stride, 10);
 }

 void idct4x4_12_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct4x4_16_add_sse2(in, out, stride, 12);
+  vpx_highbd_idct4x4_16_add_sse2(in, out, stride, 12);
 }
 #endif  // HAVE_SSE2
 #endif  // CONFIG_VP9_HIGHBITDEPTH
@@ -102,13 +104,13 @@ class Trans4x4TestBase {
    int64_t total_error = 0;
    const int count_test_block = 10000;
    for (int i = 0; i < count_test_block; ++i) {
-      DECLARE_ALIGNED_ARRAY(16, int16_t, test_input_block, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_temp_block, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+      DECLARE_ALIGNED(16, int16_t, test_input_block[kNumCoeffs]);
+      DECLARE_ALIGNED(16, tran_low_t, test_temp_block[kNumCoeffs]);
+      DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+      DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-      DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-      DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
+      DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+      DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
 #endif

      // Initialize a test block with input range [-255, 255].
@@ -142,6 +144,7 @@ class Trans4x4TestBase {
        const uint32_t diff =
            bit_depth_ == VPX_BITS_8 ? dst[j] - src[j] : dst16[j] - src16[j];
 #else
+        ASSERT_EQ(VPX_BITS_8, bit_depth_);
        const uint32_t diff = dst[j] - src[j];
 #endif
        const uint32_t error = diff * diff;
@@ -163,9 +166,9 @@ class Trans4x4TestBase {
  void RunCoeffCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 5000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_block, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, input_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_block[kNumCoeffs]);

    for (int i = 0; i < count_test_block; ++i) {
      // Initialize a test block with input range [-mask_, mask_].
@@ -184,15 +187,13 @@ class Trans4x4TestBase {
  void RunMemCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 5000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, int16_t, input_extreme_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_block, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, input_extreme_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, output_block[kNumCoeffs]);

    for (int i = 0; i < count_test_block; ++i) {
      // Initialize a test block with input range [-mask_, mask_].
      for (int j = 0; j < kNumCoeffs; ++j) {
-        input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);
        input_extreme_block[j] = rnd.Rand8() % 2 ? mask_ : -mask_;
      }
      if (i == 0) {
@@ -219,13 +220,13 @@ class Trans4x4TestBase {
  void RunInvAccuracyCheck(int limit) {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 1000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, in[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
 #endif

    for (int i = 0; i < count_test_block; ++i) {
@@ -418,14 +419,14 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    C, Trans4x4DCT,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fdct4x4_c, &idct4x4_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct4x4_c, &idct4x4_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_fdct4x4_c, &vp9_idct4x4_16_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vpx_highbd_fdct4x4_c, &idct4x4_10, 0, VPX_BITS_10),
+        make_tuple(&vpx_highbd_fdct4x4_c, &idct4x4_12, 0, VPX_BITS_12),
+        make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c, 0, VPX_BITS_8)));
 #else
 INSTANTIATE_TEST_CASE_P(
    C, Trans4x4DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct4x4_c, &vp9_idct4x4_16_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c, 0, VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH

 #if CONFIG_VP9_HIGHBITDEPTH
@@ -460,20 +461,20 @@ INSTANTIATE_TEST_CASE_P(
    ::testing::Values(
        make_tuple(&vp9_highbd_fwht4x4_c, &iwht4x4_10, 0, VPX_BITS_10),
        make_tuple(&vp9_highbd_fwht4x4_c, &iwht4x4_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_fwht4x4_c, &vp9_iwht4x4_16_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vp9_fwht4x4_c, &vpx_iwht4x4_16_add_c, 0, VPX_BITS_8)));
 #else
 INSTANTIATE_TEST_CASE_P(
    C, Trans4x4WHT,
    ::testing::Values(
-        make_tuple(&vp9_fwht4x4_c, &vp9_iwht4x4_16_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vp9_fwht4x4_c, &vpx_iwht4x4_16_add_c, 0, VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH

 #if HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    NEON, Trans4x4DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct4x4_c,
-                   &vp9_idct4x4_16_add_neon, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct4x4_c,
+                   &vpx_idct4x4_16_add_neon, 0, VPX_BITS_8)));
 #endif  // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

 #if HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@@ -491,15 +492,23 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    MMX, Trans4x4WHT,
    ::testing::Values(
-        make_tuple(&vp9_fwht4x4_mmx, &vp9_iwht4x4_16_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vp9_fwht4x4_mmx, &vpx_iwht4x4_16_add_c, 0, VPX_BITS_8)));
+#endif
+
+#if CONFIG_USE_X86INC && HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && \
+    !CONFIG_EMULATE_HARDWARE
+INSTANTIATE_TEST_CASE_P(
+    SSE2, Trans4x4WHT,
+    ::testing::Values(
+        make_tuple(&vp9_fwht4x4_c, &vpx_iwht4x4_16_add_sse2, 0, VPX_BITS_8)));
 #endif

 #if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans4x4DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct4x4_sse2,
-                   &vp9_idct4x4_16_add_sse2, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct4x4_sse2,
+                   &vpx_idct4x4_16_add_sse2, 0, VPX_BITS_8)));
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans4x4HT,
    ::testing::Values(
@@ -513,27 +522,33 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans4x4DCT,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fdct4x4_c,    &idct4x4_10_sse2, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct4x4_sse2, &idct4x4_10_sse2, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct4x4_c,    &idct4x4_12_sse2, 0, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fdct4x4_sse2, &idct4x4_12_sse2, 0, VPX_BITS_12),
-        make_tuple(&vp9_fdct4x4_sse2,      &vp9_idct4x4_16_add_c, 0,
+        make_tuple(&vpx_highbd_fdct4x4_c,    &idct4x4_10_sse2, 0, VPX_BITS_10),
+        make_tuple(&vpx_highbd_fdct4x4_sse2, &idct4x4_10_sse2, 0, VPX_BITS_10),
+        make_tuple(&vpx_highbd_fdct4x4_c,    &idct4x4_12_sse2, 0, VPX_BITS_12),
+        make_tuple(&vpx_highbd_fdct4x4_sse2, &idct4x4_12_sse2, 0, VPX_BITS_12),
+        make_tuple(&vpx_fdct4x4_sse2,      &vpx_idct4x4_16_add_c, 0,
                   VPX_BITS_8)));

 INSTANTIATE_TEST_CASE_P(
    SSE2, Trans4x4HT,
    ::testing::Values(
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_10, 1, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_10, 2, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_10, 3, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_12, 0, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_12, 1, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_12, 2, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fht4x4_sse2, &iht4x4_12, 3, VPX_BITS_12),
        make_tuple(&vp9_fht4x4_sse2, &vp9_iht4x4_16_add_c, 0, VPX_BITS_8),
        make_tuple(&vp9_fht4x4_sse2, &vp9_iht4x4_16_add_c, 1, VPX_BITS_8),
        make_tuple(&vp9_fht4x4_sse2, &vp9_iht4x4_16_add_c, 2, VPX_BITS_8),
        make_tuple(&vp9_fht4x4_sse2, &vp9_iht4x4_16_add_c, 3, VPX_BITS_8)));
 #endif  // HAVE_SSE2 && CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+
+#if HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+INSTANTIATE_TEST_CASE_P(
+    MSA, Trans4x4DCT,
+    ::testing::Values(
+        make_tuple(&vpx_fdct4x4_msa, &vpx_idct4x4_16_add_msa, 0, VPX_BITS_8)));
+INSTANTIATE_TEST_CASE_P(
+    MSA, Trans4x4HT,
+    ::testing::Values(
+        make_tuple(&vp9_fht4x4_msa, &vp9_iht4x4_16_add_msa, 0, VPX_BITS_8),
+        make_tuple(&vp9_fht4x4_msa, &vp9_iht4x4_16_add_msa, 1, VPX_BITS_8),
+        make_tuple(&vp9_fht4x4_msa, &vp9_iht4x4_16_add_msa, 2, VPX_BITS_8),
+        make_tuple(&vp9_fht4x4_msa, &vp9_iht4x4_16_add_msa, 3, VPX_BITS_8)));
+#endif  // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 }  // namespace
--- a/test/fdct8x8_test.cc
+++ b/test/fdct8x8_test.cc
@@ -13,18 +13,40 @@
 #include <string.h>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
+#include "vp9/common/vp9_scan.h"
 #include "vpx/vpx_codec.h"
 #include "vpx/vpx_integer.h"
+#include "vpx_ports/mem.h"
+
+using libvpx_test::ACMRandom;
+
+namespace {

 const int kNumCoeffs = 64;
 const double kPi = 3.141592653589793238462643383279502884;
+
+const int kSignBiasMaxDiff255 = 1500;
+const int kSignBiasMaxDiff15 = 10000;
+
+typedef void (*FdctFunc)(const int16_t *in, tran_low_t *out, int stride);
+typedef void (*IdctFunc)(const tran_low_t *in, uint8_t *out, int stride);
+typedef void (*FhtFunc)(const int16_t *in, tran_low_t *out, int stride,
+                        int tx_type);
+typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
+                        int tx_type);
+
+typedef std::tr1::tuple<FdctFunc, IdctFunc, int, vpx_bit_depth_t> Dct8x8Param;
+typedef std::tr1::tuple<FhtFunc, IhtFunc, int, vpx_bit_depth_t> Ht8x8Param;
+typedef std::tr1::tuple<IdctFunc, IdctFunc, int, vpx_bit_depth_t> Idct8x8Param;
+
 void reference_8x8_dct_1d(const double in[8], double out[8], int stride) {
  const double kInvSqrt2 = 0.707106781186547524400844362104;
  for (int k = 0; k < 8; k++) {
@@ -59,26 +81,9 @@ void reference_8x8_dct_2d(const int16_t input[kNumCoeffs],
  }
 }

-using libvpx_test::ACMRandom;
-
-namespace {
-
-const int kSignBiasMaxDiff255 = 1500;
-const int kSignBiasMaxDiff15 = 10000;
-
-typedef void (*FdctFunc)(const int16_t *in, tran_low_t *out, int stride);
-typedef void (*IdctFunc)(const tran_low_t *in, uint8_t *out, int stride);
-typedef void (*FhtFunc)(const int16_t *in, tran_low_t *out, int stride,
-                        int tx_type);
-typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
-
-typedef std::tr1::tuple<FdctFunc, IdctFunc, int, vpx_bit_depth_t> Dct8x8Param;
-typedef std::tr1::tuple<FhtFunc, IhtFunc, int, vpx_bit_depth_t> Ht8x8Param;
-typedef std::tr1::tuple<IdctFunc, IdctFunc, int, vpx_bit_depth_t> Idct8x8Param;

 void fdct8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  vp9_fdct8x8_c(in, out, stride);
+  vpx_fdct8x8_c(in, out, stride);
 }

 void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
@@ -87,11 +92,11 @@ void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {

 #if CONFIG_VP9_HIGHBITDEPTH
 void idct8x8_10(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_64_add_c(in, out, stride, 10);
+  vpx_highbd_idct8x8_64_add_c(in, out, stride, 10);
 }

 void idct8x8_12(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_64_add_c(in, out, stride, 12);
+  vpx_highbd_idct8x8_64_add_c(in, out, stride, 12);
 }

 void iht8x8_10(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
@@ -103,28 +108,28 @@ void iht8x8_12(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
 }

 void idct8x8_10_add_10_c(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_10_add_c(in, out, stride, 10);
+  vpx_highbd_idct8x8_10_add_c(in, out, stride, 10);
 }

 void idct8x8_10_add_12_c(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_10_add_c(in, out, stride, 12);
+  vpx_highbd_idct8x8_10_add_c(in, out, stride, 12);
 }

 #if HAVE_SSE2
 void idct8x8_10_add_10_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_10_add_sse2(in, out, stride, 10);
+  vpx_highbd_idct8x8_10_add_sse2(in, out, stride, 10);
 }

 void idct8x8_10_add_12_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_10_add_sse2(in, out, stride, 12);
+  vpx_highbd_idct8x8_10_add_sse2(in, out, stride, 12);
 }

 void idct8x8_64_add_10_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_64_add_sse2(in, out, stride, 10);
+  vpx_highbd_idct8x8_64_add_sse2(in, out, stride, 10);
 }

 void idct8x8_64_add_12_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  vp9_highbd_idct8x8_64_add_sse2(in, out, stride, 12);
+  vpx_highbd_idct8x8_64_add_sse2(in, out, stride, 12);
 }
 #endif  // HAVE_SSE2
 #endif  // CONFIG_VP9_HIGHBITDEPTH
@@ -139,8 +144,8 @@ class FwdTrans8x8TestBase {

  void RunSignBiasCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
-    DECLARE_ALIGNED_ARRAY(16, int16_t, test_input_block, 64);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_output_block, 64);
+    DECLARE_ALIGNED(16, int16_t, test_input_block[64]);
+    DECLARE_ALIGNED(16, tran_low_t, test_output_block[64]);
    int count_sign_block[64][2];
    const int count_test_block = 100000;

@@ -210,13 +215,13 @@ class FwdTrans8x8TestBase {
    int max_error = 0;
    int total_error = 0;
    const int count_test_block = 100000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, test_input_block, 64);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_temp_block, 64);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, 64);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, src, 64);
+    DECLARE_ALIGNED(16, int16_t, test_input_block[64]);
+    DECLARE_ALIGNED(16, tran_low_t, test_temp_block[64]);
+    DECLARE_ALIGNED(16, uint8_t, dst[64]);
+    DECLARE_ALIGNED(16, uint8_t, src[64]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, 64);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, 64);
+    DECLARE_ALIGNED(16, uint16_t, dst16[64]);
+    DECLARE_ALIGNED(16, uint16_t, src16[64]);
 #endif

    for (int i = 0; i < count_test_block; ++i) {
@@ -287,14 +292,14 @@ class FwdTrans8x8TestBase {
    int total_error = 0;
    int total_coeff_error = 0;
    const int count_test_block = 100000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, test_input_block, 64);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_temp_block, 64);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_temp_block, 64);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, 64);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, src, 64);
+    DECLARE_ALIGNED(16, int16_t, test_input_block[64]);
+    DECLARE_ALIGNED(16, tran_low_t, test_temp_block[64]);
+    DECLARE_ALIGNED(16, tran_low_t, ref_temp_block[64]);
+    DECLARE_ALIGNED(16, uint8_t, dst[64]);
+    DECLARE_ALIGNED(16, uint8_t, src[64]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, 64);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, 64);
+    DECLARE_ALIGNED(16, uint16_t, dst16[64]);
+    DECLARE_ALIGNED(16, uint16_t, src16[64]);
 #endif

    for (int i = 0; i < count_test_block; ++i) {
@@ -376,13 +381,13 @@ class FwdTrans8x8TestBase {
  void RunInvAccuracyCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 1000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, src, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, in[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, src[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, src16, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint16_t, src16[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
 #endif

    for (int i = 0; i < count_test_block; ++i) {
@@ -434,9 +439,9 @@ class FwdTrans8x8TestBase {
  void RunFwdAccuracyCheck() {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 1000;
-    DECLARE_ALIGNED_ARRAY(16, int16_t, in, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff_r, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
+    DECLARE_ALIGNED(16, int16_t, in[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, coeff_r[kNumCoeffs]);
+    DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);

    for (int i = 0; i < count_test_block; ++i) {
      double out_r[kNumCoeffs];
@@ -464,12 +469,12 @@ void CompareInvReference(IdctFunc ref_txfm, int thresh) {
    ACMRandom rnd(ACMRandom::DeterministicSeed());
    const int count_test_block = 10000;
    const int eob = 12;
-    DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, dst, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint8_t, ref, kNumCoeffs);
+    DECLARE_ALIGNED(16, tran_low_t, coeff[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, dst[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint8_t, ref[kNumCoeffs]);
 #if CONFIG_VP9_HIGHBITDEPTH
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, dst16, kNumCoeffs);
-    DECLARE_ALIGNED_ARRAY(16, uint16_t, ref16, kNumCoeffs);
+    DECLARE_ALIGNED(16, uint16_t, dst16[kNumCoeffs]);
+    DECLARE_ALIGNED(16, uint16_t, ref16[kNumCoeffs]);
 #endif
    const int16_t *scan = vp9_default_scan_orders[TX_8X8].scan;

@@ -653,14 +658,14 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    C, FwdTrans8x8DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_c, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8),
-        make_tuple(&vp9_highbd_fdct8x8_c, &idct8x8_10, 0, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct8x8_c, &idct8x8_12, 0, VPX_BITS_12)));
+        make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c, 0, VPX_BITS_8),
+        make_tuple(&vpx_highbd_fdct8x8_c, &idct8x8_10, 0, VPX_BITS_10),
+        make_tuple(&vpx_highbd_fdct8x8_c, &idct8x8_12, 0, VPX_BITS_12)));
 #else
 INSTANTIATE_TEST_CASE_P(
    C, FwdTrans8x8DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_c, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8)));
+        make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c, 0, VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH

 #if CONFIG_VP9_HIGHBITDEPTH
@@ -680,8 +685,6 @@ INSTANTIATE_TEST_CASE_P(
        make_tuple(&vp9_fht8x8_c, &vp9_iht8x8_64_add_c, 2, VPX_BITS_8),
        make_tuple(&vp9_fht8x8_c, &vp9_iht8x8_64_add_c, 3, VPX_BITS_8)));
 #else
-// TODO(jingning): re-enable after this handles the expanded range [0, 65535]
-// returned from Rand16().
 INSTANTIATE_TEST_CASE_P(
    C, FwdTrans8x8HT,
    ::testing::Values(
@@ -692,12 +695,10 @@ INSTANTIATE_TEST_CASE_P(
 #endif  // CONFIG_VP9_HIGHBITDEPTH

 #if HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
-// TODO(jingning): re-enable after this handles the expanded range [0, 65535]
-// returned from Rand16().
 INSTANTIATE_TEST_CASE_P(
    NEON, FwdTrans8x8DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_neon, &vp9_idct8x8_64_add_neon, 0,
+        make_tuple(&vpx_fdct8x8_neon, &vpx_idct8x8_64_add_neon, 0,
                   VPX_BITS_8)));
 #endif  // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

@@ -712,12 +713,10 @@ INSTANTIATE_TEST_CASE_P(
 #endif  // HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

 #if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
-// TODO(jingning): re-enable after these handle the expanded range [0, 65535]
-// returned from Rand16().
 INSTANTIATE_TEST_CASE_P(
    SSE2, FwdTrans8x8DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_sse2, &vp9_idct8x8_64_add_sse2, 0,
+        make_tuple(&vpx_fdct8x8_sse2, &vpx_idct8x8_64_add_sse2, 0,
                   VPX_BITS_8)));
 INSTANTIATE_TEST_CASE_P(
    SSE2, FwdTrans8x8HT,
@@ -732,18 +731,16 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    SSE2, FwdTrans8x8DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_sse2, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8),
-        make_tuple(&vp9_highbd_fdct8x8_c,
+        make_tuple(&vpx_fdct8x8_sse2, &vpx_idct8x8_64_add_c, 0, VPX_BITS_8),
+        make_tuple(&vpx_highbd_fdct8x8_c,
                   &idct8x8_64_add_10_sse2, 12, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct8x8_sse2,
+        make_tuple(&vpx_highbd_fdct8x8_sse2,
                   &idct8x8_64_add_10_sse2, 12, VPX_BITS_10),
-        make_tuple(&vp9_highbd_fdct8x8_c,
+        make_tuple(&vpx_highbd_fdct8x8_c,
                   &idct8x8_64_add_12_sse2, 12, VPX_BITS_12),
-        make_tuple(&vp9_highbd_fdct8x8_sse2,
+        make_tuple(&vpx_highbd_fdct8x8_sse2,
                   &idct8x8_64_add_12_sse2, 12, VPX_BITS_12)));

-// TODO(jingning): re-enable after these handle the expanded range [0, 65535]
-// returned from Rand16().
 INSTANTIATE_TEST_CASE_P(
    SSE2, FwdTrans8x8HT,
    ::testing::Values(
@@ -767,14 +764,26 @@ INSTANTIATE_TEST_CASE_P(
                   &idct8x8_64_add_12_sse2, 6225, VPX_BITS_12)));
 #endif  // HAVE_SSE2 && CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

-#if HAVE_SSSE3 && ARCH_X86_64 && !CONFIG_VP9_HIGHBITDEPTH && \
-    !CONFIG_EMULATE_HARDWARE
-// TODO(jingning): re-enable after this handles the expanded range [0, 65535]
-// returned from Rand16().
+#if HAVE_SSSE3 && CONFIG_USE_X86INC && ARCH_X86_64 && \
+    !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSSE3, FwdTrans8x8DCT,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_ssse3, &vp9_idct8x8_64_add_ssse3, 0,
+        make_tuple(&vpx_fdct8x8_ssse3, &vpx_idct8x8_64_add_ssse3, 0,
                   VPX_BITS_8)));
 #endif
+
+#if HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+INSTANTIATE_TEST_CASE_P(
+    MSA, FwdTrans8x8DCT,
+    ::testing::Values(
+        make_tuple(&vpx_fdct8x8_msa, &vpx_idct8x8_64_add_msa, 0, VPX_BITS_8)));
+INSTANTIATE_TEST_CASE_P(
+    MSA, FwdTrans8x8HT,
+    ::testing::Values(
+        make_tuple(&vp9_fht8x8_msa, &vp9_iht8x8_64_add_msa, 0, VPX_BITS_8),
+        make_tuple(&vp9_fht8x8_msa, &vp9_iht8x8_64_add_msa, 1, VPX_BITS_8),
+        make_tuple(&vp9_fht8x8_msa, &vp9_iht8x8_64_add_msa, 2, VPX_BITS_8),
+        make_tuple(&vp9_fht8x8_msa, &vp9_iht8x8_64_add_msa, 3, VPX_BITS_8)));
+#endif  // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 }  // namespace
--- a/test/frame_size_tests.cc
+++ b/test/frame_size_tests.cc
@@ -74,7 +74,7 @@ TEST_F(VP9FrameSizeTestsLarge, ValidSizes) {
  // size or almost 1 gig of memory.
  // In total the allocations will exceed 2GiB which may cause a failure with
  // mingw + wine, use a smaller size in that case.
-#if defined(_WIN32) && !defined(_WIN64)
+#if defined(_WIN32) && !defined(_WIN64) || defined(__OS2__)
  video.SetSize(4096, 3072);
 #else
  video.SetSize(4096, 4096);
--- a/test/idct8x8_test.cc
+++ b/test/idct8x8_test.cc
@@ -14,8 +14,7 @@

 #include "third_party/googletest/src/include/gtest/gtest.h"

-#include "./vp9_rtcd.h"
-
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "vpx/vpx_integer.h"

@@ -68,43 +67,6 @@ void reference_dct_2d(int16_t input[64], double output[64]) {
    output[i] *= 2;
 }

-void reference_idct_1d(double input[8], double output[8]) {
-  const double kPi = 3.141592653589793238462643383279502884;
-  const double kSqrt2 = 1.414213562373095048801688724209698;
-  for (int k = 0; k < 8; k++) {
-    output[k] = 0.0;
-    for (int n = 0; n < 8; n++) {
-      output[k] += input[n]*cos(kPi*(2*k+1)*n/16.0);
-      if (n == 0)
-        output[k] = output[k]/kSqrt2;
-    }
-  }
-}
-
-void reference_idct_2d(double input[64], int16_t output[64]) {
-  double out[64], out2[64];
-  // First transform rows
-  for (int i = 0; i < 8; ++i) {
-    double temp_in[8], temp_out[8];
-    for (int j = 0; j < 8; ++j)
-      temp_in[j] = input[j + i*8];
-    reference_idct_1d(temp_in, temp_out);
-    for (int j = 0; j < 8; ++j)
-      out[j + i*8] = temp_out[j];
-  }
-  // Then transform columns
-  for (int i = 0; i < 8; ++i) {
-    double temp_in[8], temp_out[8];
-    for (int j = 0; j < 8; ++j)
-      temp_in[j] = out[j*8 + i];
-    reference_idct_1d(temp_in, temp_out);
-    for (int j = 0; j < 8; ++j)
-      out2[j*8 + i] = temp_out[j];
-  }
-  for (int i = 0; i < 64; ++i)
-    output[i] = round(out2[i]/32);
-}
-
 TEST(VP9Idct8x8Test, AccuracyCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
  const int count_test_block = 10000;
@@ -125,7 +87,7 @@ TEST(VP9Idct8x8Test, AccuracyCheck) {
    reference_dct_2d(input, output_r);
    for (int j = 0; j < 64; ++j)
      coeff[j] = round(output_r[j]);
-    vp9_idct8x8_64_add_c(coeff, dst, 8);
+    vpx_idct8x8_64_add_c(coeff, dst, 8);
    for (int j = 0; j < 64; ++j) {
      const int diff = dst[j] - src[j];
      const int error = diff * diff;
--- a/test/idct_test.cc
+++ b/test/idct_test.cc
@@ -10,10 +10,11 @@

 #include "./vpx_config.h"
 #include "./vp8_rtcd.h"
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
+
 #include "third_party/googletest/src/include/gtest/gtest.h"

+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
 #include "vpx/vpx_integer.h"

 typedef void (*IdctFunc)(int16_t *input, unsigned char *pred_ptr,
@@ -113,4 +114,8 @@ INSTANTIATE_TEST_CASE_P(C, IDCTTest, ::testing::Values(vp8_short_idct4x4llm_c));
 INSTANTIATE_TEST_CASE_P(MMX, IDCTTest,
                        ::testing::Values(vp8_short_idct4x4llm_mmx));
 #endif
+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(MSA, IDCTTest,
+                        ::testing::Values(vp8_short_idct4x4llm_msa));
+#endif
 }
--- a/test/intrapred_test.cc
+++ b/test/intrapred_test.cc
@@ -8,15 +8,15 @@
 *  be found in the AUTHORS file in the root of the source tree.
 */

-
 #include <string.h>
-#include "test/acm_random.h"
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
+
 #include "third_party/googletest/src/include/gtest/gtest.h"

 #include "./vpx_config.h"
 #include "./vp8_rtcd.h"
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
 #include "vp8/common/blockd.h"
 #include "vpx_mem/vpx_mem.h"

@@ -299,6 +299,11 @@ INSTANTIATE_TEST_CASE_P(NEON, IntraPredYTest,
                        ::testing::Values(
                            vp8_build_intra_predictors_mby_s_neon));
 #endif
+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(MSA, IntraPredYTest,
+                        ::testing::Values(
+                            vp8_build_intra_predictors_mby_s_msa));
+#endif

 typedef void (*IntraPredUvFunc)(MACROBLOCKD *x,
                                uint8_t *uabove_row,
@@ -392,5 +397,10 @@ INSTANTIATE_TEST_CASE_P(NEON, IntraPredUVTest,
                        ::testing::Values(
                            vp8_build_intra_predictors_mbuv_s_neon));
 #endif
+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(MSA, IntraPredUVTest,
+                        ::testing::Values(
+                            vp8_build_intra_predictors_mbuv_s_msa));
+#endif

