Update aom to commit id f5bdeac22930ff4c6b219be49c843db35970b918

80 files changed, 2632 insertions, 1829 deletions

diff --git a/third_party/aom/test/active_map_refresh_test.cc b/third_party/aom/test/active_map_refresh_test.cc
index 7ee86e7e6..184692ca8 100644
--- a/third_party/aom/test/active_map_refresh_test.cc
+++ b/third_party/aom/test/active_map_refresh_test.cc
@@ -59,8 +59,8 @@ void GenerateMap(int mb_rows, int mb_cols, const aom_image_t &current,
 const int kAqModeCyclicRefresh = 3;
 
 class ActiveMapRefreshTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   ActiveMapRefreshTest() : EncoderTest(GET_PARAM(0)) {}
   virtual ~ActiveMapRefreshTest() {}
@@ -108,7 +108,7 @@ TEST_P(ActiveMapRefreshTest, Test) {
   cfg_.g_lag_in_frames = 0;
   cfg_.g_profile = 1;
   cfg_.rc_target_bitrate = 600;
-  cfg_.rc_resize_allowed = 0;
+  cfg_.rc_resize_mode = 0;
   cfg_.rc_min_quantizer = 8;
   cfg_.rc_max_quantizer = 30;
   cfg_.g_pass = AOM_RC_ONE_PASS;
diff --git a/third_party/aom/test/active_map_test.cc b/third_party/aom/test/active_map_test.cc
index a926b0faf..318a8518b 100644
--- a/third_party/aom/test/active_map_test.cc
+++ b/third_party/aom/test/active_map_test.cc
@@ -20,8 +20,8 @@
 namespace {
 
 class ActiveMapTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   static const int kWidth = 208;
   static const int kHeight = 144;
@@ -73,7 +73,7 @@ class ActiveMapTest
     // Validate that this non multiple of 64 wide clip encodes
     cfg_.g_lag_in_frames = 0;
     cfg_.rc_target_bitrate = 400;
-    cfg_.rc_resize_allowed = 0;
+    cfg_.rc_resize_mode = 0;
     cfg_.g_pass = AOM_RC_ONE_PASS;
     cfg_.rc_end_usage = AOM_CBR;
     cfg_.kf_max_dist = 90000;
diff --git a/third_party/aom/test/altref_test.cc b/third_party/aom/test/altref_test.cc
index 28b400121..dabb1475a 100644
--- a/third_party/aom/test/altref_test.cc
+++ b/third_party/aom/test/altref_test.cc
@@ -17,8 +17,8 @@
 namespace {
 
 class AltRefForcedKeyTestLarge
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   AltRefForcedKeyTestLarge()
       : EncoderTest(GET_PARAM(0)), encoding_mode_(GET_PARAM(1)),
diff --git a/third_party/aom/test/ans_codec_test.cc b/third_party/aom/test/ans_codec_test.cc
index a1b25fbda..59d352b2d 100644
--- a/third_party/aom/test/ans_codec_test.cc
+++ b/third_party/aom/test/ans_codec_test.cc
@@ -28,8 +28,8 @@ namespace {
 const char kTestVideoName[] = "niklas_1280_720_30.y4m";
 const int kTestVideoFrames = 10;
 
-class AnsCodecTest : public ::libaom_test::EncoderTest,
-                     public ::libaom_test::CodecTestWithParam<int> {
+class AnsCodecTest : public ::libaom_test::CodecTestWithParam<int>,
+                     public ::libaom_test::EncoderTest {
  protected:
   AnsCodecTest()
       : EncoderTest(GET_PARAM(0)), ans_window_size_log2_(GET_PARAM(1)) {}
diff --git a/third_party/aom/test/aq_segment_test.cc b/third_party/aom/test/aq_segment_test.cc
index b01f140a0..026b0022b 100644
--- a/third_party/aom/test/aq_segment_test.cc
+++ b/third_party/aom/test/aq_segment_test.cc
@@ -19,8 +19,8 @@
 namespace {
 
 class AqSegmentTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   AqSegmentTest() : EncoderTest(GET_PARAM(0)) {}
   virtual ~AqSegmentTest() {}
diff --git a/third_party/aom/test/arf_freq_test.cc b/third_party/aom/test/arf_freq_test.cc
index bef58b3e8..b4b17c9ab 100644
--- a/third_party/aom/test/arf_freq_test.cc
+++ b/third_party/aom/test/arf_freq_test.cc
@@ -76,9 +76,9 @@ int is_extension_y4m(const char *filename) {
 }
 
 class ArfFreqTestLarge
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith3Params<TestVideoParam,
-                                                 TestEncodeParam, int> {
+    : public ::libaom_test::CodecTestWith3Params<TestVideoParam,
+                                                 TestEncodeParam, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   ArfFreqTestLarge()
       : EncoderTest(GET_PARAM(0)), test_video_param_(GET_PARAM(1)),
@@ -99,7 +99,6 @@ class ArfFreqTestLarge
       cfg_.rc_buf_initial_sz = 500;
       cfg_.rc_buf_optimal_sz = 600;
     }
-    dec_cfg_.threads = 4;
   }
 
   virtual void BeginPassHook(unsigned int) {
diff --git a/third_party/aom/test/av1_convolve_2d_test.cc b/third_party/aom/test/av1_convolve_2d_test.cc
new file mode 100644
index 000000000..b066dd4f8
--- /dev/null
+++ b/third_party/aom/test/av1_convolve_2d_test.cc
@@ -0,0 +1,40 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include "third_party/googletest/src/googletest/include/gtest/gtest.h"
+#include "test/av1_convolve_2d_test_util.h"
+
+using std::tr1::tuple;
+using std::tr1::make_tuple;
+using libaom_test::ACMRandom;
+using libaom_test::AV1Convolve2D::AV1Convolve2DTest;
+#if CONFIG_HIGHBITDEPTH
+using libaom_test::AV1HighbdConvolve2D::AV1HighbdConvolve2DTest;
+#endif
+
+namespace {
+
+TEST_P(AV1Convolve2DTest, CheckOutput) { RunCheckOutput(GET_PARAM(3)); }
+
+INSTANTIATE_TEST_CASE_P(
+    SSE2, AV1Convolve2DTest,
+    libaom_test::AV1Convolve2D::BuildParams(av1_convolve_2d_sse2));
+
+#if CONFIG_HIGHBITDEPTH && HAVE_SSSE3
+TEST_P(AV1HighbdConvolve2DTest, CheckOutput) { RunCheckOutput(GET_PARAM(4)); }
+
+INSTANTIATE_TEST_CASE_P(SSSE3, AV1HighbdConvolve2DTest,
+                        libaom_test::AV1HighbdConvolve2D::BuildParams(
+                            av1_highbd_convolve_2d_ssse3));
+
+#endif
+
+}  // namespace
diff --git a/third_party/aom/test/av1_convolve_2d_test_util.cc b/third_party/aom/test/av1_convolve_2d_test_util.cc
new file mode 100644
index 000000000..8cec216af
--- /dev/null
+++ b/third_party/aom/test/av1_convolve_2d_test_util.cc
@@ -0,0 +1,188 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include "test/av1_convolve_2d_test_util.h"
+
+#include "av1/common/convolve.h"
+
+using std::tr1::tuple;
+using std::tr1::make_tuple;
+
+namespace libaom_test {
+
+namespace AV1Convolve2D {
+
+::testing::internal::ParamGenerator<Convolve2DParam> BuildParams(
+    convolve_2d_func filter) {
+  const Convolve2DParam params[] = {
+    make_tuple(4, 4, 20, filter),  make_tuple(8, 8, 10, filter),
+    make_tuple(64, 64, 1, filter), make_tuple(4, 16, 10, filter),
+    make_tuple(32, 8, 5, filter),
+  };
+  return ::testing::ValuesIn(params);
+}
+
+AV1Convolve2DTest::~AV1Convolve2DTest() {}
+void AV1Convolve2DTest::SetUp() { rnd_.Reset(ACMRandom::DeterministicSeed()); }
+
+void AV1Convolve2DTest::TearDown() { libaom_test::ClearSystemState(); }
+
+void AV1Convolve2DTest::RunCheckOutput(convolve_2d_func test_impl) {
+  const int w = 128, h = 128;
+  const int out_w = GET_PARAM(0), out_h = GET_PARAM(1);
+  const int num_iters = GET_PARAM(2);
+  int i, j, k;
+
+  uint8_t *input = new uint8_t[h * w];
+
+  int output_n = out_h * MAX_SB_SIZE;
+  CONV_BUF_TYPE *output = new CONV_BUF_TYPE[output_n];
+  CONV_BUF_TYPE *output2 = new CONV_BUF_TYPE[output_n];
+
+  for (i = 0; i < h; ++i)
+    for (j = 0; j < w; ++j) input[i * w + j] = rnd_.Rand8();
+
+  memset(output, 0, output_n * sizeof(CONV_BUF_TYPE));
+  memset(output2, 0, output_n * sizeof(CONV_BUF_TYPE));
+
+  int hfilter, vfilter, subx, suby;
+  for (hfilter = EIGHTTAP_REGULAR; hfilter < INTERP_FILTERS_ALL; ++hfilter) {
+    for (vfilter = EIGHTTAP_REGULAR; vfilter < INTERP_FILTERS_ALL; ++vfilter) {
+      InterpFilterParams filter_params_x =
+          av1_get_interp_filter_params((InterpFilter)hfilter);
+      InterpFilterParams filter_params_y =
+          av1_get_interp_filter_params((InterpFilter)vfilter);
+      ConvolveParams conv_params1 =
+          get_conv_params_no_round(0, 0, 0, output, MAX_SB_SIZE);
+      ConvolveParams conv_params2 =
+          get_conv_params_no_round(0, 0, 0, output2, MAX_SB_SIZE);
+
+      for (subx = 0; subx < 16; ++subx)
+        for (suby = 0; suby < 16; ++suby) {
+          for (i = 0; i < num_iters; ++i) {
+            // Choose random locations within the source block
+            int offset_r = 3 + rnd_.PseudoUniform(h - out_h - 7);
+            int offset_c = 3 + rnd_.PseudoUniform(w - out_w - 7);
+            av1_convolve_2d_c(input + offset_r * w + offset_c, w, output,