 }  // namespace
--- a/test/invalid_file_test.cc
+++ b/test/invalid_file_test.cc
@@ -112,7 +112,9 @@ TEST_P(InvalidFileTest, ReturnCode) {

 const DecodeParam kVP9InvalidFileTests[] = {
  {1, "invalid-vp90-02-v2.webm"},
+#if CONFIG_VP9_HIGHBITDEPTH
  {1, "invalid-vp90-2-00-quantizer-00.webm.ivf.s5861_r01-05_b6-.v2.ivf"},
+#endif
  {1, "invalid-vp90-03-v3.webm"},
  {1, "invalid-vp90-2-00-quantizer-11.webm.ivf.s52984_r01-05_b6-.ivf"},
  {1, "invalid-vp90-2-00-quantizer-11.webm.ivf.s52984_r01-05_b6-z.ivf"},
@@ -143,7 +145,7 @@ TEST_P(InvalidFileInvalidPeekTest, ReturnCode) {
 }

 const DecodeParam kVP9InvalidFileInvalidPeekTests[] = {
-  {1, "invalid-vp90-01-v2.webm"},
+  {1, "invalid-vp90-01-v3.webm"},
 };

 VP9_INSTANTIATE_TEST_CASE(InvalidFileInvalidPeekTest,
--- a/test/lpf_8_test.cc
+++ b/test/lpf_8_test.cc
@@ -13,13 +13,13 @@
 #include <string>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vpx_config.h"
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
 #include "vp9/common/vp9_loopfilter.h"
 #include "vpx/vpx_integer.h"
@@ -52,7 +52,7 @@ typedef void (*dual_loop_op_t)(uint8_t *s, int p, const uint8_t *blimit0,
                               const uint8_t *thresh1);
 #endif  // CONFIG_VP9_HIGHBITDEPTH

-typedef std::tr1::tuple<loop_op_t, loop_op_t, int> loop8_param_t;
+typedef std::tr1::tuple<loop_op_t, loop_op_t, int, int> loop8_param_t;
 typedef std::tr1::tuple<dual_loop_op_t, dual_loop_op_t, int> dualloop8_param_t;

 #if HAVE_SSE2
@@ -60,49 +60,49 @@ typedef std::tr1::tuple<dual_loop_op_t, dual_loop_op_t, int> dualloop8_param_t;
 void wrapper_vertical_16_sse2(uint16_t *s, int p, const uint8_t *blimit,
                              const uint8_t *limit, const uint8_t *thresh,
                              int count, int bd) {
-  vp9_highbd_lpf_vertical_16_sse2(s, p, blimit, limit, thresh, bd);
+  vpx_highbd_lpf_vertical_16_sse2(s, p, blimit, limit, thresh, bd);
 }

 void wrapper_vertical_16_c(uint16_t *s, int p, const uint8_t *blimit,
                           const uint8_t *limit, const uint8_t *thresh,
                           int count, int bd) {
-  vp9_highbd_lpf_vertical_16_c(s, p, blimit, limit, thresh, bd);
+  vpx_highbd_lpf_vertical_16_c(s, p, blimit, limit, thresh, bd);
 }

 void wrapper_vertical_16_dual_sse2(uint16_t *s, int p, const uint8_t *blimit,
                                   const uint8_t *limit, const uint8_t *thresh,
                                   int count, int bd) {
-  vp9_highbd_lpf_vertical_16_dual_sse2(s, p, blimit, limit, thresh, bd);
+  vpx_highbd_lpf_vertical_16_dual_sse2(s, p, blimit, limit, thresh, bd);
 }

 void wrapper_vertical_16_dual_c(uint16_t *s, int p, const uint8_t *blimit,
                                const uint8_t *limit, const uint8_t *thresh,
                                int count, int bd) {
-  vp9_highbd_lpf_vertical_16_dual_c(s, p, blimit, limit, thresh, bd);
+  vpx_highbd_lpf_vertical_16_dual_c(s, p, blimit, limit, thresh, bd);
 }
 #else
 void wrapper_vertical_16_sse2(uint8_t *s, int p, const uint8_t *blimit,
                              const uint8_t *limit, const uint8_t *thresh,
                              int count) {
-  vp9_lpf_vertical_16_sse2(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_sse2(s, p, blimit, limit, thresh);
 }

 void wrapper_vertical_16_c(uint8_t *s, int p, const uint8_t *blimit,
                           const uint8_t *limit, const uint8_t *thresh,
                           int count) {
-  vp9_lpf_vertical_16_c(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_c(s, p, blimit, limit, thresh);
 }

 void wrapper_vertical_16_dual_sse2(uint8_t *s, int p, const uint8_t *blimit,
                                   const uint8_t *limit, const uint8_t *thresh,
                                   int count) {
-  vp9_lpf_vertical_16_dual_sse2(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_dual_sse2(s, p, blimit, limit, thresh);
 }

 void wrapper_vertical_16_dual_c(uint8_t *s, int p, const uint8_t *blimit,
                                const uint8_t *limit, const uint8_t *thresh,
                                int count) {
-  vp9_lpf_vertical_16_dual_c(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_dual_c(s, p, blimit, limit, thresh);
 }
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif  // HAVE_SSE2
@@ -114,29 +114,43 @@ void wrapper_vertical_16_dual_c(uint8_t *s, int p, const uint8_t *blimit,
 void wrapper_vertical_16_neon(uint8_t *s, int p, const uint8_t *blimit,
                              const uint8_t *limit, const uint8_t *thresh,
                              int count) {
-  vp9_lpf_vertical_16_neon(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_neon(s, p, blimit, limit, thresh);
 }

 void wrapper_vertical_16_c(uint8_t *s, int p, const uint8_t *blimit,
                           const uint8_t *limit, const uint8_t *thresh,
                           int count) {
-  vp9_lpf_vertical_16_c(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_c(s, p, blimit, limit, thresh);
 }

 void wrapper_vertical_16_dual_neon(uint8_t *s, int p, const uint8_t *blimit,
                                   const uint8_t *limit, const uint8_t *thresh,
                                   int count) {
-  vp9_lpf_vertical_16_dual_neon(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_dual_neon(s, p, blimit, limit, thresh);
 }

 void wrapper_vertical_16_dual_c(uint8_t *s, int p, const uint8_t *blimit,
                                const uint8_t *limit, const uint8_t *thresh,
                                int count) {
-  vp9_lpf_vertical_16_dual_c(s, p, blimit, limit, thresh);
+  vpx_lpf_vertical_16_dual_c(s, p, blimit, limit, thresh);
 }
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif  // HAVE_NEON_ASM

+#if HAVE_MSA && (!CONFIG_VP9_HIGHBITDEPTH)
+void wrapper_vertical_16_msa(uint8_t *s, int p, const uint8_t *blimit,
+                             const uint8_t *limit, const uint8_t *thresh,
+                             int count) {
+  vpx_lpf_vertical_16_msa(s, p, blimit, limit, thresh);
+}
+
+void wrapper_vertical_16_c(uint8_t *s, int p, const uint8_t *blimit,
+                           const uint8_t *limit, const uint8_t *thresh,
+                           int count) {
+  vpx_lpf_vertical_16_c(s, p, blimit, limit, thresh);
+}
+#endif  // HAVE_MSA && (!CONFIG_VP9_HIGHBITDEPTH)
+
 class Loop8Test6Param : public ::testing::TestWithParam<loop8_param_t> {
 public:
  virtual ~Loop8Test6Param() {}
@@ -144,6 +158,7 @@ class Loop8Test6Param : public ::testing::TestWithParam<loop8_param_t> {
    loopfilter_op_ = GET_PARAM(0);
    ref_loopfilter_op_ = GET_PARAM(1);
    bit_depth_ = GET_PARAM(2);
+    count_ = GET_PARAM(3);
    mask_ = (1 << bit_depth_) - 1;
  }

@@ -151,6 +166,7 @@ class Loop8Test6Param : public ::testing::TestWithParam<loop8_param_t> {

 protected:
  int bit_depth_;
+  int count_;
  int mask_;
  loop_op_t loopfilter_op_;
  loop_op_t ref_loopfilter_op_;
@@ -180,11 +196,11 @@ TEST_P(Loop8Test6Param, OperationCheck) {
  const int count_test_block = number_of_iterations;
 #if CONFIG_VP9_HIGHBITDEPTH
  int32_t bd = bit_depth_;
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(16, uint16_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint16_t, ref_s[kNumCoeffs]);
 #else
-  DECLARE_ALIGNED_ARRAY(8, uint8_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(8, uint8_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(8, uint8_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(8, uint8_t, ref_s[kNumCoeffs]);
 #endif  // CONFIG_VP9_HIGHBITDEPTH
  int err_count_total = 0;
  int first_failure = -1;
@@ -206,7 +222,6 @@ TEST_P(Loop8Test6Param, OperationCheck) {
        tmp, tmp, tmp, tmp, tmp, tmp, tmp, tmp
    };
    int32_t p = kNumCoeffs/32;
-    int count = 1;

    uint16_t tmp_s[kNumCoeffs];
    int j = 0;
@@ -238,13 +253,13 @@ TEST_P(Loop8Test6Param, OperationCheck) {
      ref_s[j] = s[j];
    }
 #if CONFIG_VP9_HIGHBITDEPTH
-    ref_loopfilter_op_(ref_s + 8 + p * 8, p, blimit, limit, thresh, count, bd);
+    ref_loopfilter_op_(ref_s + 8 + p * 8, p, blimit, limit, thresh, count_, bd);
    ASM_REGISTER_STATE_CHECK(
-        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count, bd));
+        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count_, bd));
 #else
-    ref_loopfilter_op_(ref_s+8+p*8, p, blimit, limit, thresh, count);
+    ref_loopfilter_op_(ref_s+8+p*8, p, blimit, limit, thresh, count_);
    ASM_REGISTER_STATE_CHECK(
-        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count));
+        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count_));
 #endif  // CONFIG_VP9_HIGHBITDEPTH

    for (int j = 0; j < kNumCoeffs; ++j) {
@@ -266,11 +281,11 @@ TEST_P(Loop8Test6Param, ValueCheck) {
  const int count_test_block = number_of_iterations;
 #if CONFIG_VP9_HIGHBITDEPTH
  const int32_t bd = bit_depth_;
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(16, uint16_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint16_t, ref_s[kNumCoeffs]);
 #else
-  DECLARE_ALIGNED_ARRAY(8, uint8_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(8, uint8_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(8, uint8_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(8, uint8_t, ref_s[kNumCoeffs]);
 #endif  // CONFIG_VP9_HIGHBITDEPTH
  int err_count_total = 0;
  int first_failure = -1;
@@ -279,8 +294,8 @@ TEST_P(Loop8Test6Param, ValueCheck) {
  // function of sharpness_lvl and the loopfilter lvl as:
  // block_inside_limit = lvl >> ((sharpness_lvl > 0) + (sharpness_lvl > 4));
  // ...
-  // vpx_memset(lfi->lfthr[lvl].mblim, (2 * (lvl + 2) + block_inside_limit),
-  //            SIMD_WIDTH);
+  // memset(lfi->lfthr[lvl].mblim, (2 * (lvl + 2) + block_inside_limit),
+  //        SIMD_WIDTH);
  // This means that the largest value for mblim will occur when sharpness_lvl
  // is equal to 0, and lvl is equal to its greatest value (MAX_LOOP_FILTER).
  // In this case block_inside_limit will be equal to MAX_LOOP_FILTER and
@@ -305,19 +320,18 @@ TEST_P(Loop8Test6Param, ValueCheck) {
        tmp, tmp, tmp, tmp, tmp, tmp, tmp, tmp
    };
    int32_t p = kNumCoeffs / 32;
-    int count = 1;
    for (int j = 0; j < kNumCoeffs; ++j) {
      s[j] = rnd.Rand16() & mask_;
      ref_s[j] = s[j];
    }
 #if CONFIG_VP9_HIGHBITDEPTH
-    ref_loopfilter_op_(ref_s + 8 + p * 8, p, blimit, limit, thresh, count, bd);
+    ref_loopfilter_op_(ref_s + 8 + p * 8, p, blimit, limit, thresh, count_, bd);
    ASM_REGISTER_STATE_CHECK(
-        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count, bd));
+        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count_, bd));
 #else
-    ref_loopfilter_op_(ref_s+8+p*8, p, blimit, limit, thresh, count);
+    ref_loopfilter_op_(ref_s+8+p*8, p, blimit, limit, thresh, count_);
    ASM_REGISTER_STATE_CHECK(
-        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count));
+        loopfilter_op_(s + 8 + p * 8, p, blimit, limit, thresh, count_));
 #endif  // CONFIG_VP9_HIGHBITDEPTH
    for (int j = 0; j < kNumCoeffs; ++j) {
      err_count += ref_s[j] != s[j];
@@ -338,11 +352,11 @@ TEST_P(Loop8Test9Param, OperationCheck) {
  const int count_test_block = number_of_iterations;
 #if CONFIG_VP9_HIGHBITDEPTH
  const int32_t bd = bit_depth_;
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(16, uint16_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint16_t, ref_s[kNumCoeffs]);
 #else
-  DECLARE_ALIGNED_ARRAY(8,  uint8_t,  s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(8,  uint8_t,  ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(8,  uint8_t,  s[kNumCoeffs]);
+  DECLARE_ALIGNED(8,  uint8_t,  ref_s[kNumCoeffs]);
 #endif  // CONFIG_VP9_HIGHBITDEPTH
  int err_count_total = 0;
  int first_failure = -1;
@@ -440,11 +454,11 @@ TEST_P(Loop8Test9Param, ValueCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
  const int count_test_block = number_of_iterations;
 #if CONFIG_VP9_HIGHBITDEPTH
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(16, uint16_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(16, uint16_t, ref_s[kNumCoeffs]);
 #else
-  DECLARE_ALIGNED_ARRAY(8,  uint8_t, s, kNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(8,  uint8_t, ref_s, kNumCoeffs);
+  DECLARE_ALIGNED(8,  uint8_t, s[kNumCoeffs]);
+  DECLARE_ALIGNED(8,  uint8_t, ref_s[kNumCoeffs]);
 #endif  // CONFIG_VP9_HIGHBITDEPTH
  int err_count_total = 0;
  int first_failure = -1;
@@ -520,56 +534,65 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    SSE2, Loop8Test6Param,
    ::testing::Values(
-        make_tuple(&vp9_highbd_lpf_horizontal_4_sse2,
-                   &vp9_highbd_lpf_horizontal_4_c, 8),
-        make_tuple(&vp9_highbd_lpf_vertical_4_sse2,
-                   &vp9_highbd_lpf_vertical_4_c, 8),
-        make_tuple(&vp9_highbd_lpf_horizontal_8_sse2,
-                   &vp9_highbd_lpf_horizontal_8_c, 8),
-        make_tuple(&vp9_highbd_lpf_horizontal_16_sse2,
-                   &vp9_highbd_lpf_horizontal_16_c, 8),
-        make_tuple(&vp9_highbd_lpf_vertical_8_sse2,
-                   &vp9_highbd_lpf_vertical_8_c, 8),
+        make_tuple(&vpx_highbd_lpf_horizontal_4_sse2,
+                   &vpx_highbd_lpf_horizontal_4_c, 8, 1),
+        make_tuple(&vpx_highbd_lpf_vertical_4_sse2,
+                   &vpx_highbd_lpf_vertical_4_c, 8, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_8_sse2,
+                   &vpx_highbd_lpf_horizontal_8_c, 8, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+                   &vpx_highbd_lpf_horizontal_16_c, 8, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+                   &vpx_highbd_lpf_horizontal_16_c, 8, 2),
+        make_tuple(&vpx_highbd_lpf_vertical_8_sse2,
+                   &vpx_highbd_lpf_vertical_8_c, 8, 1),
        make_tuple(&wrapper_vertical_16_sse2,
-                   &wrapper_vertical_16_c, 8),
-        make_tuple(&vp9_highbd_lpf_horizontal_4_sse2,
-                   &vp9_highbd_lpf_horizontal_4_c, 10),
-        make_tuple(&vp9_highbd_lpf_vertical_4_sse2,
-                   &vp9_highbd_lpf_vertical_4_c, 10),
-        make_tuple(&vp9_highbd_lpf_horizontal_8_sse2,
-                   &vp9_highbd_lpf_horizontal_8_c, 10),
-        make_tuple(&vp9_highbd_lpf_horizontal_16_sse2,
-                   &vp9_highbd_lpf_horizontal_16_c, 10),
-        make_tuple(&vp9_highbd_lpf_vertical_8_sse2,
-                   &vp9_highbd_lpf_vertical_8_c, 10),
+                   &wrapper_vertical_16_c, 8, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_4_sse2,
+                   &vpx_highbd_lpf_horizontal_4_c, 10, 1),
+        make_tuple(&vpx_highbd_lpf_vertical_4_sse2,
+                   &vpx_highbd_lpf_vertical_4_c, 10, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_8_sse2,
+                   &vpx_highbd_lpf_horizontal_8_c, 10, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+                   &vpx_highbd_lpf_horizontal_16_c, 10, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+                   &vpx_highbd_lpf_horizontal_16_c, 10, 2),
+        make_tuple(&vpx_highbd_lpf_vertical_8_sse2,
+                   &vpx_highbd_lpf_vertical_8_c, 10, 1),
        make_tuple(&wrapper_vertical_16_sse2,
-                   &wrapper_vertical_16_c, 10),
-        make_tuple(&vp9_highbd_lpf_horizontal_4_sse2,
-                   &vp9_highbd_lpf_horizontal_4_c, 12),
-        make_tuple(&vp9_highbd_lpf_vertical_4_sse2,
-                   &vp9_highbd_lpf_vertical_4_c, 12),
-        make_tuple(&vp9_highbd_lpf_horizontal_8_sse2,
-                   &vp9_highbd_lpf_horizontal_8_c, 12),
-        make_tuple(&vp9_highbd_lpf_horizontal_16_sse2,
-                   &vp9_highbd_lpf_horizontal_16_c, 12),
-        make_tuple(&vp9_highbd_lpf_vertical_8_sse2,
-                   &vp9_highbd_lpf_vertical_8_c, 12),
+                   &wrapper_vertical_16_c, 10, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_4_sse2,
+                   &vpx_highbd_lpf_horizontal_4_c, 12, 1),
+        make_tuple(&vpx_highbd_lpf_vertical_4_sse2,
+                   &vpx_highbd_lpf_vertical_4_c, 12, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_8_sse2,
+                   &vpx_highbd_lpf_horizontal_8_c, 12, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+                   &vpx_highbd_lpf_horizontal_16_c, 12, 1),
+        make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+                   &vpx_highbd_lpf_horizontal_16_c, 12, 2),
+        make_tuple(&vpx_highbd_lpf_vertical_8_sse2,
+                   &vpx_highbd_lpf_vertical_8_c, 12, 1),
        make_tuple(&wrapper_vertical_16_sse2,
-                   &wrapper_vertical_16_c, 12),
+                   &wrapper_vertical_16_c, 12, 1),
        make_tuple(&wrapper_vertical_16_dual_sse2,
-                   &wrapper_vertical_16_dual_c, 8),
+                   &wrapper_vertical_16_dual_c, 8, 1),
        make_tuple(&wrapper_vertical_16_dual_sse2,
-                   &wrapper_vertical_16_dual_c, 10),
+                   &wrapper_vertical_16_dual_c, 10, 1),
        make_tuple(&wrapper_vertical_16_dual_sse2,
-                   &wrapper_vertical_16_dual_c, 12)));
+                   &wrapper_vertical_16_dual_c, 12, 1)));
 #else
 INSTANTIATE_TEST_CASE_P(
    SSE2, Loop8Test6Param,
    ::testing::Values(
-        make_tuple(&vp9_lpf_horizontal_8_sse2, &vp9_lpf_horizontal_8_c, 8),
-        make_tuple(&vp9_lpf_horizontal_16_sse2, &vp9_lpf_horizontal_16_c, 8),
-        make_tuple(&vp9_lpf_vertical_8_sse2, &vp9_lpf_vertical_8_c, 8),
-        make_tuple(&wrapper_vertical_16_sse2, &wrapper_vertical_16_c, 8)));
+        make_tuple(&vpx_lpf_horizontal_8_sse2, &vpx_lpf_horizontal_8_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_16_sse2, &vpx_lpf_horizontal_16_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_16_sse2, &vpx_lpf_horizontal_16_c, 8, 2),
+        make_tuple(&vpx_lpf_vertical_8_sse2, &vpx_lpf_vertical_8_c, 8, 1),
+        make_tuple(&wrapper_vertical_16_sse2, &wrapper_vertical_16_c, 8, 1),
+        make_tuple(&wrapper_vertical_16_dual_sse2,
+                   &wrapper_vertical_16_dual_c, 8, 1)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif

@@ -577,7 +600,9 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    AVX2, Loop8Test6Param,
    ::testing::Values(
-        make_tuple(&vp9_lpf_horizontal_16_avx2, &vp9_lpf_horizontal_16_c, 8)));
+        make_tuple(&vpx_lpf_horizontal_16_avx2, &vpx_lpf_horizontal_16_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_16_avx2, &vpx_lpf_horizontal_16_c, 8,
+                   2)));
 #endif

 #if HAVE_SSE2
@@ -585,42 +610,42 @@ INSTANTIATE_TEST_CASE_P(
 INSTANTIATE_TEST_CASE_P(
    SSE2, Loop8Test9Param,
    ::testing::Values(
-        make_tuple(&vp9_highbd_lpf_horizontal_4_dual_sse2,
-                   &vp9_highbd_lpf_horizontal_4_dual_c, 8),
-        make_tuple(&vp9_highbd_lpf_horizontal_8_dual_sse2,
-                   &vp9_highbd_lpf_horizontal_8_dual_c, 8),
-        make_tuple(&vp9_highbd_lpf_vertical_4_dual_sse2,
-                   &vp9_highbd_lpf_vertical_4_dual_c, 8),
-        make_tuple(&vp9_highbd_lpf_vertical_8_dual_sse2,
-                   &vp9_highbd_lpf_vertical_8_dual_c, 8),
-        make_tuple(&vp9_highbd_lpf_horizontal_4_dual_sse2,
-                   &vp9_highbd_lpf_horizontal_4_dual_c, 10),
-        make_tuple(&vp9_highbd_lpf_horizontal_8_dual_sse2,
-                   &vp9_highbd_lpf_horizontal_8_dual_c, 10),
-        make_tuple(&vp9_highbd_lpf_vertical_4_dual_sse2,
-                   &vp9_highbd_lpf_vertical_4_dual_c, 10),
-        make_tuple(&vp9_highbd_lpf_vertical_8_dual_sse2,
-                   &vp9_highbd_lpf_vertical_8_dual_c, 10),
-        make_tuple(&vp9_highbd_lpf_horizontal_4_dual_sse2,
-                   &vp9_highbd_lpf_horizontal_4_dual_c, 12),
-        make_tuple(&vp9_highbd_lpf_horizontal_8_dual_sse2,
-                   &vp9_highbd_lpf_horizontal_8_dual_c, 12),
-        make_tuple(&vp9_highbd_lpf_vertical_4_dual_sse2,
-                   &vp9_highbd_lpf_vertical_4_dual_c, 12),
-        make_tuple(&vp9_highbd_lpf_vertical_8_dual_sse2,
-                   &vp9_highbd_lpf_vertical_8_dual_c, 12)));
+        make_tuple(&vpx_highbd_lpf_horizontal_4_dual_sse2,
+                   &vpx_highbd_lpf_horizontal_4_dual_c, 8),
+        make_tuple(&vpx_highbd_lpf_horizontal_8_dual_sse2,
+                   &vpx_highbd_lpf_horizontal_8_dual_c, 8),
+        make_tuple(&vpx_highbd_lpf_vertical_4_dual_sse2,
+                   &vpx_highbd_lpf_vertical_4_dual_c, 8),
+        make_tuple(&vpx_highbd_lpf_vertical_8_dual_sse2,
+                   &vpx_highbd_lpf_vertical_8_dual_c, 8),
+        make_tuple(&vpx_highbd_lpf_horizontal_4_dual_sse2,
+                   &vpx_highbd_lpf_horizontal_4_dual_c, 10),
+        make_tuple(&vpx_highbd_lpf_horizontal_8_dual_sse2,
+                   &vpx_highbd_lpf_horizontal_8_dual_c, 10),
+        make_tuple(&vpx_highbd_lpf_vertical_4_dual_sse2,
+                   &vpx_highbd_lpf_vertical_4_dual_c, 10),
+        make_tuple(&vpx_highbd_lpf_vertical_8_dual_sse2,
+                   &vpx_highbd_lpf_vertical_8_dual_c, 10),
+        make_tuple(&vpx_highbd_lpf_horizontal_4_dual_sse2,
+                   &vpx_highbd_lpf_horizontal_4_dual_c, 12),
+        make_tuple(&vpx_highbd_lpf_horizontal_8_dual_sse2,
+                   &vpx_highbd_lpf_horizontal_8_dual_c, 12),
+        make_tuple(&vpx_highbd_lpf_vertical_4_dual_sse2,
+                   &vpx_highbd_lpf_vertical_4_dual_c, 12),
+        make_tuple(&vpx_highbd_lpf_vertical_8_dual_sse2,
+                   &vpx_highbd_lpf_vertical_8_dual_c, 12)));
 #else
 INSTANTIATE_TEST_CASE_P(
    SSE2, Loop8Test9Param,
    ::testing::Values(
-        make_tuple(&vp9_lpf_horizontal_4_dual_sse2,
-                   &vp9_lpf_horizontal_4_dual_c, 8),
-        make_tuple(&vp9_lpf_horizontal_8_dual_sse2,
-                   &vp9_lpf_horizontal_8_dual_c, 8),
-        make_tuple(&vp9_lpf_vertical_4_dual_sse2,
-                   &vp9_lpf_vertical_4_dual_c, 8),
-        make_tuple(&vp9_lpf_vertical_8_dual_sse2,
-                   &vp9_lpf_vertical_8_dual_c, 8)));
+        make_tuple(&vpx_lpf_horizontal_4_dual_sse2,
+                   &vpx_lpf_horizontal_4_dual_c, 8),
+        make_tuple(&vpx_lpf_horizontal_8_dual_sse2,
+                   &vpx_lpf_horizontal_8_dual_c, 8),
+        make_tuple(&vpx_lpf_vertical_4_dual_sse2,
+                   &vpx_lpf_vertical_4_dual_c, 8),
+        make_tuple(&vpx_lpf_vertical_8_dual_sse2,
+                   &vpx_lpf_vertical_8_dual_c, 8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif

@@ -634,35 +659,60 @@ INSTANTIATE_TEST_CASE_P(
 #if HAVE_NEON_ASM
 // Using #if inside the macro is unsupported on MSVS but the tests are not
 // currently built for MSVS with ARM and NEON.
-        make_tuple(&vp9_lpf_horizontal_16_neon,
-                   &vp9_lpf_horizontal_16_c, 8),
+        make_tuple(&vpx_lpf_horizontal_16_neon,
+                   &vpx_lpf_horizontal_16_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_16_neon,
+                   &vpx_lpf_horizontal_16_c, 8, 2),
        make_tuple(&wrapper_vertical_16_neon,
-                   &wrapper_vertical_16_c, 8),
+                   &wrapper_vertical_16_c, 8, 1),
        make_tuple(&wrapper_vertical_16_dual_neon,
-                   &wrapper_vertical_16_dual_c, 8),
-        make_tuple(&vp9_lpf_horizontal_8_neon,
-                   &vp9_lpf_horizontal_8_c, 8),
-        make_tuple(&vp9_lpf_vertical_8_neon,
-                   &vp9_lpf_vertical_8_c, 8),
+                   &wrapper_vertical_16_dual_c, 8, 1),
 #endif  // HAVE_NEON_ASM
-        make_tuple(&vp9_lpf_horizontal_4_neon,
-                   &vp9_lpf_horizontal_4_c, 8),
-        make_tuple(&vp9_lpf_vertical_4_neon,
-                   &vp9_lpf_vertical_4_c, 8)));
+        make_tuple(&vpx_lpf_horizontal_8_neon,
+                   &vpx_lpf_horizontal_8_c, 8, 1),
+        make_tuple(&vpx_lpf_vertical_8_neon,
+                   &vpx_lpf_vertical_8_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_4_neon,
+                   &vpx_lpf_horizontal_4_c, 8, 1),
+        make_tuple(&vpx_lpf_vertical_4_neon,
+                   &vpx_lpf_vertical_4_c, 8, 1)));
 INSTANTIATE_TEST_CASE_P(
    NEON, Loop8Test9Param,
    ::testing::Values(
 #if HAVE_NEON_ASM
-        make_tuple(&vp9_lpf_horizontal_8_dual_neon,
-                   &vp9_lpf_horizontal_8_dual_c, 8),
-        make_tuple(&vp9_lpf_vertical_8_dual_neon,
-                   &vp9_lpf_vertical_8_dual_c, 8),
+        make_tuple(&vpx_lpf_horizontal_8_dual_neon,
+                   &vpx_lpf_horizontal_8_dual_c, 8),
+        make_tuple(&vpx_lpf_vertical_8_dual_neon,
+                   &vpx_lpf_vertical_8_dual_c, 8),
 #endif  // HAVE_NEON_ASM
-        make_tuple(&vp9_lpf_horizontal_4_dual_neon,
-                   &vp9_lpf_horizontal_4_dual_c, 8),
-        make_tuple(&vp9_lpf_vertical_4_dual_neon,
-                   &vp9_lpf_vertical_4_dual_c, 8)));
+        make_tuple(&vpx_lpf_horizontal_4_dual_neon,
+                   &vpx_lpf_horizontal_4_dual_c, 8),
+        make_tuple(&vpx_lpf_vertical_4_dual_neon,
+                   &vpx_lpf_vertical_4_dual_c, 8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif  // HAVE_NEON

+#if HAVE_MSA && (!CONFIG_VP9_HIGHBITDEPTH)
+INSTANTIATE_TEST_CASE_P(
+    MSA, Loop8Test6Param,
+    ::testing::Values(
+        make_tuple(&vpx_lpf_horizontal_8_msa, &vpx_lpf_horizontal_8_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_16_msa, &vpx_lpf_horizontal_16_c, 8, 1),
+        make_tuple(&vpx_lpf_horizontal_16_msa, &vpx_lpf_horizontal_16_c, 8, 2),
+        make_tuple(&vpx_lpf_vertical_8_msa, &vpx_lpf_vertical_8_c, 8, 1),
+        make_tuple(&wrapper_vertical_16_msa, &wrapper_vertical_16_c, 8, 1)));
+
+INSTANTIATE_TEST_CASE_P(
+    MSA, Loop8Test9Param,
+    ::testing::Values(
+        make_tuple(&vpx_lpf_horizontal_4_dual_msa,
+                   &vpx_lpf_horizontal_4_dual_c, 8),
+        make_tuple(&vpx_lpf_horizontal_8_dual_msa,
+                   &vpx_lpf_horizontal_8_dual_c, 8),
+        make_tuple(&vpx_lpf_vertical_4_dual_msa,
+                   &vpx_lpf_vertical_4_dual_c, 8),
+        make_tuple(&vpx_lpf_vertical_8_dual_msa,
+                   &vpx_lpf_vertical_8_dual_c, 8)));
+#endif  // HAVE_MSA && (!CONFIG_VP9_HIGHBITDEPTH)
+
 }  // namespace
--- a/test/md5_helper.h
+++ b/test/md5_helper.h
@@ -42,6 +42,10 @@ class MD5 {
    }
  }

+  void Add(const uint8_t *data, size_t size) {
+    MD5Update(&md5_, data, static_cast<uint32_t>(size));
+  }
+
  const char *Get(void) {
    static const char hex[16] = {
      '0', '1', '2', '3', '4', '5', '6', '7',
--- a/test/partial_idct_test.cc
+++ b/test/partial_idct_test.cc
@@ -13,12 +13,13 @@
 #include <string.h>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_blockd.h"
 #include "vp9/common/vp9_scan.h"
 #include "vpx/vpx_integer.h"
@@ -74,16 +75,16 @@ TEST_P(PartialIDctTest, RunQuantCheck) {
      FAIL() << "Wrong Size!";
      break;
  }
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_coef_block1, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_coef_block2, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, dst1, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, dst2, kMaxNumCoeffs);
+  DECLARE_ALIGNED(16, tran_low_t, test_coef_block1[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, test_coef_block2[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, dst1[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, dst2[kMaxNumCoeffs]);

  const int count_test_block = 1000;
  const int block_size = size * size;

-  DECLARE_ALIGNED_ARRAY(16, int16_t, input_extreme_block, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, output_ref_block, kMaxNumCoeffs);
+  DECLARE_ALIGNED(16, int16_t, input_extreme_block[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, output_ref_block[kMaxNumCoeffs]);

  int max_error = 0;
  for (int i = 0; i < count_test_block; ++i) {
@@ -153,10 +154,10 @@ TEST_P(PartialIDctTest, ResultsMatch) {
      FAIL() << "Wrong Size!";
      break;
  }
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_coef_block1, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, test_coef_block2, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, dst1, kMaxNumCoeffs);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, dst2, kMaxNumCoeffs);
+  DECLARE_ALIGNED(16, tran_low_t, test_coef_block1[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, tran_low_t, test_coef_block2[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, dst1[kMaxNumCoeffs]);
+  DECLARE_ALIGNED(16, uint8_t, dst2[kMaxNumCoeffs]);
  const int count_test_block = 1000;
  const int max_coeff = 32766 / 4;
  const int block_size = size * size;
@@ -201,117 +202,142 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    C, PartialIDctTest,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c,
-                   &vp9_idct32x32_34_add_c,
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_34_add_c,
                   TX_32X32, 34),
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c,
-                   &vp9_idct32x32_1_add_c,
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_1_add_c,
                   TX_32X32, 1),
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_10_add_c,
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_10_add_c,
                   TX_16X16, 10),
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_1_add_c,
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_1_add_c,
                   TX_16X16, 1),
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_12_add_c,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_12_add_c,
                   TX_8X8, 12),
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_1_add_c,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_1_add_c,
                   TX_8X8, 1),
-        make_tuple(&vp9_fdct4x4_c,
-                   &vp9_idct4x4_16_add_c,
-                   &vp9_idct4x4_1_add_c,
+        make_tuple(&vpx_fdct4x4_c,
+                   &vpx_idct4x4_16_add_c,
+                   &vpx_idct4x4_1_add_c,
                   TX_4X4, 1)));

-#if HAVE_NEON
+#if HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    NEON, PartialIDctTest,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c,
-                   &vp9_idct32x32_1_add_neon,
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_1_add_neon,
                   TX_32X32, 1),
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_10_add_neon,
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_10_add_neon,
                   TX_16X16, 10),
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_1_add_neon,
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_1_add_neon,
                   TX_16X16, 1),
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_12_add_neon,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_12_add_neon,
                   TX_8X8, 12),
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_1_add_neon,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_1_add_neon,
                   TX_8X8, 1),
-        make_tuple(&vp9_fdct4x4_c,
-                   &vp9_idct4x4_16_add_c,
-                   &vp9_idct4x4_1_add_neon,
+        make_tuple(&vpx_fdct4x4_c,
+                   &vpx_idct4x4_16_add_c,
+                   &vpx_idct4x4_1_add_neon,
                   TX_4X4, 1)));
-#endif  // HAVE_NEON
+#endif  // HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE

 #if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSE2, PartialIDctTest,
    ::testing::Values(
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c,
-                   &vp9_idct32x32_34_add_sse2,
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_34_add_sse2,
                   TX_32X32, 34),
-        make_tuple(&vp9_fdct32x32_c,
-                   &vp9_idct32x32_1024_add_c,
-                   &vp9_idct32x32_1_add_sse2,
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_1_add_sse2,
                   TX_32X32, 1),
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_10_add_sse2,
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_10_add_sse2,
                   TX_16X16, 10),
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_1_add_sse2,
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_1_add_sse2,
                   TX_16X16, 1),
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_12_add_sse2,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_12_add_sse2,
                   TX_8X8, 12),
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_1_add_sse2,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_1_add_sse2,
                   TX_8X8, 1),
-        make_tuple(&vp9_fdct4x4_c,
-                   &vp9_idct4x4_16_add_c,
-                   &vp9_idct4x4_1_add_sse2,
+        make_tuple(&vpx_fdct4x4_c,
+                   &vpx_idct4x4_16_add_c,
+                   &vpx_idct4x4_1_add_sse2,
                   TX_4X4, 1)));
 #endif

-#if HAVE_SSSE3 && ARCH_X86_64 && !CONFIG_VP9_HIGHBITDEPTH && \
-    !CONFIG_EMULATE_HARDWARE
+#if HAVE_SSSE3 && CONFIG_USE_X86INC && ARCH_X86_64 && \
+    !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
    SSSE3_64, PartialIDctTest,
    ::testing::Values(
-        make_tuple(&vp9_fdct8x8_c,
-                   &vp9_idct8x8_64_add_c,
-                   &vp9_idct8x8_12_add_ssse3,
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_12_add_ssse3,
                   TX_8X8, 12)));
 #endif

-#if HAVE_SSSE3 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#if HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
-    SSSE3, PartialIDctTest,
+    MSA, PartialIDctTest,
    ::testing::Values(
-        make_tuple(&vp9_fdct16x16_c,
-                   &vp9_idct16x16_256_add_c,
-                   &vp9_idct16x16_10_add_ssse3,
-                   TX_16X16, 10)));
-#endif
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_34_add_msa,
+                   TX_32X32, 34),
+        make_tuple(&vpx_fdct32x32_c,
+                   &vpx_idct32x32_1024_add_c,
+                   &vpx_idct32x32_1_add_msa,
+                   TX_32X32, 1),
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_10_add_msa,
+                   TX_16X16, 10),
+        make_tuple(&vpx_fdct16x16_c,
+                   &vpx_idct16x16_256_add_c,
+                   &vpx_idct16x16_1_add_msa,
+                   TX_16X16, 1),
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_12_add_msa,
+                   TX_8X8, 10),
+        make_tuple(&vpx_fdct8x8_c,
+                   &vpx_idct8x8_64_add_c,
+                   &vpx_idct8x8_1_add_msa,
+                   TX_8X8, 1),
+        make_tuple(&vpx_fdct4x4_c,
+                   &vpx_idct4x4_16_add_c,
+                   &vpx_idct4x4_1_add_msa,
+                   TX_4X4, 1)));
+#endif  // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+
 }  // namespace
--- a/test/pp_filter_test.cc
+++ b/test/pp_filter_test.cc
@@ -63,12 +63,12 @@ TEST_P(VP8PostProcessingFilterTest, FilterOutputCheck) {
  uint8_t *const dst_image_ptr = dst_image + 8;
  uint8_t *const flimits =
      reinterpret_cast<uint8_t *>(vpx_memalign(16, block_width));
-  (void)vpx_memset(flimits, 255, block_width);
+  (void)memset(flimits, 255, block_width);

  // Initialize pixels in the input:
  //   block pixels to value 1,
  //   border pixels to value 10.
-  (void)vpx_memset(src_image, 10, input_size);
+  (void)memset(src_image, 10, input_size);
  uint8_t *pixel_ptr = src_image_ptr;
  for (int i = 0; i < block_height; ++i) {
    for (int j = 0; j < block_width; ++j) {
@@ -78,7 +78,7 @@ TEST_P(VP8PostProcessingFilterTest, FilterOutputCheck) {
  }

  // Initialize pixels in the output to 99.
-  (void)vpx_memset(dst_image, 99, output_size);
+  (void)memset(dst_image, 99, output_size);

  ASM_REGISTER_STATE_CHECK(
      GetParam()(src_image_ptr, dst_image_ptr, input_stride,
@@ -110,4 +110,9 @@ INSTANTIATE_TEST_CASE_P(SSE2, VP8PostProcessingFilterTest,
    ::testing::Values(vp8_post_proc_down_and_across_mb_row_sse2));
 #endif

+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(MSA, VP8PostProcessingFilterTest,
+    ::testing::Values(vp8_post_proc_down_and_across_mb_row_msa));
+#endif
+
 }  // namespace
--- a/test/quantize_test.cc
+++ b/test/quantize_test.cc
@@ -11,13 +11,13 @@
 #include <string.h>

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_config.h"
+#include "./vp8_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-
-#include "./vpx_config.h"
-#include "./vp8_rtcd.h"
 #include "vp8/common/blockd.h"
 #include "vp8/common/onyx.h"
 #include "vp8/encoder/block.h"
@@ -56,7 +56,7 @@ class QuantizeTestBase {

    // The full configuration is necessary to generate the quantization tables.
    VP8_CONFIG vp8_config;
-    vpx_memset(&vp8_config, 0, sizeof(vp8_config));
+    memset(&vp8_config, 0, sizeof(vp8_config));

    vp8_comp_ = vp8_create_compressor(&vp8_config);

@@ -69,8 +69,7 @@ class QuantizeTestBase {
    // Copy macroblockd from the reference to get pre-set-up dequant values.
    macroblockd_dst_ = reinterpret_cast<MACROBLOCKD *>(
        vpx_memalign(32, sizeof(*macroblockd_dst_)));
-    vpx_memcpy(macroblockd_dst_, &vp8_comp_->mb.e_mbd,
-               sizeof(*macroblockd_dst_));
+    memcpy(macroblockd_dst_, &vp8_comp_->mb.e_mbd, sizeof(*macroblockd_dst_));
    // Fix block pointers - currently they point to the blocks in the reference
    // structure.
    vp8_setup_block_dptrs(macroblockd_dst_);
@@ -79,8 +78,7 @@ class QuantizeTestBase {
  void UpdateQuantizer(int q) {
    vp8_set_quantizer(vp8_comp_, q);

-    vpx_memcpy(macroblockd_dst_, &vp8_comp_->mb.e_mbd,
-               sizeof(*macroblockd_dst_));
+    memcpy(macroblockd_dst_, &vp8_comp_->mb.e_mbd, sizeof(*macroblockd_dst_));
    vp8_setup_block_dptrs(macroblockd_dst_);
  }

@@ -194,4 +192,12 @@ INSTANTIATE_TEST_CASE_P(NEON, QuantizeTest,
                        ::testing::Values(make_tuple(&vp8_fast_quantize_b_neon,
                                                     &vp8_fast_quantize_b_c)));
 #endif  // HAVE_NEON
+
+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(
+    MSA, QuantizeTest,
+    ::testing::Values(
+        make_tuple(&vp8_fast_quantize_b_msa, &vp8_fast_quantize_b_c),
+        make_tuple(&vp8_regular_quantize_b_msa, &vp8_regular_quantize_b_c)));
+#endif  // HAVE_MSA
 }  // namespace
--- a/test/register_state_check.h
+++ b/test/register_state_check.h
@@ -96,7 +96,7 @@ class RegisterStateCheck {

 extern "C" {
 // Save the d8-d15 registers into store.
-void vp9_push_neon(int64_t *store);
+void vpx_push_neon(int64_t *store);
 }

 namespace libvpx_test {
@@ -111,7 +111,7 @@ class RegisterStateCheck {

 private:
  static bool StoreRegisters(int64_t store[8]) {
-    vp9_push_neon(store);
+    vpx_push_neon(store);
    return true;
  }

@@ -119,7 +119,7 @@ class RegisterStateCheck {
  bool Check() const {
    if (!initialized_) return false;
    int64_t post_store[8];
-    vp9_push_neon(post_store);
+    vpx_push_neon(post_store);
    for (int i = 0; i < 8; ++i) {
      EXPECT_EQ(pre_store_[i], post_store[i]) << "d"
          << i + 8 << " has been modified";
--- a/test/sad_test.cc
+++ b/test/sad_test.cc
--- a/test/set_roi.cc
+++ b/test/set_roi.cc
@@ -53,7 +53,7 @@ TEST(VP8RoiMapTest, ParameterCheck) {
  cpi.common.mb_rows = 240 >> 4;
  cpi.common.mb_cols = 320 >> 4;
  const int mbs = (cpi.common.mb_rows * cpi.common.mb_cols);
-  vpx_memset(cpi.segment_feature_data, 0, sizeof(cpi.segment_feature_data));
+  memset(cpi.segment_feature_data, 0, sizeof(cpi.segment_feature_data));

  // Segment map
  cpi.segmentation_map = reinterpret_cast<unsigned char *>(vpx_calloc(mbs, 1));
@@ -61,9 +61,9 @@ TEST(VP8RoiMapTest, ParameterCheck) {
  // Allocate memory for the source memory map.
  unsigned char *roi_map =
    reinterpret_cast<unsigned char *>(vpx_calloc(mbs, 1));
-  vpx_memset(&roi_map[mbs >> 2], 1, (mbs >> 2));
-  vpx_memset(&roi_map[mbs >> 1], 2, (mbs >> 2));
-  vpx_memset(&roi_map[mbs -(mbs >> 2)], 3, (mbs >> 2));
+  memset(&roi_map[mbs >> 2], 1, (mbs >> 2));
+  memset(&roi_map[mbs >> 1], 2, (mbs >> 2));
+  memset(&roi_map[mbs -(mbs >> 2)], 3, (mbs >> 2));

  // Do a test call with valid parameters.
  int roi_retval = vp8_set_roimap(&cpi, roi_map, cpi.common.mb_rows,
--- a/test/sixtap_predict_test.cc
+++ b/test/sixtap_predict_test.cc
@@ -11,13 +11,15 @@
 #include <math.h>
 #include <stdlib.h>
 #include <string.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_config.h"
+#include "./vp8_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-#include "third_party/googletest/src/include/gtest/gtest.h"
-#include "./vpx_config.h"
-#include "./vp8_rtcd.h"
 #include "vpx/vpx_integer.h"
 #include "vpx_mem/vpx_mem.h"

@@ -238,4 +240,16 @@ INSTANTIATE_TEST_CASE_P(
        make_tuple(8, 4, sixtap_8x4_ssse3),
        make_tuple(4, 4, sixtap_4x4_ssse3)));
 #endif
+#if HAVE_MSA
+const SixtapPredictFunc sixtap_16x16_msa = vp8_sixtap_predict16x16_msa;
+const SixtapPredictFunc sixtap_8x8_msa = vp8_sixtap_predict8x8_msa;
+const SixtapPredictFunc sixtap_8x4_msa = vp8_sixtap_predict8x4_msa;
+const SixtapPredictFunc sixtap_4x4_msa = vp8_sixtap_predict4x4_msa;
+INSTANTIATE_TEST_CASE_P(
+    MSA, SixtapPredictTest, ::testing::Values(
+        make_tuple(16, 16, sixtap_16x16_msa),
+        make_tuple(8, 8, sixtap_8x8_msa),
+        make_tuple(8, 4, sixtap_8x4_msa),
+        make_tuple(4, 4, sixtap_4x4_msa)));
+#endif
 }  // namespace
--- a/test/subtract_test.cc
+++ b/test/subtract_test.cc
@@ -1,123 +0,0 @@
-/*
- *  Copyright (c) 2012 The WebM project authors. All Rights Reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS.  All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "third_party/googletest/src/include/gtest/gtest.h"
-#include "test/acm_random.h"
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
-#include "./vpx_config.h"
-#include "./vp8_rtcd.h"
-#include "vp8/common/blockd.h"
-#include "vp8/encoder/block.h"
-#include "vpx_mem/vpx_mem.h"
-
-typedef void (*SubtractBlockFunc)(BLOCK *be, BLOCKD *bd, int pitch);
-
-namespace {
-
-class SubtractBlockTest : public ::testing::TestWithParam<SubtractBlockFunc> {
- public:
-  virtual void TearDown() {
-    libvpx_test::ClearSystemState();
-  }
-};
-
-using libvpx_test::ACMRandom;
-
-TEST_P(SubtractBlockTest, SimpleSubtract) {
-  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  BLOCK be;
-  BLOCKD bd;
-  // in libvpx, this stride is always 16
-  const int kDiffPredStride = 16;
-  const int kSrcStride[] = {32, 16, 8, 4, 0};
-  const int kBlockWidth = 4;
-  const int kBlockHeight = 4;
-
-  // Allocate... align to 16 for mmx/sse tests
-  uint8_t *source = reinterpret_cast<uint8_t*>(
-      vpx_memalign(16, kBlockHeight * kSrcStride[0] * sizeof(*source)));
-  be.src_diff = reinterpret_cast<int16_t*>(
-      vpx_memalign(16, kBlockHeight * kDiffPredStride * sizeof(*be.src_diff)));
-  bd.predictor = reinterpret_cast<unsigned char*>(
-      vpx_memalign(16, kBlockHeight * kDiffPredStride * sizeof(*bd.predictor)));
-
-  for (int i = 0; kSrcStride[i] > 0; ++i) {
-    // start at block0
-    be.src = 0;
-    be.base_src = &source;
-    be.src_stride = kSrcStride[i];
-
-    // set difference
-    int16_t *src_diff = be.src_diff;
-    for (int r = 0; r < kBlockHeight; ++r) {
-      for (int c = 0; c < kBlockWidth; ++c) {
-        src_diff[c] = static_cast<int16_t>(0xa5a5u);
-      }
-      src_diff += kDiffPredStride;
-    }
-
-    // set destination
-    uint8_t *base_src = *be.base_src;
-    for (int r = 0; r < kBlockHeight; ++r) {
-      for (int c = 0; c < kBlockWidth; ++c) {
-        base_src[c] = rnd.Rand8();
-      }
-      base_src += be.src_stride;
-    }
-
-    // set predictor
-    uint8_t *predictor = bd.predictor;
-    for (int r = 0; r < kBlockHeight; ++r) {
-      for (int c = 0; c < kBlockWidth; ++c) {
-        predictor[c] = rnd.Rand8();
-      }
-      predictor += kDiffPredStride;
-    }
-
-    ASM_REGISTER_STATE_CHECK(GetParam()(&be, &bd, kDiffPredStride));
-
-    base_src = *be.base_src;
-    src_diff = be.src_diff;
-    predictor = bd.predictor;
-    for (int r = 0; r < kBlockHeight; ++r) {
-      for (int c = 0; c < kBlockWidth; ++c) {
-        EXPECT_EQ(base_src[c], (src_diff[c] + predictor[c])) << "r = " << r
-                                                             << ", c = " << c;
-      }
-      src_diff += kDiffPredStride;
-      predictor += kDiffPredStride;
-      base_src += be.src_stride;
-    }
-  }
-  vpx_free(be.src_diff);
-  vpx_free(source);
-  vpx_free(bd.predictor);
-}
-
-INSTANTIATE_TEST_CASE_P(C, SubtractBlockTest,
-                        ::testing::Values(vp8_subtract_b_c));
-
-#if HAVE_NEON
-INSTANTIATE_TEST_CASE_P(NEON, SubtractBlockTest,
-                        ::testing::Values(vp8_subtract_b_neon));
-#endif
-
-#if HAVE_MMX
-INSTANTIATE_TEST_CASE_P(MMX, SubtractBlockTest,
-                        ::testing::Values(vp8_subtract_b_mmx));
-#endif
-
-#if HAVE_SSE2
-INSTANTIATE_TEST_CASE_P(SSE2, SubtractBlockTest,
-                        ::testing::Values(vp8_subtract_b_sse2));
-#endif
-
-}  // namespace
--- a/test/superframe_test.cc
+++ b/test/superframe_test.cc
@@ -94,4 +94,7 @@ TEST_P(SuperframeTest, TestSuperframeIndexIsOptional) {

 VP9_INSTANTIATE_TEST_CASE(SuperframeTest, ::testing::Values(
    ::libvpx_test::kTwoPassGood));
+
+VP10_INSTANTIATE_TEST_CASE(SuperframeTest, ::testing::Values(
+    ::libvpx_test::kTwoPassGood));
 }  // namespace
--- a/test/svc_test.cc
+++ b/test/svc_test.cc
@@ -63,6 +63,9 @@ class SvcTest : public ::testing::Test {
    vpx_codec_dec_cfg_t dec_cfg = vpx_codec_dec_cfg_t();
    VP9CodecFactory codec_factory;
    decoder_ = codec_factory.CreateDecoder(dec_cfg, 0);
+
+    tile_columns_ = 0;
+    tile_rows_ = 0;
  }

  virtual void TearDown() {
@@ -75,6 +78,8 @@ class SvcTest : public ::testing::Test {
        vpx_svc_init(&svc_, &codec_, vpx_codec_vp9_cx(), &codec_enc_);
    EXPECT_EQ(VPX_CODEC_OK, res);
    vpx_codec_control(&codec_, VP8E_SET_CPUUSED, 4);  // Make the test faster
+    vpx_codec_control(&codec_, VP9E_SET_TILE_COLUMNS, tile_columns_);
+    vpx_codec_control(&codec_, VP9E_SET_TILE_ROWS, tile_rows_);
    codec_initialized_ = true;
  }

@@ -108,7 +113,8 @@ class SvcTest : public ::testing::Test {
    codec_enc_.g_pass = VPX_RC_FIRST_PASS;
    InitializeEncoder();

-    libvpx_test::I420VideoSource video(test_file_name_, kWidth, kHeight,
+    libvpx_test::I420VideoSource video(test_file_name_,
+                                       codec_enc_.g_w, codec_enc_.g_h,
                                       codec_enc_.g_timebase.den,
                                       codec_enc_.g_timebase.num, 0, 30);
    video.Begin();
@@ -176,7 +182,8 @@ class SvcTest : public ::testing::Test {
    }
    InitializeEncoder();

-    libvpx_test::I420VideoSource video(test_file_name_, kWidth, kHeight,
+    libvpx_test::I420VideoSource video(test_file_name_,
+                                       codec_enc_.g_w, codec_enc_.g_h,
                                       codec_enc_.g_timebase.den,
                                       codec_enc_.g_timebase.num, 0, 30);
    video.Begin();
@@ -310,6 +317,8 @@ class SvcTest : public ::testing::Test {
  std::string test_file_name_;
  bool codec_initialized_;
  Decoder *decoder_;
+  int tile_columns_;
+  int tile_rows_;
 };