+                              MAX_SB_SIZE, out_w, out_h, &filter_params_x,
+                              &filter_params_y, subx, suby, &conv_params1);
+            test_impl(input + offset_r * w + offset_c, w, output2, MAX_SB_SIZE,
+                      out_w, out_h, &filter_params_x, &filter_params_y, subx,
+                      suby, &conv_params2);
+
+            for (j = 0; j < out_h; ++j)
+              for (k = 0; k < out_w; ++k) {
+                int idx = j * MAX_SB_SIZE + k;
+                ASSERT_EQ(output[idx], output2[idx])
+                    << "Pixel mismatch at index " << idx << " = (" << j << ", "
+                    << k << "), sub pixel offset = (" << suby << ", " << subx
+                    << ")";
+              }
+          }
+        }
+    }
+  }
+  delete[] input;
+  delete[] output;
+  delete[] output2;
+}
+}  // namespace AV1Convolve2D
+
+#if CONFIG_HIGHBITDEPTH
+namespace AV1HighbdConvolve2D {
+
+::testing::internal::ParamGenerator<HighbdConvolve2DParam> BuildParams(
+    highbd_convolve_2d_func filter) {
+  const HighbdConvolve2DParam params[] = {
+    make_tuple(4, 4, 20, 8, filter),   make_tuple(8, 8, 10, 8, filter),
+    make_tuple(64, 64, 1, 8, filter),  make_tuple(4, 16, 10, 8, filter),
+    make_tuple(32, 8, 10, 8, filter),  make_tuple(4, 4, 20, 10, filter),
+    make_tuple(8, 8, 10, 10, filter),  make_tuple(64, 64, 1, 10, filter),
+    make_tuple(4, 16, 10, 10, filter), make_tuple(32, 8, 10, 10, filter),
+    make_tuple(4, 4, 20, 12, filter),  make_tuple(8, 8, 10, 12, filter),
+    make_tuple(64, 64, 1, 12, filter), make_tuple(4, 16, 10, 12, filter),
+    make_tuple(32, 8, 10, 12, filter),
+  };
+  return ::testing::ValuesIn(params);
+}
+
+AV1HighbdConvolve2DTest::~AV1HighbdConvolve2DTest() {}
+void AV1HighbdConvolve2DTest::SetUp() {
+  rnd_.Reset(ACMRandom::DeterministicSeed());
+}
+
+void AV1HighbdConvolve2DTest::TearDown() { libaom_test::ClearSystemState(); }
+
+void AV1HighbdConvolve2DTest::RunCheckOutput(
+    highbd_convolve_2d_func test_impl) {
+  const int w = 128, h = 128;
+  const int out_w = GET_PARAM(0), out_h = GET_PARAM(1);
+  const int num_iters = GET_PARAM(2);
+  const int bd = GET_PARAM(3);
+  int i, j, k;
+
+  uint16_t *input = new uint16_t[h * w];
+
+  int output_n = out_h * MAX_SB_SIZE;
+  CONV_BUF_TYPE *output = new CONV_BUF_TYPE[output_n];
+  CONV_BUF_TYPE *output2 = new CONV_BUF_TYPE[output_n];
+
+  for (i = 0; i < h; ++i)
+    for (j = 0; j < w; ++j) input[i * w + j] = rnd_.Rand16() & ((1 << bd) - 1);
+
+  memset(output, 0, output_n * sizeof(CONV_BUF_TYPE));
+  memset(output2, 0, output_n * sizeof(CONV_BUF_TYPE));
+
+  int hfilter, vfilter, subx, suby;
+  for (hfilter = EIGHTTAP_REGULAR; hfilter < INTERP_FILTERS_ALL; ++hfilter) {
+    for (vfilter = EIGHTTAP_REGULAR; vfilter < INTERP_FILTERS_ALL; ++vfilter) {
+      InterpFilterParams filter_params_x =
+          av1_get_interp_filter_params((InterpFilter)hfilter);
+      InterpFilterParams filter_params_y =
+          av1_get_interp_filter_params((InterpFilter)vfilter);
+      ConvolveParams conv_params1 =
+          get_conv_params_no_round(0, 0, 0, output, MAX_SB_SIZE);
+      ConvolveParams conv_params2 =
+          get_conv_params_no_round(0, 0, 0, output2, MAX_SB_SIZE);
+
+      for (subx = 0; subx < 16; ++subx)
+        for (suby = 0; suby < 16; ++suby) {
+          for (i = 0; i < num_iters; ++i) {
+            // Choose random locations within the source block
+            int offset_r = 3 + rnd_.PseudoUniform(h - out_h - 7);
+            int offset_c = 3 + rnd_.PseudoUniform(w - out_w - 7);
+            av1_highbd_convolve_2d_c(input + offset_r * w + offset_c, w, output,
+                                     MAX_SB_SIZE, out_w, out_h,
+                                     &filter_params_x, &filter_params_y, subx,
+                                     suby, &conv_params1, bd);
+            test_impl(input + offset_r * w + offset_c, w, output2, MAX_SB_SIZE,
+                      out_w, out_h, &filter_params_x, &filter_params_y, subx,
+                      suby, &conv_params2, bd);
+
+            for (j = 0; j < out_h; ++j)
+              for (k = 0; k < out_w; ++k) {
+                int idx = j * MAX_SB_SIZE + k;
+                ASSERT_EQ(output[idx], output2[idx])
+                    << "Pixel mismatch at index " << idx << " = (" << j << ", "
+                    << k << "), sub pixel offset = (" << suby << ", " << subx
+                    << ")";
+              }
+          }
+        }
+    }
+  }
+  delete[] input;
+  delete[] output;
+  delete[] output2;
+}
+}  // namespace AV1HighbdConvolve2D
+#endif  // CONFIG_HIGHBITDEPTH
+}  // namespace libaom_test
diff --git a/third_party/aom/test/av1_convolve_2d_test_util.h b/third_party/aom/test/av1_convolve_2d_test_util.h
new file mode 100644
index 000000000..ed0eeb450
--- /dev/null
+++ b/third_party/aom/test/av1_convolve_2d_test_util.h
@@ -0,0 +1,87 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#ifndef TEST_HIPREC_CONVOLVE_TEST_UTIL_H_
+#define TEST_HIPREC_CONVOLVE_TEST_UTIL_H_
+
+#include "third_party/googletest/src/googletest/include/gtest/gtest.h"
+#include "test/acm_random.h"
+#include "test/util.h"
+#include "./av1_rtcd.h"
+#include "./aom_dsp_rtcd.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
+
+namespace libaom_test {
+
+namespace AV1Convolve2D {
+
+typedef void (*convolve_2d_func)(const uint8_t *src, int src_stride,
+                                 CONV_BUF_TYPE *dst, int dst_stride, int w,
+                                 int h, InterpFilterParams *filter_params_x,
+                                 InterpFilterParams *filter_params_y,
+                                 const int subpel_x_q4, const int subpel_y_q4,
+                                 ConvolveParams *conv_params);
+
+typedef std::tr1::tuple<int, int, int, convolve_2d_func> Convolve2DParam;
+
+::testing::internal::ParamGenerator<Convolve2DParam> BuildParams(
+    convolve_2d_func filter);
+
+class AV1Convolve2DTest : public ::testing::TestWithParam<Convolve2DParam> {
+ public:
+  virtual ~AV1Convolve2DTest();
+  virtual void SetUp();
+
+  virtual void TearDown();
+
+ protected:
+  void RunCheckOutput(convolve_2d_func test_impl);
+
+  libaom_test::ACMRandom rnd_;
+};
+
+}  // namespace AV1Convolve2D
+
+#if CONFIG_HIGHBITDEPTH
+namespace AV1HighbdConvolve2D {
+typedef void (*highbd_convolve_2d_func)(
+    const uint16_t *src, int src_stride, CONV_BUF_TYPE *dst, int dst_stride,
+    int w, int h, InterpFilterParams *filter_params_x,
+    InterpFilterParams *filter_params_y, const int subpel_x_q4,
+    const int subpel_y_q4, ConvolveParams *conv_params, int bd);
+
+typedef std::tr1::tuple<int, int, int, int, highbd_convolve_2d_func>
+    HighbdConvolve2DParam;
+
+::testing::internal::ParamGenerator<HighbdConvolve2DParam> BuildParams(
+    highbd_convolve_2d_func filter);
+
+class AV1HighbdConvolve2DTest
+    : public ::testing::TestWithParam<HighbdConvolve2DParam> {
+ public:
+  virtual ~AV1HighbdConvolve2DTest();
+  virtual void SetUp();
+
+  virtual void TearDown();
+
+ protected:
+  void RunCheckOutput(highbd_convolve_2d_func test_impl);
+
+  libaom_test::ACMRandom rnd_;
+};
+
+}  // namespace AV1HighbdConvolve2D
+#endif  // CONFIG_HIGHBITDEPTH
+
+}  // namespace libaom_test
+
+#endif  // TEST_HIPREC_CONVOLVE_TEST_UTIL_H_
diff --git a/third_party/aom/test/av1_convolve_optimz_test.cc b/third_party/aom/test/av1_convolve_optimz_test.cc
index c32f4cb95..95bf63f44 100644
--- a/third_party/aom/test/av1_convolve_optimz_test.cc
+++ b/third_party/aom/test/av1_convolve_optimz_test.cc
@@ -74,7 +74,7 @@ class AV1ConvolveOptimzTest : public ::testing::TestWithParam<ConvParams> {
     subpel_ = GET_PARAM(5);
     int ref = GET_PARAM(6);
     const int plane = 0;
-    conv_params_ = get_conv_params(ref, plane);
+    conv_params_ = get_conv_params(ref, ref, plane);
 