 TEST_F(SvcTest, SvcInit) {
@@ -444,6 +453,7 @@ TEST_F(SvcTest, OnePassEncodeOneFrame) {

 TEST_F(SvcTest, OnePassEncodeThreeFrames) {
  codec_enc_.g_pass = VPX_RC_ONE_PASS;
+  codec_enc_.g_lag_in_frames = 0;
  vpx_fixed_buf outputs[3];
  memset(&outputs[0], 0, sizeof(outputs));
  Pass2EncodeNFrames(NULL, 3, 2, &outputs[0]);
@@ -737,4 +747,51 @@ TEST_F(SvcTest,
  FreeBitstreamBuffers(&outputs[0], 10);
 }

+TEST_F(SvcTest, TwoPassEncode2TemporalLayersWithTiles) {
+  // First pass encode
+  std::string stats_buf;
+  vpx_svc_set_options(&svc_, "scale-factors=1/1");
+  svc_.temporal_layers = 2;
+  Pass1EncodeNFrames(10, 1, &stats_buf);
+
+  // Second pass encode
+  codec_enc_.g_pass = VPX_RC_LAST_PASS;
+  svc_.temporal_layers = 2;
+  vpx_svc_set_options(&svc_, "auto-alt-refs=1 scale-factors=1/1");
+  codec_enc_.g_w = 704;
+  codec_enc_.g_h = 144;
+  tile_columns_ = 1;
+  tile_rows_ = 1;
+  vpx_fixed_buf outputs[10];
+  memset(&outputs[0], 0, sizeof(outputs));
+  Pass2EncodeNFrames(&stats_buf, 10, 1, &outputs[0]);
+  DecodeNFrames(&outputs[0], 10);
+  FreeBitstreamBuffers(&outputs[0], 10);
+}
+
+TEST_F(SvcTest,
+       TwoPassEncode2TemporalLayersWithMultipleFrameContextsAndTiles) {
+  // First pass encode
+  std::string stats_buf;
+  vpx_svc_set_options(&svc_, "scale-factors=1/1");
+  svc_.temporal_layers = 2;
+  Pass1EncodeNFrames(10, 1, &stats_buf);
+
+  // Second pass encode
+  codec_enc_.g_pass = VPX_RC_LAST_PASS;
+  svc_.temporal_layers = 2;
+  codec_enc_.g_error_resilient = 0;
+  codec_enc_.g_w = 704;
+  codec_enc_.g_h = 144;
+  tile_columns_ = 1;
+  tile_rows_ = 1;
+  vpx_svc_set_options(&svc_, "auto-alt-refs=1 scale-factors=1/1 "
+                      "multi-frame-contexts=1");
+  vpx_fixed_buf outputs[10];
+  memset(&outputs[0], 0, sizeof(outputs));
+  Pass2EncodeNFrames(&stats_buf, 10, 1, &outputs[0]);
+  DecodeNFrames(&outputs[0], 10);
+  FreeBitstreamBuffers(&outputs[0], 10);
+}
+
 }  // namespace
--- a/test/test-data.mk
+++ b/test/test-data.mk
@@ -12,6 +12,7 @@ LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_12_420.y4m
 LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_12_422.y4m
 LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_12_444.y4m
 LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_12_440.yuv
+LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_8_420_a10-1.y4m
 LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_8_420.y4m
 LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_8_422.y4m
 LIBVPX_TEST_DATA-$(CONFIG_ENCODERS) += park_joy_90p_8_444.y4m
@@ -686,8 +687,8 @@ LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += vp93-2-20-12bit-yuv444.webm.md5
 endif  # CONFIG_VP9_HIGHBITDEPTH

 # Invalid files for testing libvpx error checking.
-LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-01-v2.webm
-LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-01-v2.webm.res
+LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-01-v3.webm
+LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-01-v3.webm.res
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-02-v2.webm
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-02-v2.webm.res
 LIBVPX_TEST_DATA-$(CONFIG_VP9_DECODER) += invalid-vp90-03-v3.webm
--- a/test/test-data.sha1
+++ b/test/test-data.sha1
--- a/test/test.mk
+++ b/test/test.mk
@@ -66,6 +66,7 @@ LIBVPX_TEST_SRCS-$(CONFIG_DECODERS)    += ../tools_common.h
 LIBVPX_TEST_SRCS-$(CONFIG_DECODERS)    += ../webmdec.cc
 LIBVPX_TEST_SRCS-$(CONFIG_DECODERS)    += ../webmdec.h
 LIBVPX_TEST_SRCS-$(CONFIG_DECODERS)    += webm_video_source.h
+LIBVPX_TEST_SRCS-$(CONFIG_VP9_DECODER) += vp9_skip_loopfilter_test.cc
 endif

 LIBVPX_TEST_SRCS-$(CONFIG_DECODERS)    += decode_api_test.cc
@@ -90,6 +91,7 @@ endif
 ## shared library builds don't make these functions accessible.
 ##
 ifeq ($(CONFIG_SHARED),)
+LIBVPX_TEST_SRCS-$(CONFIG_VP9)         += lpf_8_test.cc

 ## VP8
 ifneq ($(CONFIG_VP8_ENCODER)$(CONFIG_VP8_DECODER),)
@@ -103,7 +105,6 @@ endif
 LIBVPX_TEST_SRCS-$(CONFIG_POSTPROC)    += pp_filter_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP8_DECODER) += vp8_decrypt_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += set_roi.cc
-LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += subtract_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += variance_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += vp8_fdct4x4_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += quantize_test.cc
@@ -142,7 +143,6 @@ LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += fdct4x4_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += fdct8x8_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += variance_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_subtract_test.cc
-LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += lpf_8_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_avg_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_error_block_test.cc
 LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_quantize_test.cc
@@ -150,16 +150,23 @@ LIBVPX_TEST_SRCS-$(CONFIG_VP9)         += vp9_intrapred_test.cc

 ifeq ($(CONFIG_VP9_ENCODER),yes)
 LIBVPX_TEST_SRCS-$(CONFIG_SPATIAL_SVC) += svc_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_INTERNAL_STATS) += blockiness_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_INTERNAL_STATS) += consistency_test.cc
+
 endif

 ifeq ($(CONFIG_VP9_ENCODER)$(CONFIG_VP9_TEMPORAL_DENOISING),yesyes)
 LIBVPX_TEST_SRCS-$(HAVE_SSE2) += vp9_denoiser_sse2_test.cc
 endif
+LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_arf_freq_test.cc

 endif # VP9

 LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS)    += sad_test.cc

+TEST_INTRA_PRED_SPEED_SRCS-$(CONFIG_VP9) := test_intra_pred_speed.cc
+TEST_INTRA_PRED_SPEED_SRCS-$(CONFIG_VP9) += ../md5_utils.h ../md5_utils.c
+
 endif # CONFIG_SHARED

 include $(SRC_PATH_BARE)/test/test-data.mk
--- a/test/test_intra_pred_speed.cc
+++ b/test/test_intra_pred_speed.cc
@@ -0,0 +1,384 @@
+/*
+ *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+//  Test and time VPX intra-predictor functions
+
+#include <stdio.h>
+#include <string.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_dsp_rtcd.h"
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/md5_helper.h"
+#include "vpx/vpx_integer.h"
+#include "vpx_ports/mem.h"
+#include "vpx_ports/vpx_timer.h"
+
+// -----------------------------------------------------------------------------
+
+namespace {
+
+typedef void (*VpxPredFunc)(uint8_t *dst, ptrdiff_t y_stride,
+                            const uint8_t *above, const uint8_t *left);
+
+const int kNumVp9IntraPredFuncs = 13;
+const char *kVp9IntraPredNames[kNumVp9IntraPredFuncs] = {
+  "DC_PRED", "DC_LEFT_PRED", "DC_TOP_PRED", "DC_128_PRED", "V_PRED", "H_PRED",
+  "D45_PRED", "D135_PRED", "D117_PRED", "D153_PRED", "D207_PRED", "D63_PRED",
+  "TM_PRED"
+};
+
+void TestIntraPred(const char name[], VpxPredFunc const *pred_funcs,
+                   const char *const pred_func_names[], int num_funcs,
+                   const char *const signatures[], int block_size,
+                   int num_pixels_per_test) {
+  libvpx_test::ACMRandom rnd(libvpx_test::ACMRandom::DeterministicSeed());
+  const int kBPS = 32;
+  const int kTotalPixels = 32 * kBPS;
+  DECLARE_ALIGNED(16, uint8_t, src[kTotalPixels]);
+  DECLARE_ALIGNED(16, uint8_t, ref_src[kTotalPixels]);
+  DECLARE_ALIGNED(16, uint8_t, left[kBPS]);
+  DECLARE_ALIGNED(16, uint8_t, above_mem[2 * kBPS + 16]);
+  uint8_t *const above = above_mem + 16;
+  for (int i = 0; i < kTotalPixels; ++i) ref_src[i] = rnd.Rand8();
+  for (int i = 0; i < kBPS; ++i) left[i] = rnd.Rand8();
+  for (int i = -1; i < kBPS; ++i) above[i] = rnd.Rand8();
+  const int kNumTests = static_cast<int>(2.e10 / num_pixels_per_test);
+
+  // some code assumes the top row has been extended:
+  // d45/d63 C-code, for instance, but not the assembly.
+  // TODO(jzern): this style of extension isn't strictly necessary.
+  ASSERT_LE(block_size, kBPS);
+  memset(above + block_size, above[block_size - 1], 2 * kBPS - block_size);
+
+  for (int k = 0; k < num_funcs; ++k) {
+    if (pred_funcs[k] == NULL) continue;
+    memcpy(src, ref_src, sizeof(src));
+    vpx_usec_timer timer;
+    vpx_usec_timer_start(&timer);
+    for (int num_tests = 0; num_tests < kNumTests; ++num_tests) {
+      pred_funcs[k](src, kBPS, above, left);
+    }
+    libvpx_test::ClearSystemState();
+    vpx_usec_timer_mark(&timer);
+    const int elapsed_time =
+        static_cast<int>(vpx_usec_timer_elapsed(&timer) / 1000);
+    libvpx_test::MD5 md5;
+    md5.Add(src, sizeof(src));
+    printf("Mode %s[%12s]: %5d ms     MD5: %s\n", name, pred_func_names[k],
+           elapsed_time, md5.Get());
+    EXPECT_STREQ(signatures[k], md5.Get());
+  }
+}
+
+void TestIntraPred4(VpxPredFunc const *pred_funcs) {
+  static const int kNumVp9IntraFuncs = 13;
+  static const char *const kSignatures[kNumVp9IntraFuncs] = {
+    "4334156168b34ab599d9b5b30f522fe9",
+    "bc4649d5ba47c7ff178d92e475960fb0",
+    "8d316e5933326dcac24e1064794b5d12",
+    "a27270fed024eafd762c95de85f4da51",
+    "c33dff000d4256c2b8f3bf9e9bab14d2",
+    "44d8cddc2ad8f79b8ed3306051722b4f",
+    "eb54839b2bad6699d8946f01ec041cd0",
+    "ecb0d56ae5f677ea45127ce9d5c058e4",
+    "0b7936841f6813da818275944895b574",
+    "9117972ef64f91a58ff73e1731c81db2",
+    "c56d5e8c729e46825f46dd5d3b5d508a",
+    "c0889e2039bcf7bcb5d2f33cdca69adc",
+    "309a618577b27c648f9c5ee45252bc8f",
+  };
+  TestIntraPred("Intra4", pred_funcs, kVp9IntraPredNames, kNumVp9IntraFuncs,
+                kSignatures, 4, 4 * 4 * kNumVp9IntraFuncs);
+}
+
+void TestIntraPred8(VpxPredFunc const *pred_funcs) {
+  static const int kNumVp9IntraFuncs = 13;
+  static const char *const kSignatures[kNumVp9IntraFuncs] = {
+    "7694ddeeefed887faf9d339d18850928",
+    "7d726b1213591b99f736be6dec65065b",
+    "19c5711281357a485591aaf9c96c0a67",
+    "ba6b66877a089e71cd938e3b8c40caac",
+    "802440c93317e0f8ba93fab02ef74265",
+    "9e09a47a15deb0b9d8372824f9805080",
+    "b7c2d8c662268c0c427da412d7b0311d",
+    "78339c1c60bb1d67d248ab8c4da08b7f",
+    "5c97d70f7d47de1882a6cd86c165c8a9",
+    "8182bf60688b42205acd95e59e967157",
+    "08323400005a297f16d7e57e7fe1eaac",
+    "95f7bfc262329a5849eda66d8f7c68ce",
+    "815b75c8e0d91cc1ae766dc5d3e445a3",
+  };
+  TestIntraPred("Intra8", pred_funcs, kVp9IntraPredNames, kNumVp9IntraFuncs,
+                kSignatures, 8, 8 * 8 * kNumVp9IntraFuncs);
+}
+
+void TestIntraPred16(VpxPredFunc const *pred_funcs) {
+  static const int kNumVp9IntraFuncs = 13;
+  static const char *const kSignatures[kNumVp9IntraFuncs] = {
+    "b40dbb555d5d16a043dc361e6694fe53",
+    "fb08118cee3b6405d64c1fd68be878c6",
+    "6c190f341475c837cc38c2e566b64875",
+    "db5c34ccbe2c7f595d9b08b0dc2c698c",
+    "a62cbfd153a1f0b9fed13e62b8408a7a",
+    "143df5b4c89335e281103f610f5052e4",
+    "d87feb124107cdf2cfb147655aa0bb3c",
+    "7841fae7d4d47b519322e6a03eeed9dc",
+    "f6ebed3f71cbcf8d6d0516ce87e11093",
+    "3cc480297dbfeed01a1c2d78dd03d0c5",
+    "b9f69fa6532b372c545397dcb78ef311",
+    "a8fe1c70432f09d0c20c67bdb6432c4d",
+    "b8a41aa968ec108af447af4217cba91b",
+  };
+  TestIntraPred("Intra16", pred_funcs, kVp9IntraPredNames, kNumVp9IntraFuncs,
+                kSignatures, 16, 16 * 16 * kNumVp9IntraFuncs);
+}
+
+void TestIntraPred32(VpxPredFunc const *pred_funcs) {
+  static const int kNumVp9IntraFuncs = 13;
+  static const char *const kSignatures[kNumVp9IntraFuncs] = {
+    "558541656d84f9ae7896db655826febe",
+    "b3587a1f9a01495fa38c8cd3c8e2a1bf",
+    "4c6501e64f25aacc55a2a16c7e8f0255",
+    "b3b01379ba08916ef6b1b35f7d9ad51c",
+    "0f1eb38b6cbddb3d496199ef9f329071",
+    "911c06efb9ed1c3b4c104b232b55812f",
+    "9225beb0ddfa7a1d24eaa1be430a6654",
+    "0a6d584a44f8db9aa7ade2e2fdb9fc9e",
+    "b01c9076525216925f3456f034fb6eee",
+    "d267e20ad9e5cd2915d1a47254d3d149",
+    "ed012a4a5da71f36c2393023184a0e59",
+    "f162b51ed618d28b936974cff4391da5",
+    "9e1370c6d42e08d357d9612c93a71cfc",
+  };
+  TestIntraPred("Intra32", pred_funcs, kVp9IntraPredNames, kNumVp9IntraFuncs,
+                kSignatures, 32, 32 * 32 * kNumVp9IntraFuncs);
+}
+
+}  // namespace
+
+// Defines a test case for |arch| (e.g., C, SSE2, ...) passing the predictors
+// to |test_func|. The test name is 'arch.test_func', e.g., C.TestIntraPred4.
+#define INTRA_PRED_TEST(arch, test_func, dc, dc_left, dc_top, dc_128, v, h, \
+                        d45, d135, d117, d153, d207, d63, tm)               \
+  TEST(arch, test_func) {                                                   \
+    static const VpxPredFunc vpx_intra_pred[] = {                           \
+        dc,   dc_left, dc_top, dc_128, v,   h, d45,                         \
+        d135, d117,    d153,   d207,   d63, tm};                            \
+    test_func(vpx_intra_pred);                                              \
+  }
+
+// -----------------------------------------------------------------------------
+// 4x4
+
+INTRA_PRED_TEST(C, TestIntraPred4, vpx_dc_predictor_4x4_c,
+                vpx_dc_left_predictor_4x4_c, vpx_dc_top_predictor_4x4_c,
+                vpx_dc_128_predictor_4x4_c, vpx_v_predictor_4x4_c,
+                vpx_h_predictor_4x4_c, vpx_d45_predictor_4x4_c,
+                vpx_d135_predictor_4x4_c, vpx_d117_predictor_4x4_c,
+                vpx_d153_predictor_4x4_c, vpx_d207_predictor_4x4_c,
+                vpx_d63_predictor_4x4_c, vpx_tm_predictor_4x4_c)
+
+#if HAVE_SSE && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSE, TestIntraPred4, vpx_dc_predictor_4x4_sse,
+                vpx_dc_left_predictor_4x4_sse, vpx_dc_top_predictor_4x4_sse,
+                vpx_dc_128_predictor_4x4_sse, vpx_v_predictor_4x4_sse, NULL,
+                NULL, NULL, NULL, NULL, NULL, NULL, vpx_tm_predictor_4x4_sse)
+#endif  // HAVE_SSE && CONFIG_USE_X86INC
+
+#if HAVE_SSSE3 && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSSE3, TestIntraPred4, NULL, NULL, NULL, NULL, NULL,
+                vpx_h_predictor_4x4_ssse3, vpx_d45_predictor_4x4_ssse3, NULL,
+                NULL, vpx_d153_predictor_4x4_ssse3,
+                vpx_d207_predictor_4x4_ssse3, vpx_d63_predictor_4x4_ssse3, NULL)
+#endif  // HAVE_SSSE3 && CONFIG_USE_X86INC
+
+#if HAVE_DSPR2
+INTRA_PRED_TEST(DSPR2, TestIntraPred4, vpx_dc_predictor_4x4_dspr2, NULL, NULL,
+                NULL, NULL, vpx_h_predictor_4x4_dspr2, NULL, NULL, NULL, NULL,
+                NULL, NULL, vpx_tm_predictor_4x4_dspr2)
+#endif  // HAVE_DSPR2
+
+#if HAVE_NEON
+INTRA_PRED_TEST(NEON, TestIntraPred4, vpx_dc_predictor_4x4_neon,
+                vpx_dc_left_predictor_4x4_neon, vpx_dc_top_predictor_4x4_neon,
+                vpx_dc_128_predictor_4x4_neon, vpx_v_predictor_4x4_neon,
+                vpx_h_predictor_4x4_neon, vpx_d45_predictor_4x4_neon,
+                vpx_d135_predictor_4x4_neon, NULL, NULL, NULL, NULL,
+                vpx_tm_predictor_4x4_neon)
+#endif  // HAVE_NEON
+
+#if HAVE_MSA
+INTRA_PRED_TEST(MSA, TestIntraPred4, vpx_dc_predictor_4x4_msa,
+                vpx_dc_left_predictor_4x4_msa, vpx_dc_top_predictor_4x4_msa,
+                vpx_dc_128_predictor_4x4_msa, vpx_v_predictor_4x4_msa,
+                vpx_h_predictor_4x4_msa, NULL, NULL, NULL, NULL, NULL,
+                NULL, vpx_tm_predictor_4x4_msa)
+#endif  // HAVE_MSA
+
+// -----------------------------------------------------------------------------
+// 8x8
+
+INTRA_PRED_TEST(C, TestIntraPred8, vpx_dc_predictor_8x8_c,
+                vpx_dc_left_predictor_8x8_c, vpx_dc_top_predictor_8x8_c,
+                vpx_dc_128_predictor_8x8_c, vpx_v_predictor_8x8_c,
+                vpx_h_predictor_8x8_c, vpx_d45_predictor_8x8_c,
+                vpx_d135_predictor_8x8_c, vpx_d117_predictor_8x8_c,
+                vpx_d153_predictor_8x8_c, vpx_d207_predictor_8x8_c,
+                vpx_d63_predictor_8x8_c, vpx_tm_predictor_8x8_c)
+
+#if HAVE_SSE && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSE, TestIntraPred8, vpx_dc_predictor_8x8_sse,
+                vpx_dc_left_predictor_8x8_sse, vpx_dc_top_predictor_8x8_sse,
+                vpx_dc_128_predictor_8x8_sse, vpx_v_predictor_8x8_sse, NULL,
+                NULL, NULL, NULL, NULL, NULL, NULL, NULL)
+#endif  // HAVE_SSE && CONFIG_USE_X86INC
+
+#if HAVE_SSE2 && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSE2, TestIntraPred8, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+                NULL, NULL, NULL, NULL, NULL, vpx_tm_predictor_8x8_sse2)
+#endif  // HAVE_SSE2 && CONFIG_USE_X86INC
+
+#if HAVE_SSSE3 && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSSE3, TestIntraPred8, NULL, NULL, NULL, NULL, NULL,
+                vpx_h_predictor_8x8_ssse3, vpx_d45_predictor_8x8_ssse3, NULL,
+                NULL, vpx_d153_predictor_8x8_ssse3,
+                vpx_d207_predictor_8x8_ssse3, vpx_d63_predictor_8x8_ssse3, NULL)
+#endif  // HAVE_SSSE3 && CONFIG_USE_X86INC
+
+#if HAVE_DSPR2
+INTRA_PRED_TEST(DSPR2, TestIntraPred8, vpx_dc_predictor_8x8_dspr2, NULL, NULL,
+                NULL, NULL, vpx_h_predictor_8x8_dspr2, NULL, NULL, NULL, NULL,
+                NULL, NULL, vpx_tm_predictor_8x8_c)
+#endif  // HAVE_DSPR2
+
+#if HAVE_NEON
+INTRA_PRED_TEST(NEON, TestIntraPred8, vpx_dc_predictor_8x8_neon,
+                vpx_dc_left_predictor_8x8_neon, vpx_dc_top_predictor_8x8_neon,
+                vpx_dc_128_predictor_8x8_neon, vpx_v_predictor_8x8_neon,
+                vpx_h_predictor_8x8_neon, vpx_d45_predictor_8x8_neon, NULL,
+                NULL, NULL, NULL, NULL, vpx_tm_predictor_8x8_neon)
+
+#endif  // HAVE_NEON
+
+#if HAVE_MSA
+INTRA_PRED_TEST(MSA, TestIntraPred8, vpx_dc_predictor_8x8_msa,
+                vpx_dc_left_predictor_8x8_msa, vpx_dc_top_predictor_8x8_msa,
+                vpx_dc_128_predictor_8x8_msa, vpx_v_predictor_8x8_msa,
+                vpx_h_predictor_8x8_msa, NULL, NULL, NULL, NULL, NULL,
+                NULL, vpx_tm_predictor_8x8_msa)
+#endif  // HAVE_MSA
+
+// -----------------------------------------------------------------------------
+// 16x16
+
+INTRA_PRED_TEST(C, TestIntraPred16, vpx_dc_predictor_16x16_c,
+                vpx_dc_left_predictor_16x16_c, vpx_dc_top_predictor_16x16_c,
+                vpx_dc_128_predictor_16x16_c, vpx_v_predictor_16x16_c,
+                vpx_h_predictor_16x16_c, vpx_d45_predictor_16x16_c,
+                vpx_d135_predictor_16x16_c, vpx_d117_predictor_16x16_c,
+                vpx_d153_predictor_16x16_c, vpx_d207_predictor_16x16_c,
+                vpx_d63_predictor_16x16_c, vpx_tm_predictor_16x16_c)
+
+#if HAVE_SSE2 && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSE2, TestIntraPred16, vpx_dc_predictor_16x16_sse2,
+                vpx_dc_left_predictor_16x16_sse2,
+                vpx_dc_top_predictor_16x16_sse2,
+                vpx_dc_128_predictor_16x16_sse2, vpx_v_predictor_16x16_sse2,
+                NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+                vpx_tm_predictor_16x16_sse2)
+#endif  // HAVE_SSE2 && CONFIG_USE_X86INC
+
+#if HAVE_SSSE3 && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSSE3, TestIntraPred16, NULL, NULL, NULL, NULL, NULL,
+                vpx_h_predictor_16x16_ssse3, vpx_d45_predictor_16x16_ssse3,
+                NULL, NULL, vpx_d153_predictor_16x16_ssse3,
+                vpx_d207_predictor_16x16_ssse3, vpx_d63_predictor_16x16_ssse3,
+                NULL)
+#endif  // HAVE_SSSE3 && CONFIG_USE_X86INC
+
+#if HAVE_DSPR2
+INTRA_PRED_TEST(DSPR2, TestIntraPred16, vpx_dc_predictor_16x16_dspr2, NULL,
+                NULL, NULL, NULL, vpx_h_predictor_16x16_dspr2, NULL, NULL, NULL,
+                NULL, NULL, NULL, NULL)
+#endif  // HAVE_DSPR2
+
+#if HAVE_NEON
+INTRA_PRED_TEST(NEON, TestIntraPred16, vpx_dc_predictor_16x16_neon,
+                vpx_dc_left_predictor_16x16_neon,
+                vpx_dc_top_predictor_16x16_neon,
+                vpx_dc_128_predictor_16x16_neon, vpx_v_predictor_16x16_neon,
+                vpx_h_predictor_16x16_neon, vpx_d45_predictor_16x16_neon, NULL,
+                NULL, NULL, NULL, NULL, vpx_tm_predictor_16x16_neon)
+#endif  // HAVE_NEON
+
+#if HAVE_MSA
+INTRA_PRED_TEST(MSA, TestIntraPred16, vpx_dc_predictor_16x16_msa,
+                vpx_dc_left_predictor_16x16_msa, vpx_dc_top_predictor_16x16_msa,
+                vpx_dc_128_predictor_16x16_msa, vpx_v_predictor_16x16_msa,
+                vpx_h_predictor_16x16_msa, NULL, NULL, NULL, NULL, NULL,
+                NULL, vpx_tm_predictor_16x16_msa)
+#endif  // HAVE_MSA
+
+// -----------------------------------------------------------------------------
+// 32x32
+
+INTRA_PRED_TEST(C, TestIntraPred32, vpx_dc_predictor_32x32_c,
+                vpx_dc_left_predictor_32x32_c, vpx_dc_top_predictor_32x32_c,
+                vpx_dc_128_predictor_32x32_c, vpx_v_predictor_32x32_c,
+                vpx_h_predictor_32x32_c, vpx_d45_predictor_32x32_c,
+                vpx_d135_predictor_32x32_c, vpx_d117_predictor_32x32_c,
+                vpx_d153_predictor_32x32_c, vpx_d207_predictor_32x32_c,
+                vpx_d63_predictor_32x32_c, vpx_tm_predictor_32x32_c)
+
+#if HAVE_SSE2 && CONFIG_USE_X86INC
+#if ARCH_X86_64
+INTRA_PRED_TEST(SSE2, TestIntraPred32, vpx_dc_predictor_32x32_sse2,
+                vpx_dc_left_predictor_32x32_sse2,
+                vpx_dc_top_predictor_32x32_sse2,
+                vpx_dc_128_predictor_32x32_sse2, vpx_v_predictor_32x32_sse2,
+                NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+                vpx_tm_predictor_32x32_sse2)
+#else
+INTRA_PRED_TEST(SSE2, TestIntraPred32, vpx_dc_predictor_32x32_sse2,
+                vpx_dc_left_predictor_32x32_sse2,
+                vpx_dc_top_predictor_32x32_sse2,
+                vpx_dc_128_predictor_32x32_sse2, vpx_v_predictor_32x32_sse2,
+                NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL)
+#endif  // ARCH_X86_64
+#endif  // HAVE_SSE2 && CONFIG_USE_X86INC
+
+#if HAVE_SSSE3 && CONFIG_USE_X86INC
+INTRA_PRED_TEST(SSSE3, TestIntraPred32, NULL, NULL, NULL, NULL, NULL,
+                vpx_h_predictor_32x32_ssse3, vpx_d45_predictor_32x32_ssse3,
+                NULL, NULL, vpx_d153_predictor_32x32_ssse3,
+                vpx_d207_predictor_32x32_ssse3, vpx_d63_predictor_32x32_ssse3,
+                NULL)
+#endif  // HAVE_SSSE3 && CONFIG_USE_X86INC
+
+#if HAVE_NEON
+INTRA_PRED_TEST(NEON, TestIntraPred32, vpx_dc_predictor_32x32_neon,
+                vpx_dc_left_predictor_32x32_neon,
+                vpx_dc_top_predictor_32x32_neon,
+                vpx_dc_128_predictor_32x32_neon, vpx_v_predictor_32x32_neon,
+                vpx_h_predictor_32x32_neon, NULL, NULL, NULL, NULL, NULL, NULL,
+                vpx_tm_predictor_32x32_neon)
+#endif  // HAVE_NEON
+
+#if HAVE_MSA
+INTRA_PRED_TEST(MSA, TestIntraPred32, vpx_dc_predictor_32x32_msa,
+                vpx_dc_left_predictor_32x32_msa, vpx_dc_top_predictor_32x32_msa,
+                vpx_dc_128_predictor_32x32_msa, vpx_v_predictor_32x32_msa,
+                vpx_h_predictor_32x32_msa, NULL, NULL, NULL, NULL, NULL,
+                NULL, vpx_tm_predictor_32x32_msa)
+#endif  // HAVE_MSA
+
+#include "test/test_libvpx.cc"
--- a/test/test_libvpx.cc
+++ b/test/test_libvpx.cc
@@ -8,6 +8,9 @@
 *  be found in the AUTHORS file in the root of the source tree.
 */
 #include <string>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
 #include "./vpx_config.h"
 #if ARCH_X86 || ARCH_X86_64
 #include "vpx_ports/x86.h"
@@ -15,13 +18,15 @@
 extern "C" {
 #if CONFIG_VP8
 extern void vp8_rtcd();
-#endif
+#endif  // CONFIG_VP8
 #if CONFIG_VP9
 extern void vp9_rtcd();
-#endif
+#endif  // CONFIG_VP9
+extern void vpx_dsp_rtcd();
+extern void vpx_scale_rtcd();
 }
-#include "third_party/googletest/src/include/gtest/gtest.h"