     alloc_ = new uint8_t[maxBlockSize * 4];
     src_ = alloc_ + (vertiOffset * maxWidth);
diff --git a/third_party/aom/test/av1_convolve_test.cc b/third_party/aom/test/av1_convolve_test.cc
index 9ea662381..3947c7166 100644
--- a/third_party/aom/test/av1_convolve_test.cc
+++ b/third_party/aom/test/av1_convolve_test.cc
@@ -145,11 +145,11 @@ class Av1ConvolveTest : public ::testing::TestWithParam<ConvolveParam> {
 };
 
 int bsize_ls[] = { 1, 2, 4, 8, 16, 32, 64, 3, 7, 15, 31, 63 };
-int bsize_num = sizeof(bsize_ls) / sizeof(bsize_ls[0]);
+int bsize_num = NELEMENTS(bsize_ls);
 
 TEST_P(Av1ConvolveTest, av1_convolve_vert) {
   const int y_step_q4 = 16;
-  ConvolveParams conv_params = get_conv_params(0, 0);
+  ConvolveParams conv_params = get_conv_params(0, 0, 0);
 
   int in_stride, out_stride, ref_out_stride, avg_out_stride, ref_avg_out_stride;
   uint8_t *in = add_input(MAX_SB_SIZE, MAX_SB_SIZE, &in_stride);
@@ -172,6 +172,7 @@ TEST_P(Av1ConvolveTest, av1_convolve_vert) {
                               ref_out, ref_out_stride, w, h);
 
         conv_params.ref = 0;
+        conv_params.do_average = 0;
         cfs_->vf_(in, in_stride, out, out_stride, w, h, param_vert, subpel_y_q4,
                   y_step_q4, &conv_params);
         EXPECT_EQ(match(out, out_stride, ref_out, ref_out_stride, w, h), 1)
@@ -186,6 +187,7 @@ TEST_P(Av1ConvolveTest, av1_convolve_vert) {
           }
         }
         conv_params.ref = 1;
+        conv_params.do_average = 1;
         cfs_->vf_(in, in_stride, avg_out, avg_out_stride, w, h, param_vert,
                   subpel_y_q4, y_step_q4, &conv_params);
         EXPECT_EQ(match(avg_out, avg_out_stride, ref_avg_out,
@@ -200,7 +202,7 @@ TEST_P(Av1ConvolveTest, av1_convolve_vert) {
 
 TEST_P(Av1ConvolveTest, av1_convolve_horiz) {
   const int x_step_q4 = 16;
-  ConvolveParams conv_params = get_conv_params(0, 0);
+  ConvolveParams conv_params = get_conv_params(0, 0, 0);
 
   int in_stride, out_stride, ref_out_stride, avg_out_stride, ref_avg_out_stride;
   uint8_t *in = add_input(MAX_SB_SIZE, MAX_SB_SIZE, &in_stride);
@@ -223,6 +225,7 @@ TEST_P(Av1ConvolveTest, av1_convolve_horiz) {
                                ref_out, ref_out_stride, w, h);
 
         conv_params.ref = 0;
+        conv_params.do_average = 0;
         cfs_->hf_(in, in_stride, out, out_stride, w, h, param_horiz,
                   subpel_x_q4, x_step_q4, &conv_params);
         EXPECT_EQ(match(out, out_stride, ref_out, ref_out_stride, w, h), 1)
@@ -237,6 +240,7 @@ TEST_P(Av1ConvolveTest, av1_convolve_horiz) {
           }
         }
         conv_params.ref = 1;
+        conv_params.do_average = 1;
         cfs_->hf_(in, in_stride, avg_out, avg_out_stride, w, h, param_horiz,
                   subpel_x_q4, x_step_q4, &conv_params);
         EXPECT_EQ(match(avg_out, avg_out_stride, ref_avg_out,
diff --git a/third_party/aom/test/av1_dct_test.cc b/third_party/aom/test/av1_dct_test.cc
index 691cc8b79..8ce7a79d4 100644
--- a/third_party/aom/test/av1_dct_test.cc
+++ b/third_party/aom/test/av1_dct_test.cc
@@ -23,6 +23,9 @@
 #define CONFIG_COEFFICIENT_RANGE_CHECKING 1
 #define AV1_DCT_GTEST
 #include "av1/encoder/dct.c"
+#if CONFIG_DAALA_DCT4 || CONFIG_DAALA_DCT8
+#include "av1/common/daala_tx.c"
+#endif
 
 using libaom_test::ACMRandom;
 
diff --git a/third_party/aom/test/av1_ext_tile_test.cc b/third_party/aom/test/av1_ext_tile_test.cc
index f96447965..034b07167 100644
--- a/third_party/aom/test/av1_ext_tile_test.cc
+++ b/third_party/aom/test/av1_ext_tile_test.cc
@@ -30,11 +30,11 @@ const int kTIleSizeInPixels = (kTileSize << 6);
 const int kImgWidth = 704;
 const int kImgHeight = 576;
 
-// This test tests "tile_encoding_mode = TILE_VR" case. The TILE_NORMAL case is
-// tested by the tile_independence test.
+// This test tests large scale tile coding case. Non-large-scale tile coding
+// is tested by the tile_independence test.
 class AV1ExtTileTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   AV1ExtTileTest()
       : EncoderTest(GET_PARAM(0)), encoding_mode_(GET_PARAM(1)),
@@ -43,6 +43,7 @@ class AV1ExtTileTest
     aom_codec_dec_cfg_t cfg = aom_codec_dec_cfg_t();
     cfg.w = kImgWidth;
     cfg.h = kImgHeight;
+    cfg.allow_lowbitdepth = 1;
 
     decoder_ = codec_->CreateDecoder(cfg, 0);
     decoder_->Control(AV1_SET_DECODE_TILE_ROW, -1);
@@ -83,11 +84,15 @@ class AV1ExtTileTest
       // The tile size is 64x64.
       encoder->Control(AV1E_SET_TILE_COLUMNS, kTileSize);
       encoder->Control(AV1E_SET_TILE_ROWS, kTileSize);
-      encoder->Control(AV1E_SET_TILE_ENCODING_MODE, 1);  // TILE_VR
+      // TODO(yunqingwang): test single_tile_decoding = 0.
+      encoder->Control(AV1E_SET_SINGLE_TILE_DECODING, 1);
 #if CONFIG_EXT_PARTITION
       // Always use 64x64 max partition.
       encoder->Control(AV1E_SET_SUPERBLOCK_SIZE, AOM_SUPERBLOCK_SIZE_64X64);
 #endif
+#if CONFIG_LOOPFILTERING_ACROSS_TILES
+      encoder->Control(AV1E_SET_TILE_LOOPFILTER, 0);
+#endif
     }
 
     if (video->frame() == 1) {
@@ -182,6 +187,7 @@ TEST_P(AV1ExtTileTest, DecoderResultTest) {
                                        kImgHeight, 30, 1, 0, kLimit);
   cfg_.rc_target_bitrate = 500;
   cfg_.g_error_resilient = AOM_ERROR_RESILIENT_DEFAULT;
+  cfg_.large_scale_tile = 1;
   cfg_.g_lag_in_frames = 0;
   cfg_.g_threads = 1;
 
diff --git a/third_party/aom/test/av1_fht16x16_test.cc b/third_party/aom/test/av1_fht16x16_test.cc
index e1032ef24..c0f6974c6 100644
--- a/third_party/aom/test/av1_fht16x16_test.cc
+++ b/third_party/aom/test/av1_fht16x16_test.cc
@@ -25,18 +25,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht16x16Param;
 
-void fht16x16_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht16x16_c(in, out, stride, tx_type);
+void fht16x16_ref(const int16_t *in, tran_low_t *out, int stride,
+                  TxfmParam *txfm_param) {
+  av1_fht16x16_c(in, out, stride, txfm_param);
 }
 
 void iht16x16_ref(const tran_low_t *in, uint8_t *dest, int stride,
-                  int tx_type) {
-  av1_iht16x16_256_add_c(in, dest, stride, tx_type);
+                  const TxfmParam *txfm_param) {
+  av1_iht16x16_256_add_c(in, dest, stride, txfm_param);
 }
 
 #if CONFIG_HIGHBITDEPTH
@@ -62,7 +63,6 @@ class AV1Trans16x16HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 16;
     height_ = 16;
     fwd_txfm_ref = fht16x16_ref;
@@ -70,16 +70,17 @@ class AV1Trans16x16HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht16x32_test.cc b/third_party/aom/test/av1_fht16x32_test.cc
index 43d025327..099a312e3 100644
--- a/third_party/aom/test/av1_fht16x32_test.cc
+++ b/third_party/aom/test/av1_fht16x32_test.cc
@@ -25,17 +25,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht16x32Param;
 
-void fht16x32_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht16x32_c(in, out, stride, tx_type);
+void fht16x32_ref(const int16_t *in, tran_low_t *out, int stride,
+                  TxfmParam *txfm_param) {
+  av1_fht16x32_c(in, out, stride, txfm_param);
 }
 