+#if ARCH_X86 || ARCH_X86_64
 static void append_negative_gtest_filter(const char *str) {
  std::string filter = ::testing::FLAGS_gtest_filter;
  // Negative patterns begin with one '-' followed by a ':' separated list.
@@ -29,6 +34,7 @@ static void append_negative_gtest_filter(const char *str) {
  filter += str;
  ::testing::FLAGS_gtest_filter = filter;
 }
+#endif  // ARCH_X86 || ARCH_X86_64

 int main(int argc, char **argv) {
  ::testing::InitGoogleTest(&argc, argv);
@@ -36,22 +42,22 @@ int main(int argc, char **argv) {
 #if ARCH_X86 || ARCH_X86_64
  const int simd_caps = x86_simd_caps();
  if (!(simd_caps & HAS_MMX))
-    append_negative_gtest_filter(":MMX/*");
+    append_negative_gtest_filter(":MMX.*:MMX/*");
  if (!(simd_caps & HAS_SSE))
-    append_negative_gtest_filter(":SSE/*");
+    append_negative_gtest_filter(":SSE.*:SSE/*");
  if (!(simd_caps & HAS_SSE2))
-    append_negative_gtest_filter(":SSE2/*");
+    append_negative_gtest_filter(":SSE2.*:SSE2/*");
  if (!(simd_caps & HAS_SSE3))
-    append_negative_gtest_filter(":SSE3/*");
+    append_negative_gtest_filter(":SSE3.*:SSE3/*");
  if (!(simd_caps & HAS_SSSE3))
-    append_negative_gtest_filter(":SSSE3/*");
+    append_negative_gtest_filter(":SSSE3.*:SSSE3/*");
  if (!(simd_caps & HAS_SSE4_1))
-    append_negative_gtest_filter(":SSE4_1/*");
+    append_negative_gtest_filter(":SSE4_1.*:SSE4_1/*");
  if (!(simd_caps & HAS_AVX))
-    append_negative_gtest_filter(":AVX/*");
+    append_negative_gtest_filter(":AVX.*:AVX/*");
  if (!(simd_caps & HAS_AVX2))
-    append_negative_gtest_filter(":AVX2/*");
-#endif
+    append_negative_gtest_filter(":AVX2.*:AVX2/*");
+#endif  // ARCH_X86 || ARCH_X86_64

 #if !CONFIG_SHARED
 // Shared library builds don't support whitebox tests
@@ -59,11 +65,13 @@ int main(int argc, char **argv) {

 #if CONFIG_VP8
  vp8_rtcd();
-#endif
+#endif  // CONFIG_VP8
 #if CONFIG_VP9
  vp9_rtcd();
-#endif
-#endif
+#endif  // CONFIG_VP9
+  vpx_dsp_rtcd();
+  vpx_scale_rtcd();
+#endif  // !CONFIG_SHARED

  return RUN_ALL_TESTS();
 }
--- a/test/test_vector_test.cc
+++ b/test/test_vector_test.cc
@@ -135,6 +135,7 @@ TEST_P(TestVectorTest, MD5Match) {

 // Test VP8 decode in serial mode with single thread.
 // NOTE: VP8 only support serial mode.
+#if CONFIG_VP8_DECODER
 VP8_INSTANTIATE_TEST_CASE(
    TestVectorTest,
    ::testing::Combine(
@@ -143,8 +144,10 @@ VP8_INSTANTIATE_TEST_CASE(
        ::testing::ValuesIn(libvpx_test::kVP8TestVectors,
                            libvpx_test::kVP8TestVectors +
                                libvpx_test::kNumVP8TestVectors)));
+#endif  // CONFIG_VP8_DECODER

 // Test VP9 decode in serial mode with single thread.
+#if CONFIG_VP9_DECODER
 VP9_INSTANTIATE_TEST_CASE(
    TestVectorTest,
    ::testing::Combine(
@@ -154,8 +157,6 @@ VP9_INSTANTIATE_TEST_CASE(
                            libvpx_test::kVP9TestVectors +
                                libvpx_test::kNumVP9TestVectors)));

-
-#if CONFIG_VP9_DECODER
 // Test VP9 decode in frame parallel mode with different number of threads.
 INSTANTIATE_TEST_CASE_P(
    VP9MultiThreadedFrameParallel, TestVectorTest,
--- a/test/test_vectors.cc
+++ b/test/test_vectors.cc
@@ -165,7 +165,10 @@ const char *const kVP9TestVectors[] = {
  "vp90-2-11-size-351x287.webm", "vp90-2-11-size-351x288.webm",
  "vp90-2-11-size-352x287.webm", "vp90-2-12-droppable_1.ivf",
  "vp90-2-12-droppable_2.ivf", "vp90-2-12-droppable_3.ivf",
+#if !CONFIG_SIZE_LIMIT || \
+    (DECODE_WIDTH_LIMIT >= 20400 && DECODE_HEIGHT_LIMIT >= 120)
  "vp90-2-13-largescaling.webm",
+#endif
  "vp90-2-14-resize-fp-tiles-1-16.webm",
  "vp90-2-14-resize-fp-tiles-1-2-4-8-16.webm",
  "vp90-2-14-resize-fp-tiles-1-2.webm", "vp90-2-14-resize-fp-tiles-1-4.webm",
--- a/test/tile_independence_test.cc
+++ b/test/tile_independence_test.cc
@@ -104,4 +104,5 @@ TEST_P(TileIndependenceTest, MD5Match) {

 VP9_INSTANTIATE_TEST_CASE(TileIndependenceTest, ::testing::Range(0, 2, 1));

+VP10_INSTANTIATE_TEST_CASE(TileIndependenceTest, ::testing::Range(0, 2, 1));
 }  // namespace
--- a/test/tools_common.sh
+++ b/test/tools_common.sh
@@ -402,11 +402,15 @@ VP9_IVF_FILE="${LIBVPX_TEST_DATA_PATH}/vp90-2-09-subpixel-00.ivf"

 VP9_WEBM_FILE="${LIBVPX_TEST_DATA_PATH}/vp90-2-00-quantizer-00.webm"
 VP9_FPM_WEBM_FILE="${LIBVPX_TEST_DATA_PATH}/vp90-2-07-frame_parallel-1.webm"
+VP9_LT_50_FRAMES_WEBM_FILE="${LIBVPX_TEST_DATA_PATH}/vp90-2-02-size-32x08.webm"

 YUV_RAW_INPUT="${LIBVPX_TEST_DATA_PATH}/hantro_collage_w352h288.yuv"
 YUV_RAW_INPUT_WIDTH=352
 YUV_RAW_INPUT_HEIGHT=288

+Y4M_NOSQ_PAR_INPUT="${LIBVPX_TEST_DATA_PATH}/park_joy_90p_8_420_a10-1.y4m"
+Y4M_720P_INPUT="${LIBVPX_TEST_DATA_PATH}/niklas_1280_720_30.y4m"
+
 # Setup a trap function to clean up after tests complete.
 trap cleanup EXIT

@@ -428,6 +432,7 @@ vlog "$(basename "${0%.*}") test configuration:
  VPX_TEST_VERBOSE_OUTPUT=${VPX_TEST_VERBOSE_OUTPUT}
  YUV_RAW_INPUT=${YUV_RAW_INPUT}
  YUV_RAW_INPUT_WIDTH=${YUV_RAW_INPUT_WIDTH}
-  YUV_RAW_INPUT_HEIGHT=${YUV_RAW_INPUT_HEIGHT}"
+  YUV_RAW_INPUT_HEIGHT=${YUV_RAW_INPUT_HEIGHT}
+  Y4M_NOSQ_PAR_INPUT=${Y4M_NOSQ_PAR_INPUT}"

 fi  # End $VPX_TEST_TOOLS_COMMON_SH pseudo include guard.
--- a/test/util.h
+++ b/test/util.h
@@ -19,8 +19,7 @@
 // Macros
 #define GET_PARAM(k) std::tr1::get< k >(GetParam())

-static double compute_psnr(const vpx_image_t *img1,
-                           const vpx_image_t *img2) {
+inline double compute_psnr(const vpx_image_t *img1, const vpx_image_t *img2) {
  assert((img1->fmt == img2->fmt) &&
         (img1->d_w == img2->d_w) &&
         (img1->d_h == img2->d_h));
--- a/test/variance_test.cc
+++ b/test/variance_test.cc
--- a/test/video_source.h
+++ b/test/video_source.h
@@ -48,7 +48,7 @@ static std::string GetDataPath() {
 #undef TO_STRING
 #undef STRINGIFY

-static FILE *OpenTestDataFile(const std::string& file_name) {
+inline FILE *OpenTestDataFile(const std::string& file_name) {
  const std::string path_to_source = GetDataPath() + "/" + file_name;
  return fopen(path_to_source.c_str(), "rb");
 }
--- a/test/vp8_boolcoder_test.cc
+++ b/test/vp8_boolcoder_test.cc
@@ -16,12 +16,12 @@
 #include <string.h>
 #include <sys/types.h>

-#include "test/acm_random.h"
 #include "third_party/googletest/src/include/gtest/gtest.h"
-#include "vpx/vpx_integer.h"

-#include "vp8/encoder/boolhuff.h"
+#include "test/acm_random.h"
 #include "vp8/decoder/dboolhuff.h"
+#include "vp8/encoder/boolhuff.h"
+#include "vpx/vpx_integer.h"

 namespace {
 const int num_tests = 10;
--- a/test/vp8_denoiser_sse2_test.cc
+++ b/test/vp8_denoiser_sse2_test.cc
@@ -52,13 +52,13 @@ TEST_P(VP8DenoiserTest, BitexactCheck) {
  // mc_avg_block is the denoised reference block,
  // avg_block_c is the denoised result from C code,
  // avg_block_sse2 is the denoised result from SSE2 code.
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, sig_block_c, kNumPixels);
+  DECLARE_ALIGNED(16, uint8_t, sig_block_c[kNumPixels]);
  // Since in VP8 denoiser, the source signal will be changed,
  // we need another copy of the source signal as the input of sse2 code.
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, sig_block_sse2, kNumPixels);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, mc_avg_block, kNumPixels);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, avg_block_c, kNumPixels);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, avg_block_sse2, kNumPixels);
+  DECLARE_ALIGNED(16, uint8_t, sig_block_sse2[kNumPixels]);
+  DECLARE_ALIGNED(16, uint8_t, mc_avg_block[kNumPixels]);
+  DECLARE_ALIGNED(16, uint8_t, avg_block_c[kNumPixels]);
+  DECLARE_ALIGNED(16, uint8_t, avg_block_sse2[kNumPixels]);

  for (int i = 0; i < count_test_block; ++i) {
    // Generate random motion magnitude, 20% of which exceed the threshold.
--- a/test/vp8_fdct4x4_test.cc
+++ b/test/vp8_fdct4x4_test.cc
@@ -15,10 +15,10 @@
 #include <string.h>
 #include <sys/types.h>

-#include "./vp8_rtcd.h"
-
-#include "test/acm_random.h"
 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp8_rtcd.h"
+#include "test/acm_random.h"
 #include "vpx/vpx_integer.h"

 namespace {
--- a/test/vp9_arf_freq_test.cc
+++ b/test/vp9_arf_freq_test.cc
@@ -0,0 +1,238 @@
+/*
+ *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "test/codec_factory.h"
+#include "test/encode_test_driver.h"
+#include "test/util.h"
+#include "test/y4m_video_source.h"
+#include "test/yuv_video_source.h"
+#include "vp9/encoder/vp9_ratectrl.h"
+
+namespace {
+
+const unsigned int kFrames = 100;
+const int kBitrate = 500;
+
+#define ARF_NOT_SEEN               1000001
+#define ARF_SEEN_ONCE              1000000
+
+typedef struct {
+  const char *filename;
+  unsigned int width;
+  unsigned int height;
+  unsigned int framerate_num;
+  unsigned int framerate_den;
+  unsigned int input_bit_depth;
+  vpx_img_fmt fmt;
+  vpx_bit_depth_t bit_depth;
+  unsigned int profile;
+} TestVideoParam;
+
+typedef struct {
+  libvpx_test::TestMode mode;
+  int cpu_used;
+} TestEncodeParam;
+
+const TestVideoParam kTestVectors[] = {
+  // artificially increase framerate to trigger default check
+  {"hantro_collage_w352h288.yuv", 352, 288, 5000, 1,
+    8, VPX_IMG_FMT_I420, VPX_BITS_8, 0},
+  {"hantro_collage_w352h288.yuv", 352, 288, 30, 1,
+    8, VPX_IMG_FMT_I420, VPX_BITS_8, 0},
+  {"rush_hour_444.y4m", 352, 288, 30, 1,
+    8, VPX_IMG_FMT_I444, VPX_BITS_8, 1},
+#if CONFIG_VP9_HIGHBITDEPTH
+  // Add list of profile 2/3 test videos here ...
+#endif  // CONFIG_VP9_HIGHBITDEPTH
+};
+
+const TestEncodeParam kEncodeVectors[] = {
+  {::libvpx_test::kOnePassGood, 2},
+  {::libvpx_test::kOnePassGood, 5},
+  {::libvpx_test::kTwoPassGood, 1},
+  {::libvpx_test::kTwoPassGood, 2},
+  {::libvpx_test::kTwoPassGood, 5},
+  {::libvpx_test::kRealTime, 5},
+};
+
+const int kMinArfVectors[] = {
+  // NOTE: 0 refers to the default built-in logic in:
+  //       vp9_rc_get_default_min_gf_interval(...)
+  0, 4, 8, 12, 15
+};
+
+int is_extension_y4m(const char *filename) {
+  const char *dot = strrchr(filename, '.');
+  if (!dot || dot == filename)
+    return 0;
+  else
+    return !strcmp(dot, ".y4m");
+}
+
+class ArfFreqTest
+    : public ::libvpx_test::EncoderTest,
+      public ::libvpx_test::CodecTestWith3Params<TestVideoParam, \
+                                                 TestEncodeParam, int> {
+ protected:
+  ArfFreqTest()
+      : EncoderTest(GET_PARAM(0)),
+        test_video_param_(GET_PARAM(1)),
+        test_encode_param_(GET_PARAM(2)),
+        min_arf_requested_(GET_PARAM(3)) {
+  }
+
+  virtual ~ArfFreqTest() {}
+
+  virtual void SetUp() {
+    InitializeConfig();
+    SetMode(test_encode_param_.mode);
+    if (test_encode_param_.mode != ::libvpx_test::kRealTime) {
+      cfg_.g_lag_in_frames = 25;
+      cfg_.rc_end_usage = VPX_VBR;
+    } else {
+      cfg_.g_lag_in_frames = 0;
+      cfg_.rc_end_usage = VPX_CBR;
+      cfg_.rc_buf_sz = 1000;
+      cfg_.rc_buf_initial_sz = 500;
+      cfg_.rc_buf_optimal_sz = 600;
+    }
+    dec_cfg_.threads = 4;
+  }
+
+  virtual void BeginPassHook(unsigned int) {
+    min_run_ = ARF_NOT_SEEN;
+    run_of_visible_frames_ = 0;
+  }
+
+  int GetNumFramesInPkt(const vpx_codec_cx_pkt_t *pkt) {
+    const uint8_t *buffer = reinterpret_cast<uint8_t*>(pkt->data.frame.buf);
+    const uint8_t marker = buffer[pkt->data.frame.sz - 1];
+    const int mag = ((marker >> 3) & 3) + 1;
+    int frames = (marker & 0x7) + 1;
+    const unsigned int index_sz = 2 + mag  * frames;
+    // Check for superframe or not.
+    // Assume superframe has only one visible frame, the rest being
+    // invisible. If superframe index is not found, then there is only
+    // one frame.
+    if (!((marker & 0xe0) == 0xc0 &&
+          pkt->data.frame.sz >= index_sz &&
+          buffer[pkt->data.frame.sz - index_sz] == marker)) {
+      frames = 1;
+    }
+    return frames;
+  }
+
+  virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+    if (pkt->kind != VPX_CODEC_CX_FRAME_PKT)
+      return;
+    const int frames = GetNumFramesInPkt(pkt);
+    if (frames == 1) {
+      run_of_visible_frames_++;
+    } else if (frames == 2) {
+      if (min_run_ == ARF_NOT_SEEN) {
+        min_run_ = ARF_SEEN_ONCE;
+      } else if (min_run_ == ARF_SEEN_ONCE ||
+                 run_of_visible_frames_ < min_run_) {
+        min_run_ = run_of_visible_frames_;
+      }
+      run_of_visible_frames_ = 1;
+    } else {
+      min_run_ = 0;
+      run_of_visible_frames_ = 1;
+    }
+  }
+
+  virtual void PreEncodeFrameHook(::libvpx_test::VideoSource *video,
+                                  ::libvpx_test::Encoder *encoder) {
+    if (video->frame() == 0) {
+      encoder->Control(VP9E_SET_FRAME_PARALLEL_DECODING, 1);
+      encoder->Control(VP9E_SET_TILE_COLUMNS, 4);
+      encoder->Control(VP8E_SET_CPUUSED, test_encode_param_.cpu_used);
+      encoder->Control(VP9E_SET_MIN_GF_INTERVAL, min_arf_requested_);
+      if (test_encode_param_.mode != ::libvpx_test::kRealTime) {
+        encoder->Control(VP8E_SET_ENABLEAUTOALTREF, 1);
+        encoder->Control(VP8E_SET_ARNR_MAXFRAMES, 7);
+        encoder->Control(VP8E_SET_ARNR_STRENGTH, 5);
+        encoder->Control(VP8E_SET_ARNR_TYPE, 3);
+      }
+    }
+  }
+
+  int GetMinVisibleRun() const {
+    return min_run_;
+  }
+
+  int GetMinArfDistanceRequested() const {
+    if (min_arf_requested_)
+      return min_arf_requested_;
+    else
+      return vp9_rc_get_default_min_gf_interval(
+          test_video_param_.width, test_video_param_.height,
+          (double)test_video_param_.framerate_num /
+          test_video_param_.framerate_den);
+  }
+
+  TestVideoParam test_video_param_;
+  TestEncodeParam test_encode_param_;
+
+ private:
+  int min_arf_requested_;
+  int min_run_;
+  int run_of_visible_frames_;
+};
+
+TEST_P(ArfFreqTest, MinArfFreqTest) {
+  cfg_.rc_target_bitrate = kBitrate;
+  cfg_.g_error_resilient = 0;
+  cfg_.g_profile = test_video_param_.profile;
+  cfg_.g_input_bit_depth = test_video_param_.input_bit_depth;
+  cfg_.g_bit_depth = test_video_param_.bit_depth;
+  init_flags_ = VPX_CODEC_USE_PSNR;
+  if (cfg_.g_bit_depth > 8)
+    init_flags_ |= VPX_CODEC_USE_HIGHBITDEPTH;
+
+  libvpx_test::VideoSource *video;
+  if (is_extension_y4m(test_video_param_.filename)) {
+    video = new libvpx_test::Y4mVideoSource(test_video_param_.filename,
+                                            0, kFrames);
+  } else {
+    video = new libvpx_test::YUVVideoSource(test_video_param_.filename,
+                                            test_video_param_.fmt,
+                                            test_video_param_.width,
+                                            test_video_param_.height,
+                                            test_video_param_.framerate_num,
+                                            test_video_param_.framerate_den,
+                                            0, kFrames);
+  }
+
+  ASSERT_NO_FATAL_FAILURE(RunLoop(video));
+  const int min_run = GetMinVisibleRun();
+  const int min_arf_dist_requested = GetMinArfDistanceRequested();
+  if (min_run != ARF_NOT_SEEN && min_run != ARF_SEEN_ONCE) {
+    const int min_arf_dist = min_run + 1;
+    EXPECT_GE(min_arf_dist, min_arf_dist_requested);
+  }
+  delete(video);
+}
+
+VP9_INSTANTIATE_TEST_CASE(
+    ArfFreqTest,
+    ::testing::ValuesIn(kTestVectors),
+    ::testing::ValuesIn(kEncodeVectors),
+    ::testing::ValuesIn(kMinArfVectors));
+
+VP10_INSTANTIATE_TEST_CASE(
+    ArfFreqTest,
+    ::testing::ValuesIn(kTestVectors),
+    ::testing::ValuesIn(kEncodeVectors),
+    ::testing::ValuesIn(kMinArfVectors));
+}  // namespace
--- a/test/vp9_avg_test.cc
+++ b/test/vp9_avg_test.cc
@@ -8,22 +8,22 @@
 *  be found in the AUTHORS file in the root of the source tree.
 */

-
-#include <string.h>
 #include <limits.h>
 #include <stdio.h>
+#include <string.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"

 #include "./vpx_config.h"
 #if CONFIG_VP9_ENCODER
 #include "./vp9_rtcd.h"
 #endif
-#include "vpx_mem/vpx_mem.h"

 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "vpx_mem/vpx_mem.h"

 using libvpx_test::ACMRandom;

@@ -121,6 +121,79 @@ class AverageTest
  }
 };

+typedef void (*IntProRowFunc)(int16_t hbuf[16], uint8_t const *ref,
+                              const int ref_stride, const int height);
+
+typedef std::tr1::tuple<int, IntProRowFunc, IntProRowFunc> IntProRowParam;
+
+class IntProRowTest
+    : public AverageTestBase,
+      public ::testing::WithParamInterface<IntProRowParam> {
+ public:
+  IntProRowTest()
+    : AverageTestBase(16, GET_PARAM(0)),
+      hbuf_asm_(NULL),
+      hbuf_c_(NULL) {
+    asm_func_ = GET_PARAM(1);
+    c_func_ = GET_PARAM(2);
+  }
+
+ protected:
+  virtual void SetUp() {
+    hbuf_asm_ = reinterpret_cast<int16_t*>(
+        vpx_memalign(kDataAlignment, sizeof(*hbuf_asm_) * 16));
+    hbuf_c_ = reinterpret_cast<int16_t*>(
+        vpx_memalign(kDataAlignment, sizeof(*hbuf_c_) * 16));
+  }
+
+  virtual void TearDown() {
+    vpx_free(hbuf_c_);
+    hbuf_c_ = NULL;
+    vpx_free(hbuf_asm_);
+    hbuf_asm_ = NULL;
+  }
+
+  void RunComparison() {
+    ASM_REGISTER_STATE_CHECK(c_func_(hbuf_c_, source_data_, 0, height_));
+    ASM_REGISTER_STATE_CHECK(asm_func_(hbuf_asm_, source_data_, 0, height_));
+    EXPECT_EQ(0, memcmp(hbuf_c_, hbuf_asm_, sizeof(*hbuf_c_) * 16))
+        << "Output mismatch";
+  }
+
+ private:
+  IntProRowFunc asm_func_;
+  IntProRowFunc c_func_;
+  int16_t *hbuf_asm_;
+  int16_t *hbuf_c_;
+};
+
+typedef int16_t (*IntProColFunc)(uint8_t const *ref, const int width);
+
+typedef std::tr1::tuple<int, IntProColFunc, IntProColFunc> IntProColParam;
+
+class IntProColTest
+    : public AverageTestBase,
+      public ::testing::WithParamInterface<IntProColParam> {
+ public:
+  IntProColTest() : AverageTestBase(GET_PARAM(0), 1), sum_asm_(0), sum_c_(0) {
+    asm_func_ = GET_PARAM(1);
+    c_func_ = GET_PARAM(2);
+  }
+
+ protected:
+  void RunComparison() {
+    ASM_REGISTER_STATE_CHECK(sum_c_ = c_func_(source_data_, width_));
+    ASM_REGISTER_STATE_CHECK(sum_asm_ = asm_func_(source_data_, width_));
+    EXPECT_EQ(sum_c_, sum_asm_) << "Output mismatch";
+  }
+
+ private:
+  IntProColFunc asm_func_;
+  IntProColFunc c_func_;
+  int16_t sum_asm_;
+  int16_t sum_c_;
+};
+

 uint8_t* AverageTestBase::source_data_ = NULL;