-void iht16x32_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht16x32_512_add_c(in, out, stride, tx_type);
+void iht16x32_ref(const tran_low_t *in, uint8_t *out, int stride,
+                  const TxfmParam *txfm_param) {
+  av1_iht16x32_512_add_c(in, out, stride, txfm_param);
 }
 
 class AV1Trans16x32HT : public libaom_test::TransformTestBase,
@@ -46,7 +48,6 @@ class AV1Trans16x32HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 16;
     height_ = 32;
     fwd_txfm_ref = fht16x32_ref;
@@ -54,16 +55,17 @@ class AV1Trans16x32HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht16x8_test.cc b/third_party/aom/test/av1_fht16x8_test.cc
index d99bec5eb..8277e2865 100644
--- a/third_party/aom/test/av1_fht16x8_test.cc
+++ b/third_party/aom/test/av1_fht16x8_test.cc
@@ -25,17 +25,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht16x8Param;
 
-void fht16x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht16x8_c(in, out, stride, tx_type);
+void fht16x8_ref(const int16_t *in, tran_low_t *out, int stride,
+                 TxfmParam *txfm_param) {
+  av1_fht16x8_c(in, out, stride, txfm_param);
 }
 
-void iht16x8_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht16x8_128_add_c(in, out, stride, tx_type);
+void iht16x8_ref(const tran_low_t *in, uint8_t *out, int stride,
+                 const TxfmParam *txfm_param) {
+  av1_iht16x8_128_add_c(in, out, stride, txfm_param);
 }
 
 class AV1Trans16x8HT : public libaom_test::TransformTestBase,
@@ -46,7 +48,6 @@ class AV1Trans16x8HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 16;
     height_ = 8;
     inv_txfm_ref = iht16x8_ref;
@@ -54,16 +55,17 @@ class AV1Trans16x8HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht32x16_test.cc b/third_party/aom/test/av1_fht32x16_test.cc
index e38283f86..1c70fd4fc 100644
--- a/third_party/aom/test/av1_fht32x16_test.cc
+++ b/third_party/aom/test/av1_fht32x16_test.cc
@@ -25,17 +25,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht32x16Param;
 
-void fht32x16_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht32x16_c(in, out, stride, tx_type);
+void fht32x16_ref(const int16_t *in, tran_low_t *out, int stride,
+                  TxfmParam *txfm_param) {
+  av1_fht32x16_c(in, out, stride, txfm_param);
 }
 
-void iht32x16_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht32x16_512_add_c(in, out, stride, tx_type);
+void iht32x16_ref(const tran_low_t *in, uint8_t *out, int stride,
+                  const TxfmParam *txfm_param) {
+  av1_iht32x16_512_add_c(in, out, stride, txfm_param);
 }
 
 class AV1Trans32x16HT : public libaom_test::TransformTestBase,
@@ -46,7 +48,6 @@ class AV1Trans32x16HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 32;
     height_ = 16;
     fwd_txfm_ref = fht32x16_ref;
@@ -54,16 +55,17 @@ class AV1Trans32x16HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht32x32_test.cc b/third_party/aom/test/av1_fht32x32_test.cc
index cdd915337..e96ffffc2 100644
--- a/third_party/aom/test/av1_fht32x32_test.cc
+++ b/third_party/aom/test/av1_fht32x32_test.cc
@@ -25,13 +25,14 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht32x32Param;
 
-void fht32x32_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht32x32_c(in, out, stride, tx_type);
+void fht32x32_ref(const int16_t *in, tran_low_t *out, int stride,
+                  TxfmParam *txfm_param) {
+  av1_fht32x32_c(in, out, stride, txfm_param);
 }
 
 #if CONFIG_HIGHBITDEPTH
@@ -51,11 +52,11 @@ void highbd_fht32x32_ref(const int16_t *in, int32_t *out, int stride,
 
 #if HAVE_SSE2 || HAVE_AVX2
 void dummy_inv_txfm(const tran_low_t *in, uint8_t *out, int stride,
-                    int tx_type) {
+                    const TxfmParam *txfm_param) {
   (void)in;
   (void)out;
   (void)stride;
-  (void)tx_type;
+  (void)txfm_param;
 }
 #endif
 
@@ -67,23 +68,23 @@ class AV1Trans32x32HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 32;
     height_ = 32;
     fwd_txfm_ref = fht32x32_ref;
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht4x4_test.cc b/third_party/aom/test/av1_fht4x4_test.cc
index 42837d3a4..f49d7368d 100644
--- a/third_party/aom/test/av1_fht4x4_test.cc
+++ b/third_party/aom/test/av1_fht4x4_test.cc
@@ -25,17 +25,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht4x4Param;
 
-void fht4x4_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht4x4_c(in, out, stride, tx_type);
+void fht4x4_ref(const int16_t *in, tran_low_t *out, int stride,
+                TxfmParam *txfm_param) {
+  av1_fht4x4_c(in, out, stride, txfm_param);
 }
 