@@ -143,6 +216,36 @@ TEST_P(AverageTest, Random) {
  }
 }

+TEST_P(IntProRowTest, MinValue) {
+  FillConstant(0);
+  RunComparison();
+}
+
+TEST_P(IntProRowTest, MaxValue) {
+  FillConstant(255);
+  RunComparison();
+}
+
+TEST_P(IntProRowTest, Random) {
+  FillRandom();
+  RunComparison();
+}
+
+TEST_P(IntProColTest, MinValue) {
+  FillConstant(0);
+  RunComparison();
+}
+
+TEST_P(IntProColTest, MaxValue) {
+  FillConstant(255);
+  RunComparison();
+}
+
+TEST_P(IntProColTest, Random) {
+  FillRandom();
+  RunComparison();
+}
+
 using std::tr1::make_tuple;

 INSTANTIATE_TEST_CASE_P(
@@ -151,7 +254,6 @@ INSTANTIATE_TEST_CASE_P(
        make_tuple(16, 16, 1, 8, &vp9_avg_8x8_c),
        make_tuple(16, 16, 1, 4, &vp9_avg_4x4_c)));

-
 #if HAVE_SSE2
 INSTANTIATE_TEST_CASE_P(
    SSE2, AverageTest,
@@ -163,6 +265,17 @@ INSTANTIATE_TEST_CASE_P(
        make_tuple(16, 16, 5, 4, &vp9_avg_4x4_sse2),
        make_tuple(32, 32, 15, 4, &vp9_avg_4x4_sse2)));

+INSTANTIATE_TEST_CASE_P(
+    SSE2, IntProRowTest, ::testing::Values(
+        make_tuple(16, &vp9_int_pro_row_sse2, &vp9_int_pro_row_c),
+        make_tuple(32, &vp9_int_pro_row_sse2, &vp9_int_pro_row_c),
+        make_tuple(64, &vp9_int_pro_row_sse2, &vp9_int_pro_row_c)));
+
+INSTANTIATE_TEST_CASE_P(
+    SSE2, IntProColTest, ::testing::Values(
+        make_tuple(16, &vp9_int_pro_col_sse2, &vp9_int_pro_col_c),
+        make_tuple(32, &vp9_int_pro_col_sse2, &vp9_int_pro_col_c),
+        make_tuple(64, &vp9_int_pro_col_sse2, &vp9_int_pro_col_c)));
 #endif

 #if HAVE_NEON
@@ -173,6 +286,29 @@ INSTANTIATE_TEST_CASE_P(
        make_tuple(16, 16, 5, 8, &vp9_avg_8x8_neon),
        make_tuple(32, 32, 15, 8, &vp9_avg_8x8_neon)));

+INSTANTIATE_TEST_CASE_P(
+    NEON, IntProRowTest, ::testing::Values(
+        make_tuple(16, &vp9_int_pro_row_neon, &vp9_int_pro_row_c),
+        make_tuple(32, &vp9_int_pro_row_neon, &vp9_int_pro_row_c),
+        make_tuple(64, &vp9_int_pro_row_neon, &vp9_int_pro_row_c)));
+
+INSTANTIATE_TEST_CASE_P(
+    NEON, IntProColTest, ::testing::Values(
+        make_tuple(16, &vp9_int_pro_col_neon, &vp9_int_pro_col_c),
+        make_tuple(32, &vp9_int_pro_col_neon, &vp9_int_pro_col_c),
+        make_tuple(64, &vp9_int_pro_col_neon, &vp9_int_pro_col_c)));
+#endif
+
+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(
+    MSA, AverageTest,
+    ::testing::Values(
+        make_tuple(16, 16, 0, 8, &vp9_avg_8x8_msa),
+        make_tuple(16, 16, 5, 8, &vp9_avg_8x8_msa),
+        make_tuple(32, 32, 15, 8, &vp9_avg_8x8_msa),
+        make_tuple(16, 16, 0, 4, &vp9_avg_4x4_msa),
+        make_tuple(16, 16, 5, 4, &vp9_avg_4x4_msa),
+        make_tuple(32, 32, 15, 4, &vp9_avg_4x4_msa)));
 #endif

 }  // namespace
--- a/test/vp9_boolcoder_test.cc
+++ b/test/vp9_boolcoder_test.cc
@@ -14,11 +14,10 @@

 #include "third_party/googletest/src/include/gtest/gtest.h"

-#include "vp9/decoder/vp9_reader.h"
-#include "vp9/encoder/vp9_writer.h"
-
 #include "test/acm_random.h"
 #include "vpx/vpx_integer.h"
+#include "vpx_dsp/bitreader.h"
+#include "vpx_dsp/bitwriter.h"

 using libvpx_test::ACMRandom;

@@ -50,9 +49,9 @@ TEST(VP9, TestBitIO) {
        const int random_seed = 6432;
        const int kBufferSize = 10000;
        ACMRandom bit_rnd(random_seed);
-        vp9_writer bw;
+        vpx_writer bw;
        uint8_t bw_buffer[kBufferSize];
-        vp9_start_encode(&bw, bw_buffer);
+        vpx_start_encode(&bw, bw_buffer);

        int bit = (bit_method == 0) ? 0 : (bit_method == 1) ? 1 : 0;
        for (int i = 0; i < kBitsToTest; ++i) {
@@ -61,16 +60,16 @@ TEST(VP9, TestBitIO) {
          } else if (bit_method == 3) {
            bit = bit_rnd(2);
          }
-          vp9_write(&bw, bit, static_cast<int>(probas[i]));
+          vpx_write(&bw, bit, static_cast<int>(probas[i]));
        }

-        vp9_stop_encode(&bw);
+        vpx_stop_encode(&bw);

        // First bit should be zero
        GTEST_ASSERT_EQ(bw_buffer[0] & 0x80, 0);

-        vp9_reader br;
-        vp9_reader_init(&br, bw_buffer, kBufferSize, NULL, NULL);
+        vpx_reader br;
+        vpx_reader_init(&br, bw_buffer, kBufferSize, NULL, NULL);
        bit_rnd.Reset(random_seed);
        for (int i = 0; i < kBitsToTest; ++i) {
          if (bit_method == 2) {
@@ -78,7 +77,7 @@ TEST(VP9, TestBitIO) {
          } else if (bit_method == 3) {
            bit = bit_rnd(2);
          }
-          GTEST_ASSERT_EQ(vp9_read(&br, probas[i]), bit)
+          GTEST_ASSERT_EQ(vpx_read(&br, probas[i]), bit)
              << "pos: " << i << " / " << kBitsToTest
              << " bit_method: " << bit_method
              << " method: " << method;
--- a/test/vp9_denoiser_sse2_test.cc
+++ b/test/vp9_denoiser_sse2_test.cc
@@ -52,10 +52,10 @@ TEST_P(VP9DenoiserTest, BitexactCheck) {
  // mc_avg_block is the denoised reference block,
  // avg_block_c is the denoised result from C code,
  // avg_block_sse2 is the denoised result from SSE2 code.
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, sig_block, kNumPixels);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, mc_avg_block, kNumPixels);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, avg_block_c, kNumPixels);
-  DECLARE_ALIGNED_ARRAY(16, uint8_t, avg_block_sse2, kNumPixels);
+  DECLARE_ALIGNED(16, uint8_t, sig_block[kNumPixels]);
+  DECLARE_ALIGNED(16, uint8_t, mc_avg_block[kNumPixels]);
+  DECLARE_ALIGNED(16, uint8_t, avg_block_c[kNumPixels]);
+  DECLARE_ALIGNED(16, uint8_t, avg_block_sse2[kNumPixels]);

  for (int i = 0; i < count_test_block; ++i) {
    // Generate random motion magnitude, 20% of which exceed the threshold.
--- a/test/vp9_encoder_parms_get_to_decoder.cc
+++ b/test/vp9_encoder_parms_get_to_decoder.cc
@@ -8,12 +8,13 @@
 *  be found in the AUTHORS file in the root of the source tree.
 */

+#include "third_party/googletest/src/include/gtest/gtest.h"
+
 #include "test/codec_factory.h"
 #include "test/encode_test_driver.h"
+#include "test/util.h"
 #include "test/y4m_video_source.h"
 #include "test/yuv_video_source.h"
-#include "test/util.h"
-#include "third_party/googletest/src/include/gtest/gtest.h"
 #include "vp9/decoder/vp9_decoder.h"

 typedef vpx_codec_stream_info_t vp9_stream_info_t;
@@ -85,17 +86,17 @@ int is_extension_y4m(const char *filename) {
    return !strcmp(dot, ".y4m");
 }

-class Vp9EncoderParmsGetToDecoder
+class VpxEncoderParmsGetToDecoder
    : public ::libvpx_test::EncoderTest,
      public ::libvpx_test::CodecTestWith2Params<EncodeParameters, \
                                                 EncodePerfTestVideo> {
 protected:
-  Vp9EncoderParmsGetToDecoder()
+  VpxEncoderParmsGetToDecoder()
      : EncoderTest(GET_PARAM(0)),
        encode_parms(GET_PARAM(1)) {
  }

-  virtual ~Vp9EncoderParmsGetToDecoder() {}
+  virtual ~VpxEncoderParmsGetToDecoder() {}

  virtual void SetUp() {
    InitializeConfig();
@@ -165,7 +166,7 @@ class Vp9EncoderParmsGetToDecoder

 // TODO(hkuang): This test conflicts with frame parallel decode. So disable it
 // for now until fix.
-TEST_P(Vp9EncoderParmsGetToDecoder, DISABLED_BitstreamParms) {
+TEST_P(VpxEncoderParmsGetToDecoder, DISABLED_BitstreamParms) {
  init_flags_ = VPX_CODEC_USE_PSNR;

  libvpx_test::VideoSource *video;
@@ -186,8 +187,12 @@ TEST_P(Vp9EncoderParmsGetToDecoder, DISABLED_BitstreamParms) {
 }

 VP9_INSTANTIATE_TEST_CASE(
-    Vp9EncoderParmsGetToDecoder,
+    VpxEncoderParmsGetToDecoder,
    ::testing::ValuesIn(kVP9EncodeParameterSet),
    ::testing::ValuesIn(kVP9EncodePerfTestVectors));

+VP10_INSTANTIATE_TEST_CASE(
+    VpxEncoderParmsGetToDecoder,
+    ::testing::ValuesIn(kVP9EncodeParameterSet),
+    ::testing::ValuesIn(kVP9EncodePerfTestVectors));
 }  // namespace
--- a/test/vp9_end_to_end_test.cc
+++ b/test/vp9_end_to_end_test.cc
@@ -8,12 +8,13 @@
 *  be found in the AUTHORS file in the root of the source tree.
 */

+#include "third_party/googletest/src/include/gtest/gtest.h"
+
 #include "test/codec_factory.h"
 #include "test/encode_test_driver.h"
+#include "test/util.h"
 #include "test/y4m_video_source.h"
 #include "test/yuv_video_source.h"
-#include "test/util.h"
-#include "third_party/googletest/src/include/gtest/gtest.h"

 namespace {

@@ -186,4 +187,9 @@ VP9_INSTANTIATE_TEST_CASE(
    ::testing::ValuesIn(kTestVectors),
    ::testing::ValuesIn(kCpuUsedVectors));

+VP10_INSTANTIATE_TEST_CASE(
+    EndToEndTestLarge,
+    ::testing::ValuesIn(kEncodingModeVectors),
+    ::testing::ValuesIn(kTestVectors),
+    ::testing::ValuesIn(kCpuUsedVectors));
 }  // namespace
--- a/test/vp9_error_block_test.cc
+++ b/test/vp9_error_block_test.cc
@@ -14,13 +14,14 @@

 #include "third_party/googletest/src/include/gtest/gtest.h"

+#include "./vpx_config.h"
+#include "./vp9_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-#include "./vpx_config.h"
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
+#include "vpx/vpx_codec.h"
 #include "vpx/vpx_integer.h"

 using libvpx_test::ACMRandom;
@@ -57,8 +58,8 @@ class ErrorBlockTest

 TEST_P(ErrorBlockTest, OperationCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff,   4096);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, dqcoeff, 4096);
+  DECLARE_ALIGNED(16, tran_low_t, coeff[4096]);
+  DECLARE_ALIGNED(16, tran_low_t, dqcoeff[4096]);
  int err_count_total = 0;
  int first_failure = -1;
  intptr_t block_size;
@@ -90,8 +91,8 @@ TEST_P(ErrorBlockTest, OperationCheck) {

 TEST_P(ErrorBlockTest, ExtremeValues) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff,   4096);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, dqcoeff, 4096);
+  DECLARE_ALIGNED(16, tran_low_t, coeff[4096]);
+  DECLARE_ALIGNED(16, tran_low_t, dqcoeff[4096]);
  int err_count_total = 0;
  int first_failure = -1;
  intptr_t block_size;
--- a/test/vp9_ethread_test.cc
+++ b/test/vp9_ethread_test.cc
@@ -18,11 +18,11 @@
 #include "test/y4m_video_source.h"

 namespace {
-class VP9EncoderThreadTest
+class VPxEncoderThreadTest
    : public ::libvpx_test::EncoderTest,
      public ::libvpx_test::CodecTestWith2Params<libvpx_test::TestMode, int> {
 protected:
-  VP9EncoderThreadTest()
+  VPxEncoderThreadTest()
      : EncoderTest(GET_PARAM(0)),
        encoder_initialized_(false),
        tiles_(2),
@@ -36,7 +36,7 @@ class VP9EncoderThreadTest

    md5_.clear();
  }
-  virtual ~VP9EncoderThreadTest() {
+  virtual ~VPxEncoderThreadTest() {
    delete decoder_;
  }

@@ -105,7 +105,7 @@ class VP9EncoderThreadTest
  std::vector<std::string> md5_;
 };

-TEST_P(VP9EncoderThreadTest, EncoderResultTest) {
+TEST_P(VPxEncoderThreadTest, EncoderResultTest) {
  std::vector<std::string> single_thr_md5, multi_thr_md5;

  ::libvpx_test::Y4mVideoSource video("niklas_1280_720_30.y4m", 15, 20);
@@ -130,8 +130,13 @@ TEST_P(VP9EncoderThreadTest, EncoderResultTest) {
 }

 VP9_INSTANTIATE_TEST_CASE(
-    VP9EncoderThreadTest,
+    VPxEncoderThreadTest,
    ::testing::Values(::libvpx_test::kTwoPassGood, ::libvpx_test::kOnePassGood,
                      ::libvpx_test::kRealTime),
    ::testing::Range(1, 9));
+
+VP10_INSTANTIATE_TEST_CASE(
+    VPxEncoderThreadTest,
+    ::testing::Values(::libvpx_test::kTwoPassGood, ::libvpx_test::kOnePassGood),
+    ::testing::Range(1, 3));
 }  // namespace
--- a/test/vp9_frame_parallel_test.cc
+++ b/test/vp9_frame_parallel_test.cc
@@ -29,7 +29,7 @@ using std::string;

 #if CONFIG_WEBM_IO

-struct FileList {
+struct PauseFileList {
  const char *name;
  // md5 sum for decoded frames which does not include skipped frames.
  const char *expected_md5;
@@ -39,7 +39,8 @@ struct FileList {
 // Decodes |filename| with |num_threads|. Pause at the specified frame_num,
 // seek to next key frame and then continue decoding until the end. Return
 // the md5 of the decoded frames which does not include skipped frames.
-string DecodeFile(const string &filename, int num_threads, int pause_num) {
+string DecodeFileWithPause(const string &filename, int num_threads,
+                           int pause_num) {
  libvpx_test::WebMVideoSource video(filename);
  video.Init();
  int in_frames = 0;
@@ -92,12 +93,12 @@ string DecodeFile(const string &filename, int num_threads, int pause_num) {
  return string(md5.Get());
 }

-void DecodeFiles(const FileList files[]) {
-  for (const FileList *iter = files; iter->name != NULL; ++iter) {
+void DecodeFilesWithPause(const PauseFileList files[]) {
+  for (const PauseFileList *iter = files; iter->name != NULL; ++iter) {
    SCOPED_TRACE(iter->name);
    for (int t = 2; t <= 8; ++t) {
      EXPECT_EQ(iter->expected_md5,
-                DecodeFile(iter->name, t, iter->pause_frame_num))
+                DecodeFileWithPause(iter->name, t, iter->pause_frame_num))
          << "threads = " << t;
    }
  }
@@ -106,19 +107,19 @@ void DecodeFiles(const FileList files[]) {
 TEST(VP9MultiThreadedFrameParallel, PauseSeekResume) {
  // vp90-2-07-frame_parallel-1.webm is a 40 frame video file with
  // one key frame for every ten frames.
-  static const FileList files[] = {
+  static const PauseFileList files[] = {
    { "vp90-2-07-frame_parallel-1.webm",
-      "6ea7c3875d67252e7caf2bc6e75b36b1", 6},
+      "6ea7c3875d67252e7caf2bc6e75b36b1", 6 },
    { "vp90-2-07-frame_parallel-1.webm",
-      "4bb634160c7356a8d7d4299b6dc83a45", 12},
+      "4bb634160c7356a8d7d4299b6dc83a45", 12 },
    { "vp90-2-07-frame_parallel-1.webm",
-      "89772591e6ef461f9fa754f916c78ed8", 26},
-    { NULL, NULL, 0},
+      "89772591e6ef461f9fa754f916c78ed8", 26 },
+    { NULL, NULL, 0 },
  };
-  DecodeFiles(files);
+  DecodeFilesWithPause(files);
 }

-struct InvalidFileList {
+struct FileList {
  const char *name;
  // md5 sum for decoded frames which does not include corrupted frames.
  const char *expected_md5;
@@ -128,8 +129,8 @@ struct InvalidFileList {

 // Decodes |filename| with |num_threads|. Return the md5 of the decoded
 // frames which does not include corrupted frames.
-string DecodeInvalidFile(const string &filename, int num_threads,
-                         int expected_frame_count) {
+string DecodeFile(const string &filename, int num_threads,
+                  int expected_frame_count) {
  libvpx_test::WebMVideoSource video(filename);
  video.Init();

@@ -173,37 +174,47 @@ string DecodeInvalidFile(const string &filename, int num_threads,
  return string(md5.Get());
 }

-void DecodeInvalidFiles(const InvalidFileList files[]) {
-  for (const InvalidFileList *iter = files; iter->name != NULL; ++iter) {
+void DecodeFiles(const FileList files[]) {
+  for (const FileList *iter = files; iter->name != NULL; ++iter) {
    SCOPED_TRACE(iter->name);
    for (int t = 2; t <= 8; ++t) {
      EXPECT_EQ(iter->expected_md5,
-                DecodeInvalidFile(iter->name, t, iter->expected_frame_count))
+                DecodeFile(iter->name, t, iter->expected_frame_count))
          << "threads = " << t;
    }
  }
 }

 TEST(VP9MultiThreadedFrameParallel, InvalidFileTest) {
-  static const InvalidFileList files[] = {
+  static const FileList files[] = {
    // invalid-vp90-2-07-frame_parallel-1.webm is a 40 frame video file with
    // one key frame for every ten frames. The 11th frame has corrupted data.
    { "invalid-vp90-2-07-frame_parallel-1.webm",
-      "0549d0f45f60deaef8eb708e6c0eb6cb", 30},
+      "0549d0f45f60deaef8eb708e6c0eb6cb", 30 },
    // invalid-vp90-2-07-frame_parallel-2.webm is a 40 frame video file with
    // one key frame for every ten frames. The 1st and 31st frames have
    // corrupted data.
    { "invalid-vp90-2-07-frame_parallel-2.webm",
-      "6a1f3cf6f9e7a364212fadb9580d525e", 20},
+      "6a1f3cf6f9e7a364212fadb9580d525e", 20 },
    // invalid-vp90-2-07-frame_parallel-3.webm is a 40 frame video file with
    // one key frame for every ten frames. The 5th and 13th frames have
    // corrupted data.
    { "invalid-vp90-2-07-frame_parallel-3.webm",
-      "8256544308de926b0681e04685b98677", 27},
-    { NULL, NULL, 0},
+      "8256544308de926b0681e04685b98677", 27 },
+    { NULL, NULL, 0 },
  };
-  DecodeInvalidFiles(files);
+  DecodeFiles(files);
 }

+TEST(VP9MultiThreadedFrameParallel, ValidFileTest) {
+  static const FileList files[] = {
+#if CONFIG_VP9_HIGHBITDEPTH
+    { "vp92-2-20-10bit-yuv420.webm",
+      "a16b99df180c584e8db2ffeda987d293", 10 },
+#endif
+    { NULL, NULL, 0 },
+  };
+  DecodeFiles(files);
+}
 #endif  // CONFIG_WEBM_IO
 }  // namespace
--- a/test/vp9_intrapred_test.cc
+++ b/test/vp9_intrapred_test.cc
@@ -10,17 +10,17 @@

 #include <string>

-#include "test/acm_random.h"
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
 #include "third_party/googletest/src/include/gtest/gtest.h"

 #include "./vpx_config.h"
-#include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
+#include "test/util.h"
 #include "vp9/common/vp9_blockd.h"
 #include "vp9/common/vp9_pred_common.h"
 #include "vpx_mem/vpx_mem.h"
-#include "test/util.h"

 namespace {

@@ -120,10 +120,10 @@ class VP9IntraPredTest

 TEST_P(VP9IntraPredTest, IntraPredTests) {
  // max block size is 32
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, left_col, 2*32);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, above_data, 2*32+32);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, dst, 3 * 32 * 32);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_dst, 3 * 32 * 32);
+  DECLARE_ALIGNED(16, uint16_t, left_col[2*32]);
+  DECLARE_ALIGNED(16, uint16_t, above_data[2*32+32]);
+  DECLARE_ALIGNED(16, uint16_t, dst[3 * 32 * 32]);
+  DECLARE_ALIGNED(16, uint16_t, ref_dst[3 * 32 * 32]);
  RunTest(left_col, above_data, dst, ref_dst);
 }

@@ -131,168 +131,171 @@ using std::tr1::make_tuple;

 #if HAVE_SSE2
 #if CONFIG_VP9_HIGHBITDEPTH
+#if CONFIG_USE_X86INC
 #if ARCH_X86_64
 INSTANTIATE_TEST_CASE_P(SSE2_TO_C_8, VP9IntraPredTest,
                        ::testing::Values(
-                            make_tuple(&vp9_highbd_dc_predictor_32x32_sse2,
-                                       &vp9_highbd_dc_predictor_32x32_c, 32, 8),
-                            make_tuple(&vp9_highbd_tm_predictor_16x16_sse2,
-                                       &vp9_highbd_tm_predictor_16x16_c, 16, 8),
-                            make_tuple(&vp9_highbd_tm_predictor_32x32_sse2,
-                                       &vp9_highbd_tm_predictor_32x32_c, 32, 8),
-                            make_tuple(&vp9_highbd_dc_predictor_4x4_sse,
-                                       &vp9_highbd_dc_predictor_4x4_c, 4, 8),
-                            make_tuple(&vp9_highbd_dc_predictor_8x8_sse2,
-                                       &vp9_highbd_dc_predictor_8x8_c, 8, 8),
-                            make_tuple(&vp9_highbd_dc_predictor_16x16_sse2,
-                                       &vp9_highbd_dc_predictor_16x16_c, 16, 8),
-                            make_tuple(&vp9_highbd_v_predictor_4x4_sse,
-                                       &vp9_highbd_v_predictor_4x4_c, 4, 8),
-                            make_tuple(&vp9_highbd_v_predictor_8x8_sse2,
-                                       &vp9_highbd_v_predictor_8x8_c, 8, 8),
-                            make_tuple(&vp9_highbd_v_predictor_16x16_sse2,
-                                       &vp9_highbd_v_predictor_16x16_c, 16, 8),
-                            make_tuple(&vp9_highbd_v_predictor_32x32_sse2,
-                                       &vp9_highbd_v_predictor_32x32_c, 32, 8),
-                            make_tuple(&vp9_highbd_tm_predictor_4x4_sse,
-                                       &vp9_highbd_tm_predictor_4x4_c, 4, 8),
-                            make_tuple(&vp9_highbd_tm_predictor_8x8_sse2,
-                                       &vp9_highbd_tm_predictor_8x8_c, 8, 8)));
+                            make_tuple(&vpx_highbd_dc_predictor_32x32_sse2,
+                                       &vpx_highbd_dc_predictor_32x32_c, 32, 8),
+                            make_tuple(&vpx_highbd_tm_predictor_16x16_sse2,
+                                       &vpx_highbd_tm_predictor_16x16_c, 16, 8),
+                            make_tuple(&vpx_highbd_tm_predictor_32x32_sse2,
+                                       &vpx_highbd_tm_predictor_32x32_c, 32, 8),
+                            make_tuple(&vpx_highbd_dc_predictor_4x4_sse,
+                                       &vpx_highbd_dc_predictor_4x4_c, 4, 8),
+                            make_tuple(&vpx_highbd_dc_predictor_8x8_sse2,
+                                       &vpx_highbd_dc_predictor_8x8_c, 8, 8),
+                            make_tuple(&vpx_highbd_dc_predictor_16x16_sse2,
+                                       &vpx_highbd_dc_predictor_16x16_c, 16, 8),
+                            make_tuple(&vpx_highbd_v_predictor_4x4_sse,
+                                       &vpx_highbd_v_predictor_4x4_c, 4, 8),
+                            make_tuple(&vpx_highbd_v_predictor_8x8_sse2,
+                                       &vpx_highbd_v_predictor_8x8_c, 8, 8),
+                            make_tuple(&vpx_highbd_v_predictor_16x16_sse2,
+                                       &vpx_highbd_v_predictor_16x16_c, 16, 8),
+                            make_tuple(&vpx_highbd_v_predictor_32x32_sse2,
+                                       &vpx_highbd_v_predictor_32x32_c, 32, 8),
+                            make_tuple(&vpx_highbd_tm_predictor_4x4_sse,
+                                       &vpx_highbd_tm_predictor_4x4_c, 4, 8),
+                            make_tuple(&vpx_highbd_tm_predictor_8x8_sse2,
+                                       &vpx_highbd_tm_predictor_8x8_c, 8, 8)));
 #else
 INSTANTIATE_TEST_CASE_P(SSE2_TO_C_8, VP9IntraPredTest,
                        ::testing::Values(
-                            make_tuple(&vp9_highbd_dc_predictor_4x4_sse,
-                                       &vp9_highbd_dc_predictor_4x4_c, 4, 8),
-                            make_tuple(&vp9_highbd_dc_predictor_8x8_sse2,
-                                       &vp9_highbd_dc_predictor_8x8_c, 8, 8),
-                            make_tuple(&vp9_highbd_dc_predictor_16x16_sse2,
-                                       &vp9_highbd_dc_predictor_16x16_c, 16, 8),
-                            make_tuple(&vp9_highbd_v_predictor_4x4_sse,
-                                       &vp9_highbd_v_predictor_4x4_c, 4, 8),
-                            make_tuple(&vp9_highbd_v_predictor_8x8_sse2,
-                                       &vp9_highbd_v_predictor_8x8_c, 8, 8),
-                            make_tuple(&vp9_highbd_v_predictor_16x16_sse2,
-                                       &vp9_highbd_v_predictor_16x16_c, 16, 8),
-                            make_tuple(&vp9_highbd_v_predictor_32x32_sse2,
-                                       &vp9_highbd_v_predictor_32x32_c, 32, 8),
-                            make_tuple(&vp9_highbd_tm_predictor_4x4_sse,
-                                       &vp9_highbd_tm_predictor_4x4_c, 4, 8),
-                            make_tuple(&vp9_highbd_tm_predictor_8x8_sse2,
-                                       &vp9_highbd_tm_predictor_8x8_c, 8, 8)));
-#endif
+                            make_tuple(&vpx_highbd_dc_predictor_4x4_sse,
+                                       &vpx_highbd_dc_predictor_4x4_c, 4, 8),
+                            make_tuple(&vpx_highbd_dc_predictor_8x8_sse2,
+                                       &vpx_highbd_dc_predictor_8x8_c, 8, 8),
+                            make_tuple(&vpx_highbd_dc_predictor_16x16_sse2,
+                                       &vpx_highbd_dc_predictor_16x16_c, 16, 8),
+                            make_tuple(&vpx_highbd_v_predictor_4x4_sse,
+                                       &vpx_highbd_v_predictor_4x4_c, 4, 8),
+                            make_tuple(&vpx_highbd_v_predictor_8x8_sse2,
+                                       &vpx_highbd_v_predictor_8x8_c, 8, 8),
+                            make_tuple(&vpx_highbd_v_predictor_16x16_sse2,
+                                       &vpx_highbd_v_predictor_16x16_c, 16, 8),
+                            make_tuple(&vpx_highbd_v_predictor_32x32_sse2,
+                                       &vpx_highbd_v_predictor_32x32_c, 32, 8),
+                            make_tuple(&vpx_highbd_tm_predictor_4x4_sse,
+                                       &vpx_highbd_tm_predictor_4x4_c, 4, 8),
+                            make_tuple(&vpx_highbd_tm_predictor_8x8_sse2,
+                                       &vpx_highbd_tm_predictor_8x8_c, 8, 8)));
+#endif  // !ARCH_X86_64
+
 #if ARCH_X86_64
 INSTANTIATE_TEST_CASE_P(SSE2_TO_C_10, VP9IntraPredTest,
                        ::testing::Values(
-                            make_tuple(&vp9_highbd_dc_predictor_32x32_sse2,
-                                       &vp9_highbd_dc_predictor_32x32_c, 32,
+                            make_tuple(&vpx_highbd_dc_predictor_32x32_sse2,
+                                       &vpx_highbd_dc_predictor_32x32_c, 32,
                                       10),
-                            make_tuple(&vp9_highbd_tm_predictor_16x16_sse2,
-                                       &vp9_highbd_tm_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_tm_predictor_16x16_sse2,
+                                       &vpx_highbd_tm_predictor_16x16_c, 16,
                                       10),
-                            make_tuple(&vp9_highbd_tm_predictor_32x32_sse2,
-                                       &vp9_highbd_tm_predictor_32x32_c, 32,
+                            make_tuple(&vpx_highbd_tm_predictor_32x32_sse2,
+                                       &vpx_highbd_tm_predictor_32x32_c, 32,
                                       10),
-                            make_tuple(&vp9_highbd_dc_predictor_4x4_sse,
-                                       &vp9_highbd_dc_predictor_4x4_c, 4, 10),
-                            make_tuple(&vp9_highbd_dc_predictor_8x8_sse2,
-                                       &vp9_highbd_dc_predictor_8x8_c, 8, 10),
-                            make_tuple(&vp9_highbd_dc_predictor_16x16_sse2,
-                                       &vp9_highbd_dc_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_dc_predictor_4x4_sse,
+                                       &vpx_highbd_dc_predictor_4x4_c, 4, 10),
+                            make_tuple(&vpx_highbd_dc_predictor_8x8_sse2,
+                                       &vpx_highbd_dc_predictor_8x8_c, 8, 10),
+                            make_tuple(&vpx_highbd_dc_predictor_16x16_sse2,
+                                       &vpx_highbd_dc_predictor_16x16_c, 16,
                                       10),
-                            make_tuple(&vp9_highbd_v_predictor_4x4_sse,
-                                       &vp9_highbd_v_predictor_4x4_c, 4, 10),
-                            make_tuple(&vp9_highbd_v_predictor_8x8_sse2,
-                                       &vp9_highbd_v_predictor_8x8_c, 8, 10),
-                            make_tuple(&vp9_highbd_v_predictor_16x16_sse2,
-                                       &vp9_highbd_v_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_v_predictor_4x4_sse,
+                                       &vpx_highbd_v_predictor_4x4_c, 4, 10),
+                            make_tuple(&vpx_highbd_v_predictor_8x8_sse2,
+                                       &vpx_highbd_v_predictor_8x8_c, 8, 10),
+                            make_tuple(&vpx_highbd_v_predictor_16x16_sse2,
+                                       &vpx_highbd_v_predictor_16x16_c, 16,
                                       10),
-                            make_tuple(&vp9_highbd_v_predictor_32x32_sse2,
-                                       &vp9_highbd_v_predictor_32x32_c, 32,
+                            make_tuple(&vpx_highbd_v_predictor_32x32_sse2,
+                                       &vpx_highbd_v_predictor_32x32_c, 32,
                                       10),
-                            make_tuple(&vp9_highbd_tm_predictor_4x4_sse,
-                                       &vp9_highbd_tm_predictor_4x4_c, 4, 10),
-                            make_tuple(&vp9_highbd_tm_predictor_8x8_sse2,
-                                       &vp9_highbd_tm_predictor_8x8_c, 8, 10)));
+                            make_tuple(&vpx_highbd_tm_predictor_4x4_sse,
+                                       &vpx_highbd_tm_predictor_4x4_c, 4, 10),
+                            make_tuple(&vpx_highbd_tm_predictor_8x8_sse2,
+                                       &vpx_highbd_tm_predictor_8x8_c, 8, 10)));
 #else
 INSTANTIATE_TEST_CASE_P(SSE2_TO_C_10, VP9IntraPredTest,
                        ::testing::Values(
-                            make_tuple(&vp9_highbd_dc_predictor_4x4_sse,
-                                       &vp9_highbd_dc_predictor_4x4_c, 4, 10),
-                            make_tuple(&vp9_highbd_dc_predictor_8x8_sse2,
-                                       &vp9_highbd_dc_predictor_8x8_c, 8, 10),
-                            make_tuple(&vp9_highbd_dc_predictor_16x16_sse2,
-                                       &vp9_highbd_dc_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_dc_predictor_4x4_sse,
+                                       &vpx_highbd_dc_predictor_4x4_c, 4, 10),
+                            make_tuple(&vpx_highbd_dc_predictor_8x8_sse2,
+                                       &vpx_highbd_dc_predictor_8x8_c, 8, 10),
+                            make_tuple(&vpx_highbd_dc_predictor_16x16_sse2,
+                                       &vpx_highbd_dc_predictor_16x16_c, 16,
                                       10),
-                            make_tuple(&vp9_highbd_v_predictor_4x4_sse,
-                                       &vp9_highbd_v_predictor_4x4_c, 4, 10),
-                            make_tuple(&vp9_highbd_v_predictor_8x8_sse2,
-                                       &vp9_highbd_v_predictor_8x8_c, 8, 10),
-                            make_tuple(&vp9_highbd_v_predictor_16x16_sse2,
-                                       &vp9_highbd_v_predictor_16x16_c, 16, 10),
-                            make_tuple(&vp9_highbd_v_predictor_32x32_sse2,
-                                       &vp9_highbd_v_predictor_32x32_c, 32, 10),
-                            make_tuple(&vp9_highbd_tm_predictor_4x4_sse,
-                                       &vp9_highbd_tm_predictor_4x4_c, 4, 10),
-                            make_tuple(&vp9_highbd_tm_predictor_8x8_sse2,
-                                       &vp9_highbd_tm_predictor_8x8_c, 8, 10)));
-#endif
+                            make_tuple(&vpx_highbd_v_predictor_4x4_sse,
+                                       &vpx_highbd_v_predictor_4x4_c, 4, 10),
+                            make_tuple(&vpx_highbd_v_predictor_8x8_sse2,
+                                       &vpx_highbd_v_predictor_8x8_c, 8, 10),
+                            make_tuple(&vpx_highbd_v_predictor_16x16_sse2,
+                                       &vpx_highbd_v_predictor_16x16_c, 16, 10),
+                            make_tuple(&vpx_highbd_v_predictor_32x32_sse2,
+                                       &vpx_highbd_v_predictor_32x32_c, 32, 10),
+                            make_tuple(&vpx_highbd_tm_predictor_4x4_sse,
+                                       &vpx_highbd_tm_predictor_4x4_c, 4, 10),
+                            make_tuple(&vpx_highbd_tm_predictor_8x8_sse2,
+                                       &vpx_highbd_tm_predictor_8x8_c, 8, 10)));
+#endif  // !ARCH_X86_64

 #if ARCH_X86_64
 INSTANTIATE_TEST_CASE_P(SSE2_TO_C_12, VP9IntraPredTest,
                        ::testing::Values(
-                            make_tuple(&vp9_highbd_dc_predictor_32x32_sse2,
-                                       &vp9_highbd_dc_predictor_32x32_c, 32,
+                            make_tuple(&vpx_highbd_dc_predictor_32x32_sse2,
+                                       &vpx_highbd_dc_predictor_32x32_c, 32,
                                       12),
-                            make_tuple(&vp9_highbd_tm_predictor_16x16_sse2,
-                                       &vp9_highbd_tm_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_tm_predictor_16x16_sse2,
+                                       &vpx_highbd_tm_predictor_16x16_c, 16,
                                       12),
-                            make_tuple(&vp9_highbd_tm_predictor_32x32_sse2,
-                                       &vp9_highbd_tm_predictor_32x32_c, 32,
+                            make_tuple(&vpx_highbd_tm_predictor_32x32_sse2,
+                                       &vpx_highbd_tm_predictor_32x32_c, 32,
                                       12),
-                            make_tuple(&vp9_highbd_dc_predictor_4x4_sse,
-                                       &vp9_highbd_dc_predictor_4x4_c, 4, 12),
-                            make_tuple(&vp9_highbd_dc_predictor_8x8_sse2,
-                                       &vp9_highbd_dc_predictor_8x8_c, 8, 12),
-                            make_tuple(&vp9_highbd_dc_predictor_16x16_sse2,
-                                       &vp9_highbd_dc_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_dc_predictor_4x4_sse,
+                                       &vpx_highbd_dc_predictor_4x4_c, 4, 12),
+                            make_tuple(&vpx_highbd_dc_predictor_8x8_sse2,
+                                       &vpx_highbd_dc_predictor_8x8_c, 8, 12),
+                            make_tuple(&vpx_highbd_dc_predictor_16x16_sse2,
+                                       &vpx_highbd_dc_predictor_16x16_c, 16,
                                       12),
-                            make_tuple(&vp9_highbd_v_predictor_4x4_sse,
-                                       &vp9_highbd_v_predictor_4x4_c, 4, 12),
-                            make_tuple(&vp9_highbd_v_predictor_8x8_sse2,
-                                       &vp9_highbd_v_predictor_8x8_c, 8, 12),
-                            make_tuple(&vp9_highbd_v_predictor_16x16_sse2,
-                                       &vp9_highbd_v_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_v_predictor_4x4_sse,
+                                       &vpx_highbd_v_predictor_4x4_c, 4, 12),
+                            make_tuple(&vpx_highbd_v_predictor_8x8_sse2,
+                                       &vpx_highbd_v_predictor_8x8_c, 8, 12),
+                            make_tuple(&vpx_highbd_v_predictor_16x16_sse2,
+                                       &vpx_highbd_v_predictor_16x16_c, 16,
                                       12),
-                            make_tuple(&vp9_highbd_v_predictor_32x32_sse2,
-                                       &vp9_highbd_v_predictor_32x32_c, 32,
+                            make_tuple(&vpx_highbd_v_predictor_32x32_sse2,
+                                       &vpx_highbd_v_predictor_32x32_c, 32,
                                       12),
-                            make_tuple(&vp9_highbd_tm_predictor_4x4_sse,
-                                       &vp9_highbd_tm_predictor_4x4_c, 4, 12),
-                            make_tuple(&vp9_highbd_tm_predictor_8x8_sse2,
-                                       &vp9_highbd_tm_predictor_8x8_c, 8, 12)));
+                            make_tuple(&vpx_highbd_tm_predictor_4x4_sse,
+                                       &vpx_highbd_tm_predictor_4x4_c, 4, 12),
+                            make_tuple(&vpx_highbd_tm_predictor_8x8_sse2,
+                                       &vpx_highbd_tm_predictor_8x8_c, 8, 12)));
 #else
 INSTANTIATE_TEST_CASE_P(SSE2_TO_C_12, VP9IntraPredTest,
                        ::testing::Values(
-                            make_tuple(&vp9_highbd_dc_predictor_4x4_sse,
-                                       &vp9_highbd_dc_predictor_4x4_c, 4, 12),
-                            make_tuple(&vp9_highbd_dc_predictor_8x8_sse2,
-                                       &vp9_highbd_dc_predictor_8x8_c, 8, 12),
-                            make_tuple(&vp9_highbd_dc_predictor_16x16_sse2,
-                                       &vp9_highbd_dc_predictor_16x16_c, 16,
+                            make_tuple(&vpx_highbd_dc_predictor_4x4_sse,
+                                       &vpx_highbd_dc_predictor_4x4_c, 4, 12),
+                            make_tuple(&vpx_highbd_dc_predictor_8x8_sse2,
+                                       &vpx_highbd_dc_predictor_8x8_c, 8, 12),
+                            make_tuple(&vpx_highbd_dc_predictor_16x16_sse2,
+                                       &vpx_highbd_dc_predictor_16x16_c, 16,
                                       12),
-                            make_tuple(&vp9_highbd_v_predictor_4x4_sse,
-                                       &vp9_highbd_v_predictor_4x4_c, 4, 12),
-                            make_tuple(&vp9_highbd_v_predictor_8x8_sse2,
-                                       &vp9_highbd_v_predictor_8x8_c, 8, 12),
-                            make_tuple(&vp9_highbd_v_predictor_16x16_sse2,
-                                       &vp9_highbd_v_predictor_16x16_c, 16, 12),
-                            make_tuple(&vp9_highbd_v_predictor_32x32_sse2,
-                                       &vp9_highbd_v_predictor_32x32_c, 32, 12),
-                            make_tuple(&vp9_highbd_tm_predictor_4x4_sse,
-                                       &vp9_highbd_tm_predictor_4x4_c, 4, 12),
-                            make_tuple(&vp9_highbd_tm_predictor_8x8_sse2,
-                                       &vp9_highbd_tm_predictor_8x8_c, 8, 12)));
-#endif
+                            make_tuple(&vpx_highbd_v_predictor_4x4_sse,
+                                       &vpx_highbd_v_predictor_4x4_c, 4, 12),
+                            make_tuple(&vpx_highbd_v_predictor_8x8_sse2,
+                                       &vpx_highbd_v_predictor_8x8_c, 8, 12),
+                            make_tuple(&vpx_highbd_v_predictor_16x16_sse2,
+                                       &vpx_highbd_v_predictor_16x16_c, 16, 12),
+                            make_tuple(&vpx_highbd_v_predictor_32x32_sse2,
+                                       &vpx_highbd_v_predictor_32x32_c, 32, 12),
+                            make_tuple(&vpx_highbd_tm_predictor_4x4_sse,
+                                       &vpx_highbd_tm_predictor_4x4_c, 4, 12),
+                            make_tuple(&vpx_highbd_tm_predictor_8x8_sse2,
+                                       &vpx_highbd_tm_predictor_8x8_c, 8, 12)));
+#endif  // !ARCH_X86_64
+#endif  // CONFIG_USE_X86INC
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif  // HAVE_SSE2
 }  // namespace
--- a/test/vp9_lossless_test.cc
+++ b/test/vp9_lossless_test.cc
@@ -7,8 +7,10 @@
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */
-#include "./vpx_config.h"
+
 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vpx_config.h"
 #include "test/codec_factory.h"
 #include "test/encode_test_driver.h"
 #include "test/i420_video_source.h"
@@ -125,4 +127,8 @@ VP9_INSTANTIATE_TEST_CASE(LosslessTest,
                          ::testing::Values(::libvpx_test::kRealTime,
                                            ::libvpx_test::kOnePassGood,
                                            ::libvpx_test::kTwoPassGood));
+
+VP10_INSTANTIATE_TEST_CASE(LosslessTest,
+                           ::testing::Values(::libvpx_test::kOnePassGood,
+                                             ::libvpx_test::kTwoPassGood));
 }  // namespace
--- a/test/vp9_quantize_test.cc
+++ b/test/vp9_quantize_test.cc
@@ -14,13 +14,15 @@

 #include "third_party/googletest/src/include/gtest/gtest.h"

+#include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
 #include "test/util.h"
-#include "./vpx_config.h"
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_entropy.h"
+#include "vp9/common/vp9_scan.h"
+#include "vpx/vpx_codec.h"
 #include "vpx/vpx_integer.h"

 using libvpx_test::ACMRandom;
@@ -80,18 +82,18 @@ class VP9Quantize32Test : public ::testing::TestWithParam<QuantizeParam> {

 TEST_P(VP9QuantizeTest, OperationCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, zbin_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, round_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_shift_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, qcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, dqcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_qcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_dqcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, dequant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, eob_ptr, 1);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_eob_ptr, 1);
+  DECLARE_ALIGNED(16, tran_low_t, coeff_ptr[256]);
+  DECLARE_ALIGNED(16, int16_t, zbin_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, round_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_shift_ptr[2]);
+  DECLARE_ALIGNED(16, tran_low_t, qcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, tran_low_t, dqcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_qcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_dqcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, int16_t, dequant_ptr[2]);
+  DECLARE_ALIGNED(16, uint16_t, eob_ptr[1]);
+  DECLARE_ALIGNED(16, uint16_t, ref_eob_ptr[1]);
  int err_count_total = 0;
  int first_failure = -1;
  for (int i = 0; i < number_of_iterations; ++i) {
@@ -139,18 +141,18 @@ TEST_P(VP9QuantizeTest, OperationCheck) {

 TEST_P(VP9Quantize32Test, OperationCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, zbin_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, round_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_shift_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, qcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, dqcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_qcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_dqcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, dequant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, eob_ptr, 1);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_eob_ptr, 1);
+  DECLARE_ALIGNED(16, tran_low_t, coeff_ptr[1024]);
+  DECLARE_ALIGNED(16, int16_t, zbin_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, round_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_shift_ptr[2]);
+  DECLARE_ALIGNED(16, tran_low_t, qcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, tran_low_t, dqcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_qcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_dqcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, int16_t, dequant_ptr[2]);
+  DECLARE_ALIGNED(16, uint16_t, eob_ptr[1]);
+  DECLARE_ALIGNED(16, uint16_t, ref_eob_ptr[1]);
  int err_count_total = 0;
  int first_failure = -1;
  for (int i = 0; i < number_of_iterations; ++i) {
@@ -198,18 +200,18 @@ TEST_P(VP9Quantize32Test, OperationCheck) {

 TEST_P(VP9QuantizeTest, EOBCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, zbin_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, round_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_shift_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, qcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, dqcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_qcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_dqcoeff_ptr, 256);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, dequant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, eob_ptr, 1);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_eob_ptr, 1);
+  DECLARE_ALIGNED(16, tran_low_t, coeff_ptr[256]);
+  DECLARE_ALIGNED(16, int16_t, zbin_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, round_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_shift_ptr[2]);
+  DECLARE_ALIGNED(16, tran_low_t, qcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, tran_low_t, dqcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_qcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_dqcoeff_ptr[256]);
+  DECLARE_ALIGNED(16, int16_t, dequant_ptr[2]);
+  DECLARE_ALIGNED(16, uint16_t, eob_ptr[1]);
+  DECLARE_ALIGNED(16, uint16_t, ref_eob_ptr[1]);
  int err_count_total = 0;
  int first_failure = -1;
  for (int i = 0; i < number_of_iterations; ++i) {
@@ -262,18 +264,18 @@ TEST_P(VP9QuantizeTest, EOBCheck) {

 TEST_P(VP9Quantize32Test, EOBCheck) {
  ACMRandom rnd(ACMRandom::DeterministicSeed());
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, coeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, zbin_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, round_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, quant_shift_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, qcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, dqcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_qcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, tran_low_t, ref_dqcoeff_ptr, 1024);
-  DECLARE_ALIGNED_ARRAY(16, int16_t, dequant_ptr, 2);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, eob_ptr, 1);
-  DECLARE_ALIGNED_ARRAY(16, uint16_t, ref_eob_ptr, 1);
+  DECLARE_ALIGNED(16, tran_low_t, coeff_ptr[1024]);
+  DECLARE_ALIGNED(16, int16_t, zbin_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, round_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_ptr[2]);
+  DECLARE_ALIGNED(16, int16_t, quant_shift_ptr[2]);
+  DECLARE_ALIGNED(16, tran_low_t, qcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, tran_low_t, dqcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_qcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, tran_low_t, ref_dqcoeff_ptr[1024]);
+  DECLARE_ALIGNED(16, int16_t, dequant_ptr[2]);
+  DECLARE_ALIGNED(16, uint16_t, eob_ptr[1]);
+  DECLARE_ALIGNED(16, uint16_t, ref_eob_ptr[1]);
  int err_count_total = 0;
  int first_failure = -1;
  for (int i = 0; i < number_of_iterations; ++i) {
@@ -329,21 +331,21 @@ using std::tr1::make_tuple;
 INSTANTIATE_TEST_CASE_P(
    SSE2, VP9QuantizeTest,
    ::testing::Values(
-        make_tuple(&vp9_highbd_quantize_b_sse2,
-                   &vp9_highbd_quantize_b_c, VPX_BITS_8),
-        make_tuple(&vp9_highbd_quantize_b_sse2,
-                   &vp9_highbd_quantize_b_c, VPX_BITS_10),
-        make_tuple(&vp9_highbd_quantize_b_sse2,
-                   &vp9_highbd_quantize_b_c, VPX_BITS_12)));
+        make_tuple(&vpx_highbd_quantize_b_sse2,
+                   &vpx_highbd_quantize_b_c, VPX_BITS_8),
+        make_tuple(&vpx_highbd_quantize_b_sse2,
+                   &vpx_highbd_quantize_b_c, VPX_BITS_10),
+        make_tuple(&vpx_highbd_quantize_b_sse2,
+                   &vpx_highbd_quantize_b_c, VPX_BITS_12)));
 INSTANTIATE_TEST_CASE_P(
    SSE2, VP9Quantize32Test,
    ::testing::Values(
-        make_tuple(&vp9_highbd_quantize_b_32x32_sse2,
-                   &vp9_highbd_quantize_b_32x32_c, VPX_BITS_8),
-        make_tuple(&vp9_highbd_quantize_b_32x32_sse2,
-                   &vp9_highbd_quantize_b_32x32_c, VPX_BITS_10),
-        make_tuple(&vp9_highbd_quantize_b_32x32_sse2,
-                   &vp9_highbd_quantize_b_32x32_c, VPX_BITS_12)));
+        make_tuple(&vpx_highbd_quantize_b_32x32_sse2,
+                   &vpx_highbd_quantize_b_32x32_c, VPX_BITS_8),
+        make_tuple(&vpx_highbd_quantize_b_32x32_sse2,
+                   &vpx_highbd_quantize_b_32x32_c, VPX_BITS_10),
+        make_tuple(&vpx_highbd_quantize_b_32x32_sse2,
+                   &vpx_highbd_quantize_b_32x32_c, VPX_BITS_12)));
 #endif  // HAVE_SSE2
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 }  // namespace
--- a/test/vp9_skip_loopfilter_test.cc
+++ b/test/vp9_skip_loopfilter_test.cc
@@ -0,0 +1,180 @@
+/*
+ *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <string>
+
+#include "test/codec_factory.h"
+#include "test/decode_test_driver.h"
+#include "test/md5_helper.h"
+#include "test/util.h"
+#include "test/webm_video_source.h"
+
+namespace {
+
+const char kVp9TestFile[] = "vp90-2-08-tile_1x8_frame_parallel.webm";
+const char kVp9Md5File[] = "vp90-2-08-tile_1x8_frame_parallel.webm.md5";
+
+// Class for testing shutting off the loop filter.
+class SkipLoopFilterTest {
+ public:
+  SkipLoopFilterTest()
+      : video_(NULL),
+        decoder_(NULL),
+        md5_file_(NULL) {}
+
+  ~SkipLoopFilterTest() {
+    if (md5_file_ != NULL)
+      fclose(md5_file_);
+    delete decoder_;
+    delete video_;
+  }
+
+  // If |threads| > 0 then set the decoder with that number of threads.
+  void Init(int num_threads) {
+    expected_md5_[0] = '\0';
+    junk_[0] = '\0';
+    video_ = new libvpx_test::WebMVideoSource(kVp9TestFile);
+    ASSERT_TRUE(video_ != NULL);
+    video_->Init();
+    video_->Begin();
+
+    vpx_codec_dec_cfg_t cfg = vpx_codec_dec_cfg_t();
+    if (num_threads > 0)
+      cfg.threads = num_threads;
+    decoder_ = new libvpx_test::VP9Decoder(cfg, 0);
+    ASSERT_TRUE(decoder_ != NULL);
+
+    OpenMd5File(kVp9Md5File);
+  }
+
+  // Set the VP9 skipLoopFilter control value.
+  void SetSkipLoopFilter(int value, vpx_codec_err_t expected_value) {
+    decoder_->Control(VP9_SET_SKIP_LOOP_FILTER, value, expected_value);
+  }
+
+  vpx_codec_err_t DecodeOneFrame() {
+    const vpx_codec_err_t res =
+        decoder_->DecodeFrame(video_->cxdata(), video_->frame_size());
+    if (res == VPX_CODEC_OK) {
+      ReadMd5();
+      video_->Next();
+    }
+    return res;
+  }
+
+  vpx_codec_err_t DecodeRemainingFrames() {
+    for (; video_->cxdata() != NULL; video_->Next()) {
+      const vpx_codec_err_t res =
+          decoder_->DecodeFrame(video_->cxdata(), video_->frame_size());
+      if (res != VPX_CODEC_OK)
+        return res;
+      ReadMd5();
+    }
+    return VPX_CODEC_OK;
+  }
+
+  // Checks if MD5 matches or doesn't.
+  void CheckMd5(bool matches) {
+    libvpx_test::DxDataIterator dec_iter = decoder_->GetDxData();
+    const vpx_image_t *img = dec_iter.Next();
+    CheckMd5Vpx(*img, matches);
+  }
+
+ private:
+  // TODO(fgalligan): Move the MD5 testing code into another class.
+  void OpenMd5File(const std::string &md5_file_name) {
+    md5_file_ = libvpx_test::OpenTestDataFile(md5_file_name);
+    ASSERT_TRUE(md5_file_ != NULL) << "MD5 file open failed. Filename: "
+        << md5_file_name;
+  }
+
+  // Reads the next line of the MD5 file.
+  void ReadMd5() {
+    ASSERT_TRUE(md5_file_ != NULL);
+    const int res = fscanf(md5_file_, "%s  %s", expected_md5_, junk_);
+    ASSERT_NE(EOF, res) << "Read md5 data failed";
+    expected_md5_[32] = '\0';
+  }
+
+  // Checks if the last read MD5 matches |img| or doesn't.
+  void CheckMd5Vpx(const vpx_image_t &img, bool matches) {
+    ::libvpx_test::MD5 md5_res;
+    md5_res.Add(&img);
+    const char *const actual_md5 = md5_res.Get();
+
+    // Check MD5.
+    if (matches)
+      ASSERT_STREQ(expected_md5_, actual_md5) << "MD5 checksums don't match";
+    else
+      ASSERT_STRNE(expected_md5_, actual_md5) << "MD5 checksums match";
+  }
+
+  libvpx_test::WebMVideoSource *video_;
+  libvpx_test::VP9Decoder *decoder_;
+  FILE *md5_file_;
+  char expected_md5_[33];
+  char junk_[128];
+};
+
+TEST(SkipLoopFilterTest, ShutOffLoopFilter) {
+  const int non_zero_value = 1;
+  const int num_threads = 0;
+  SkipLoopFilterTest skip_loop_filter;
+  skip_loop_filter.Init(num_threads);
+  skip_loop_filter.SetSkipLoopFilter(non_zero_value, VPX_CODEC_OK);
+  ASSERT_EQ(VPX_CODEC_OK, skip_loop_filter.DecodeRemainingFrames());
+  skip_loop_filter.CheckMd5(false);
+}
+
+TEST(SkipLoopFilterTest, ShutOffLoopFilterSingleThread) {
+  const int non_zero_value = 1;
+  const int num_threads = 1;
+  SkipLoopFilterTest skip_loop_filter;
+  skip_loop_filter.Init(num_threads);
+  skip_loop_filter.SetSkipLoopFilter(non_zero_value, VPX_CODEC_OK);
+  ASSERT_EQ(VPX_CODEC_OK, skip_loop_filter.DecodeRemainingFrames());
+  skip_loop_filter.CheckMd5(false);
+}
+
+TEST(SkipLoopFilterTest, ShutOffLoopFilter8Threads) {
+  const int non_zero_value = 1;
+  const int num_threads = 8;
+  SkipLoopFilterTest skip_loop_filter;
+  skip_loop_filter.Init(num_threads);
+  skip_loop_filter.SetSkipLoopFilter(non_zero_value, VPX_CODEC_OK);
+  ASSERT_EQ(VPX_CODEC_OK, skip_loop_filter.DecodeRemainingFrames());
+  skip_loop_filter.CheckMd5(false);
+}
+
+TEST(SkipLoopFilterTest, WithLoopFilter) {
+  const int non_zero_value = 1;
+  const int num_threads = 0;
+  SkipLoopFilterTest skip_loop_filter;
+  skip_loop_filter.Init(num_threads);
+  skip_loop_filter.SetSkipLoopFilter(non_zero_value, VPX_CODEC_OK);
+  skip_loop_filter.SetSkipLoopFilter(0, VPX_CODEC_OK);
+  ASSERT_EQ(VPX_CODEC_OK, skip_loop_filter.DecodeRemainingFrames());
+  skip_loop_filter.CheckMd5(true);
+}
+
+TEST(SkipLoopFilterTest, ToggleLoopFilter) {
+  const int num_threads = 0;
+  SkipLoopFilterTest skip_loop_filter;
+  skip_loop_filter.Init(num_threads);
+
+  for (int i = 0; i < 10; ++i) {
+    skip_loop_filter.SetSkipLoopFilter(i % 2, VPX_CODEC_OK);
+    ASSERT_EQ(VPX_CODEC_OK, skip_loop_filter.DecodeOneFrame());
+  }
+  ASSERT_EQ(VPX_CODEC_OK, skip_loop_filter.DecodeRemainingFrames());
+  skip_loop_filter.CheckMd5(false);
+}
+
+}  // namespace
--- a/test/vp9_subtract_test.cc
+++ b/test/vp9_subtract_test.cc
@@ -9,11 +9,13 @@
 */

 #include "third_party/googletest/src/include/gtest/gtest.h"
+
+#include "./vp9_rtcd.h"
+#include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "test/acm_random.h"
 #include "test/clear_system_state.h"
 #include "test/register_state_check.h"
-#include "./vpx_config.h"
-#include "./vp9_rtcd.h"
 #include "vp9/common/vp9_blockd.h"
 #include "vpx_mem/vpx_mem.h"

@@ -89,15 +91,19 @@ TEST_P(VP9SubtractBlockTest, SimpleSubtract) {
 }

 INSTANTIATE_TEST_CASE_P(C, VP9SubtractBlockTest,
-                        ::testing::Values(vp9_subtract_block_c));
+                        ::testing::Values(vpx_subtract_block_c));

 #if HAVE_SSE2 && CONFIG_USE_X86INC
 INSTANTIATE_TEST_CASE_P(SSE2, VP9SubtractBlockTest,
-                        ::testing::Values(vp9_subtract_block_sse2));
+                        ::testing::Values(vpx_subtract_block_sse2));
 #endif
 #if HAVE_NEON
 INSTANTIATE_TEST_CASE_P(NEON, VP9SubtractBlockTest,
-                        ::testing::Values(vp9_subtract_block_neon));
+                        ::testing::Values(vpx_subtract_block_neon));
+#endif
+#if HAVE_MSA
+INSTANTIATE_TEST_CASE_P(MSA, VP9SubtractBlockTest,
+                        ::testing::Values(vpx_subtract_block_msa));
 #endif

 }  // namespace vp9
--- a/test/vp9_thread_test.cc
+++ b/test/vp9_thread_test.cc
@@ -18,33 +18,33 @@
 #if CONFIG_WEBM_IO
 #include "test/webm_video_source.h"
 #endif
-#include "vp9/common/vp9_thread.h"
+#include "vpx_util/vpx_thread.h"

 namespace {

 using std::string;

-class VP9WorkerThreadTest : public ::testing::TestWithParam<bool> {
+class VPxWorkerThreadTest : public ::testing::TestWithParam<bool> {
 protected:
-  virtual ~VP9WorkerThreadTest() {}
+  virtual ~VPxWorkerThreadTest() {}
  virtual void SetUp() {
-    vp9_get_worker_interface()->init(&worker_);
+    vpx_get_worker_interface()->init(&worker_);
  }

  virtual void TearDown() {
-    vp9_get_worker_interface()->end(&worker_);
+    vpx_get_worker_interface()->end(&worker_);
  }

-  void Run(VP9Worker* worker) {
+  void Run(VPxWorker* worker) {
    const bool synchronous = GetParam();
    if (synchronous) {
-      vp9_get_worker_interface()->execute(worker);
+      vpx_get_worker_interface()->execute(worker);
    } else {
-      vp9_get_worker_interface()->launch(worker);
+      vpx_get_worker_interface()->launch(worker);
    }
  }

-  VP9Worker worker_;
+  VPxWorker worker_;
 };

 int ThreadHook(void* data, void* return_value) {
@@ -53,12 +53,12 @@ int ThreadHook(void* data, void* return_value) {
  return *reinterpret_cast<int*>(return_value);
 }

-TEST_P(VP9WorkerThreadTest, HookSuccess) {
+TEST_P(VPxWorkerThreadTest, HookSuccess) {
  // should be a no-op.
-  EXPECT_NE(vp9_get_worker_interface()->sync(&worker_), 0);
+  EXPECT_NE(vpx_get_worker_interface()->sync(&worker_), 0);

  for (int i = 0; i < 2; ++i) {
-    EXPECT_NE(vp9_get_worker_interface()->reset(&worker_), 0);
+    EXPECT_NE(vpx_get_worker_interface()->reset(&worker_), 0);

    int hook_data = 0;
    int return_value = 1;  // return successfully from the hook
@@ -67,17 +67,17 @@ TEST_P(VP9WorkerThreadTest, HookSuccess) {
    worker_.data2 = &return_value;

    Run(&worker_);
-    EXPECT_NE(vp9_get_worker_interface()->sync(&worker_), 0);
+    EXPECT_NE(vpx_get_worker_interface()->sync(&worker_), 0);
    EXPECT_FALSE(worker_.had_error);
    EXPECT_EQ(5, hook_data);

    // should be a no-op.
-    EXPECT_NE(vp9_get_worker_interface()->sync(&worker_), 0);
+    EXPECT_NE(vpx_get_worker_interface()->sync(&worker_), 0);
  }
 }

-TEST_P(VP9WorkerThreadTest, HookFailure) {
-  EXPECT_NE(vp9_get_worker_interface()->reset(&worker_), 0);
+TEST_P(VPxWorkerThreadTest, HookFailure) {
+  EXPECT_NE(vpx_get_worker_interface()->reset(&worker_), 0);

  int hook_data = 0;
  int return_value = 0;  // return failure from the hook
@@ -86,29 +86,29 @@ TEST_P(VP9WorkerThreadTest, HookFailure) {
  worker_.data2 = &return_value;

  Run(&worker_);
-  EXPECT_FALSE(vp9_get_worker_interface()->sync(&worker_));
+  EXPECT_FALSE(vpx_get_worker_interface()->sync(&worker_));
  EXPECT_EQ(1, worker_.had_error);

  // Ensure _reset() clears the error and _launch() can be called again.
  return_value = 1;
-  EXPECT_NE(vp9_get_worker_interface()->reset(&worker_), 0);
+  EXPECT_NE(vpx_get_worker_interface()->reset(&worker_), 0);
  EXPECT_FALSE(worker_.had_error);
-  vp9_get_worker_interface()->launch(&worker_);
-  EXPECT_NE(vp9_get_worker_interface()->sync(&worker_), 0);
+  vpx_get_worker_interface()->launch(&worker_);
+  EXPECT_NE(vpx_get_worker_interface()->sync(&worker_), 0);
  EXPECT_FALSE(worker_.had_error);
 }

-TEST_P(VP9WorkerThreadTest, EndWithoutSync) {
+TEST_P(VPxWorkerThreadTest, EndWithoutSync) {
  // Create a large number of threads to increase the chances of detecting a
  // race. Doing more work in the hook is no guarantee as any race would occur
  // post hook execution in the main thread loop driver.
  static const int kNumWorkers = 64;
-  VP9Worker workers[kNumWorkers];
+  VPxWorker workers[kNumWorkers];
  int hook_data[kNumWorkers];
  int return_value[kNumWorkers];

  for (int n = 0; n < kNumWorkers; ++n) {
-    vp9_get_worker_interface()->init(&workers[n]);
+    vpx_get_worker_interface()->init(&workers[n]);
    return_value[n] = 1;  // return successfully from the hook
    workers[n].hook = ThreadHook;
    workers[n].data1 = &hook_data[n];
@@ -117,7 +117,7 @@ TEST_P(VP9WorkerThreadTest, EndWithoutSync) {

  for (int i = 0; i < 2; ++i) {
    for (int n = 0; n < kNumWorkers; ++n) {
-      EXPECT_NE(vp9_get_worker_interface()->reset(&workers[n]), 0);
+      EXPECT_NE(vpx_get_worker_interface()->reset(&workers[n]), 0);
      hook_data[n] = 0;
    }

@@ -126,16 +126,16 @@ TEST_P(VP9WorkerThreadTest, EndWithoutSync) {
    }

    for (int n = kNumWorkers - 1; n >= 0; --n) {
-      vp9_get_worker_interface()->end(&workers[n]);
+      vpx_get_worker_interface()->end(&workers[n]);
    }
  }
 }

-TEST(VP9WorkerThreadTest, TestInterfaceAPI) {
-  EXPECT_EQ(0, vp9_set_worker_interface(NULL));
-  EXPECT_TRUE(vp9_get_worker_interface() != NULL);
+TEST(VPxWorkerThreadTest, TestInterfaceAPI) {
+  EXPECT_EQ(0, vpx_set_worker_interface(NULL));
+  EXPECT_TRUE(vpx_get_worker_interface() != NULL);
  for (int i = 0; i < 6; ++i) {
-    VP9WorkerInterface winterface = *vp9_get_worker_interface();
+    VPxWorkerInterface winterface = *vpx_get_worker_interface();
    switch (i) {
      default:
      case 0: winterface.init = NULL; break;
@@ -145,7 +145,7 @@ TEST(VP9WorkerThreadTest, TestInterfaceAPI) {
      case 4: winterface.execute = NULL; break;
      case 5: winterface.end = NULL; break;
    }
-    EXPECT_EQ(0, vp9_set_worker_interface(&winterface));
+    EXPECT_EQ(0, vpx_set_worker_interface(&winterface));
  }
 }

@@ -202,21 +202,21 @@ void DecodeFiles(const FileList files[]) {
 // hang.
 namespace impl {

-void Init(VP9Worker *const worker) { memset(worker, 0, sizeof(*worker)); }
-int Reset(VP9Worker *const /*worker*/) { return 1; }
-int Sync(VP9Worker *const worker) { return !worker->had_error; }
+void Init(VPxWorker *const worker) { memset(worker, 0, sizeof(*worker)); }
+int Reset(VPxWorker *const /*worker*/) { return 1; }
+int Sync(VPxWorker *const worker) { return !worker->had_error; }

-void Execute(VP9Worker *const worker) {
+void Execute(VPxWorker *const worker) {
  worker->had_error |= !worker->hook(worker->data1, worker->data2);
 }

-void Launch(VP9Worker *const worker) { Execute(worker); }
-void End(VP9Worker *const /*worker*/) {}
+void Launch(VPxWorker *const worker) { Execute(worker); }
+void End(VPxWorker *const /*worker*/) {}

 }  // namespace impl

-TEST(VP9WorkerThreadTest, TestSerialInterface) {
-  static const VP9WorkerInterface serial_interface = {
+TEST(VPxWorkerThreadTest, TestSerialInterface) {
+  static const VPxWorkerInterface serial_interface = {
    impl::Init, impl::Reset, impl::Sync, impl::Launch, impl::Execute, impl::End
  };
  // TODO(jzern): Avoid using a file that will use the row-based thread
@@ -225,13 +225,13 @@ TEST(VP9WorkerThreadTest, TestSerialInterface) {
  // progress in the row above before proceeding.
  static const char expected_md5[] = "b35a1b707b28e82be025d960aba039bc";
  static const char filename[] = "vp90-2-03-size-226x226.webm";
-  VP9WorkerInterface default_interface = *vp9_get_worker_interface();
+  VPxWorkerInterface default_interface = *vpx_get_worker_interface();

-  EXPECT_NE(vp9_set_worker_interface(&serial_interface), 0);
+  EXPECT_NE(vpx_set_worker_interface(&serial_interface), 0);
  EXPECT_EQ(expected_md5, DecodeFile(filename, 2));

  // Reset the interface.
-  EXPECT_NE(vp9_set_worker_interface(&default_interface), 0);
+  EXPECT_NE(vpx_set_worker_interface(&default_interface), 0);
  EXPECT_EQ(expected_md5, DecodeFile(filename, 2));
 }

@@ -309,6 +309,6 @@ TEST(VP9DecodeMultiThreadedTest, Decode3) {
 }
 #endif  // CONFIG_WEBM_IO

-INSTANTIATE_TEST_CASE_P(Synchronous, VP9WorkerThreadTest, ::testing::Bool());
+INSTANTIATE_TEST_CASE_P(Synchronous, VPxWorkerThreadTest, ::testing::Bool());

 }  // namespace
--- a/test/vpx_scale_test.cc
+++ b/test/vpx_scale_test.cc
@@ -10,11 +10,10 @@

 #include "third_party/googletest/src/include/gtest/gtest.h"

-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
-
 #include "./vpx_config.h"
 #include "./vpx_scale_rtcd.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
 #include "vpx_mem/vpx_mem.h"
 #include "vpx_scale/yv12config.h"

@@ -33,10 +32,10 @@ class VpxScaleBase {
  void ResetImage(int width, int height) {
    width_ = width;
    height_ = height;
-    vpx_memset(&img_, 0, sizeof(img_));
+    memset(&img_, 0, sizeof(img_));
    ASSERT_EQ(0, vp8_yv12_alloc_frame_buffer(&img_, width_, height_,
                                             VP8BORDERINPIXELS));
-    vpx_memset(img_.buffer_alloc, kBufFiller, img_.frame_size);
+    memset(img_.buffer_alloc, kBufFiller, img_.frame_size);
    FillPlane(img_.y_buffer, img_.y_crop_width, img_.y_crop_height,
              img_.y_stride);
    FillPlane(img_.u_buffer, img_.uv_crop_width, img_.uv_crop_height,
@@ -44,15 +43,15 @@ class VpxScaleBase {
    FillPlane(img_.v_buffer, img_.uv_crop_width, img_.uv_crop_height,
              img_.uv_stride);

-    vpx_memset(&ref_img_, 0, sizeof(ref_img_));
+    memset(&ref_img_, 0, sizeof(ref_img_));
    ASSERT_EQ(0, vp8_yv12_alloc_frame_buffer(&ref_img_, width_, height_,
                                             VP8BORDERINPIXELS));
-    vpx_memset(ref_img_.buffer_alloc, kBufFiller, ref_img_.frame_size);
+    memset(ref_img_.buffer_alloc, kBufFiller, ref_img_.frame_size);

-    vpx_memset(&cpy_img_, 0, sizeof(cpy_img_));
+    memset(&cpy_img_, 0, sizeof(cpy_img_));
    ASSERT_EQ(0, vp8_yv12_alloc_frame_buffer(&cpy_img_, width_, height_,
                                             VP8BORDERINPIXELS));
-    vpx_memset(cpy_img_.buffer_alloc, kBufFiller, cpy_img_.frame_size);
+    memset(cpy_img_.buffer_alloc, kBufFiller, cpy_img_.frame_size);
    ReferenceCopyFrame();
  }

@@ -87,8 +86,8 @@ class VpxScaleBase {

    // Fill the border pixels from the nearest image pixel.
    for (int y = 0; y < crop_height; ++y) {
-      vpx_memset(left, left[padding], padding);
-      vpx_memset(right, right[-1], right_extend);
+      memset(left, left[padding], padding);
+      memset(right, right[-1], right_extend);
      left += stride;
      right += stride;
    }
@@ -101,13 +100,13 @@ class VpxScaleBase {

    // The first row was already extended to the left and right. Copy it up.
    for (int y = 0; y < padding; ++y) {
-      vpx_memcpy(top, left, extend_width);
+      memcpy(top, left, extend_width);
      top += stride;
    }

    uint8_t *bottom = left + (crop_height * stride);
    for (int y = 0; y <  bottom_extend; ++y) {
-      vpx_memcpy(bottom, left + (crop_height - 1) * stride, extend_width);
+      memcpy(bottom, left + (crop_height - 1) * stride, extend_width);
      bottom += stride;
    }
  }
--- a/test/vpxdec.sh
+++ b/test/vpxdec.sh
@@ -17,7 +17,8 @@
 # Environment check: Make sure input is available.
 vpxdec_verify_environment() {
  if [ ! -e "${VP8_IVF_FILE}" ] || [ ! -e "${VP9_WEBM_FILE}" ] || \
-    [ ! -e "${VP9_FPM_WEBM_FILE}" ] ; then
+    [ ! -e "${VP9_FPM_WEBM_FILE}" ] || \
+    [ ! -e "${VP9_LT_50_FRAMES_WEBM_FILE}" ] ; then
    elog "Libvpx test data must exist in LIBVPX_TEST_DATA_PATH."
    return 1
  fi
@@ -87,12 +88,29 @@ vpxdec_vp9_webm_frame_parallel() {
        --frame-parallel
    done
  fi
+}

+vpxdec_vp9_webm_less_than_50_frames() {
+  # ensure that reaching eof in webm_guess_framerate doesn't result in invalid
+  # frames in actual webm_read_frame calls.
+  if [ "$(vpxdec_can_decode_vp9)" = "yes" ] && \
+     [ "$(webm_io_available)" = "yes" ]; then
+    local readonly decoder="$(vpx_tool_path vpxdec)"
+    local readonly expected=10
+    local readonly num_frames=$(${VPX_TEST_PREFIX} "${decoder}" \
+      "${VP9_LT_50_FRAMES_WEBM_FILE}" --summary --noblit 2>&1 \
+      | awk '/^[0-9]+ decoded frames/ { print $1 }')
+    if [ "$num_frames" -ne "$expected" ]; then
+      elog "Output frames ($num_frames) != expected ($expected)"
+      return 1
+    fi
+  fi
 }

 vpxdec_tests="vpxdec_vp8_ivf
              vpxdec_vp8_ivf_pipe_input
              vpxdec_vp9_webm
-              vpxdec_vp9_webm_frame_parallel"
+              vpxdec_vp9_webm_frame_parallel
+              vpxdec_vp9_webm_less_than_50_frames"

 run_tests vpxdec_verify_environment "${vpxdec_tests}"
--- a/test/vpxenc.sh
+++ b/test/vpxenc.sh
@@ -23,6 +23,13 @@ vpxenc_verify_environment() {
    elog "The file ${YUV_RAW_INPUT##*/} must exist in LIBVPX_TEST_DATA_PATH."
    return 1
  fi
+  if [ "$(vpxenc_can_encode_vp9)" = "yes" ]; then
+    if [ ! -e "${Y4M_NOSQ_PAR_INPUT}" ]; then
+      elog "The file ${Y4M_NOSQ_PAR_INPUT##*/} must exist in"
+      elog "LIBVPX_TEST_DATA_PATH."
+      return 1
+    fi
+  fi
  if [ -z "$(vpx_tool_path vpxenc)" ]; then
    elog "vpxenc not found. It must exist in LIBVPX_BIN_PATH or its parent."
    return 1
@@ -49,6 +56,14 @@ yuv_input_hantro_collage() {
       --height="${YUV_RAW_INPUT_HEIGHT}""
 }

+y4m_input_non_square_par() {
+  echo ""${Y4M_NOSQ_PAR_INPUT}""
+}
+
+y4m_input_720p() {
+  echo ""${Y4M_720P_INPUT}""
+}
+
 # Echo default vpxenc real time encoding params. $1 is the codec, which defaults
 # to vp8 if unspecified.
 vpxenc_rt_params() {
@@ -57,7 +72,7 @@ vpxenc_rt_params() {
    --buf-initial-sz=500
    --buf-optimal-sz=600
    --buf-sz=1000
-    --cpu-used=-5
+    --cpu-used=-6
    --end-usage=cbr
    --error-resilient=1
    --kf-max-dist=90000
@@ -247,6 +262,63 @@ vpxenc_vp9_webm_rt() {
  fi
 }

+vpxenc_vp9_webm_rt_multithread_tiled() {
+  if [ "$(vpxenc_can_encode_vp9)" = "yes" ] && \
+     [ "$(webm_io_available)" = "yes" ]; then
+    local readonly output="${VPX_TEST_OUTPUT_DIR}/vp9_rt_multithread_tiled.webm"
+    local readonly tilethread_min=2
+    local readonly tilethread_max=4
+    local readonly num_threads="$(seq ${tilethread_min} ${tilethread_max})"
+    local readonly num_tile_cols="$(seq ${tilethread_min} ${tilethread_max})"
+
+    for threads in ${num_threads}; do
+      for tile_cols in ${num_tile_cols}; do
+        vpxenc $(y4m_input_720p) \
+          $(vpxenc_rt_params vp9) \
+          --threads=${threads} \
+          --tile-columns=${tile_cols} \
+          --output="${output}"
+      done
+    done
+
+    if [ ! -e "${output}" ]; then
+      elog "Output file does not exist."
+      return 1
+    fi
+
+    rm "${output}"
+  fi
+}
+
+vpxenc_vp9_webm_rt_multithread_tiled_frameparallel() {
+  if [ "$(vpxenc_can_encode_vp9)" = "yes" ] && \
+     [ "$(webm_io_available)" = "yes" ]; then
+    local readonly output="${VPX_TEST_OUTPUT_DIR}/vp9_rt_mt_t_fp.webm"
+    local readonly tilethread_min=2
+    local readonly tilethread_max=4
+    local readonly num_threads="$(seq ${tilethread_min} ${tilethread_max})"
+    local readonly num_tile_cols="$(seq ${tilethread_min} ${tilethread_max})"
+
+    for threads in ${num_threads}; do
+      for tile_cols in ${num_tile_cols}; do
+        vpxenc $(y4m_input_720p) \
+          $(vpxenc_rt_params vp9) \
+          --threads=${threads} \
+          --tile-columns=${tile_cols} \
+          --frame-parallel=1 \
+          --output="${output}"
+      done
+    done
+
+    if [ ! -e "${output}" ]; then
+      elog "Output file does not exist."
+      return 1
+    fi
+
+    rm "${output}"
+  fi
+}
+
 vpxenc_vp9_webm_2pass() {
  if [ "$(vpxenc_can_encode_vp9)" = "yes" ] && \
     [ "$(webm_io_available)" = "yes" ]; then
@@ -320,6 +392,23 @@ vpxenc_vp9_webm_lag10_frames20() {
  fi
 }

+# TODO(fgalligan): Test that DisplayWidth is different than video width.
+vpxenc_vp9_webm_non_square_par() {
+  if [ "$(vpxenc_can_encode_vp9)" = "yes" ] && \
+     [ "$(webm_io_available)" = "yes" ]; then
+    local readonly output="${VPX_TEST_OUTPUT_DIR}/vp9_non_square_par.webm"
+    vpxenc $(y4m_input_non_square_par) \
+      --codec=vp9 \
+      --limit="${TEST_FRAMES}" \
+      --output="${output}"
+
+    if [ ! -e "${output}" ]; then
+      elog "Output file does not exist."
+      return 1
+    fi
+  fi
+}
+
 vpxenc_tests="vpxenc_vp8_ivf
              vpxenc_vp8_webm
              vpxenc_vp8_webm_rt
@@ -329,9 +418,12 @@ vpxenc_tests="vpxenc_vp8_ivf
              vpxenc_vp9_ivf
              vpxenc_vp9_webm
              vpxenc_vp9_webm_rt
+              vpxenc_vp9_webm_rt_multithread_tiled
+              vpxenc_vp9_webm_rt_multithread_tiled_frameparallel
              vpxenc_vp9_webm_2pass
              vpxenc_vp9_ivf_lossless
              vpxenc_vp9_ivf_minq0_maxq0
-              vpxenc_vp9_webm_lag10_frames20"
+              vpxenc_vp9_webm_lag10_frames20
+              vpxenc_vp9_webm_non_square_par"

 run_tests vpxenc_verify_environment "${vpxenc_tests}"
--- a/Show More
+++ b/Show More