-void iht4x4_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht4x4_16_add_c(in, out, stride, tx_type);
+void iht4x4_ref(const tran_low_t *in, uint8_t *out, int stride,
+                const TxfmParam *txfm_param) {
+  av1_iht4x4_16_add_c(in, out, stride, txfm_param);
 }
 
 #if CONFIG_HIGHBITDEPTH
@@ -62,7 +64,6 @@ class AV1Trans4x4HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 4;
     height_ = 4;
     fwd_txfm_ref = fht4x4_ref;
@@ -70,16 +71,17 @@ class AV1Trans4x4HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht4x8_test.cc b/third_party/aom/test/av1_fht4x8_test.cc
index a899c8739..e447d8e2e 100644
--- a/third_party/aom/test/av1_fht4x8_test.cc
+++ b/third_party/aom/test/av1_fht4x8_test.cc
@@ -25,17 +25,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht4x8Param;
 
-void fht4x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht4x8_c(in, out, stride, tx_type);
+void fht4x8_ref(const int16_t *in, tran_low_t *out, int stride,
+                TxfmParam *txfm_param) {
+  av1_fht4x8_c(in, out, stride, txfm_param);
 }
 
-void iht4x8_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht4x8_32_add_c(in, out, stride, tx_type);
+void iht4x8_ref(const tran_low_t *in, uint8_t *out, int stride,
+                const TxfmParam *txfm_param) {
+  av1_iht4x8_32_add_c(in, out, stride, txfm_param);
 }
 
 class AV1Trans4x8HT : public libaom_test::TransformTestBase,
@@ -46,7 +48,6 @@ class AV1Trans4x8HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 4;
     height_ = 8;
     fwd_txfm_ref = fht4x8_ref;
@@ -54,16 +55,17 @@ class AV1Trans4x8HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht64x64_test.cc b/third_party/aom/test/av1_fht64x64_test.cc
index cde1d0ca3..61ea9f1f3 100644
--- a/third_party/aom/test/av1_fht64x64_test.cc
+++ b/third_party/aom/test/av1_fht64x64_test.cc
@@ -26,18 +26,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht64x64Param;
 
-void fht64x64_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht64x64_c(in, out, stride, tx_type);
+void fht64x64_ref(const int16_t *in, tran_low_t *out, int stride,
+                  TxfmParam *txfm_param) {
+  av1_fht64x64_c(in, out, stride, txfm_param);
 }
 
 void iht64x64_ref(const tran_low_t *in, uint8_t *dest, int stride,
-                  int tx_type) {
-  av1_iht64x64_4096_add_c(in, dest, stride, tx_type);
+                  const TxfmParam *txfm_param) {
+  av1_iht64x64_4096_add_c(in, dest, stride, txfm_param);
 }
 
 class AV1Trans64x64HT : public libaom_test::TransformTestBase,
@@ -48,7 +49,6 @@ class AV1Trans64x64HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 64;
     height_ = 64;
     fwd_txfm_ref = fht64x64_ref;
@@ -56,16 +56,17 @@ class AV1Trans64x64HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht8x16_test.cc b/third_party/aom/test/av1_fht8x16_test.cc
index ace9a8f47..11f085885 100644
--- a/third_party/aom/test/av1_fht8x16_test.cc
+++ b/third_party/aom/test/av1_fht8x16_test.cc
@@ -24,17 +24,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht8x16Param;
 
-void fht8x16_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht8x16_c(in, out, stride, tx_type);
+void fht8x16_ref(const int16_t *in, tran_low_t *out, int stride,
+                 TxfmParam *txfm_param) {
+  av1_fht8x16_c(in, out, stride, txfm_param);
 }
 
-void iht8x16_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht8x16_128_add_c(in, out, stride, tx_type);
+void iht8x16_ref(const tran_low_t *in, uint8_t *out, int stride,
+                 const TxfmParam *txfm_param) {
+  av1_iht8x16_128_add_c(in, out, stride, txfm_param);
 }
 
 class AV1Trans8x16HT : public libaom_test::TransformTestBase,
@@ -45,7 +47,6 @@ class AV1Trans8x16HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 8;
     height_ = 16;
     inv_txfm_ref = iht8x16_ref;
@@ -53,16 +54,17 @@ class AV1Trans8x16HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht8x4_test.cc b/third_party/aom/test/av1_fht8x4_test.cc
index 9bf4ff647..c797421af 100644
--- a/third_party/aom/test/av1_fht8x4_test.cc
+++ b/third_party/aom/test/av1_fht8x4_test.cc
@@ -24,17 +24,19 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using std::tr1::tuple;
 using libaom_test::FhtFunc;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht8x4Param;
 
-void fht8x4_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht8x4_c(in, out, stride, tx_type);
+void fht8x4_ref(const int16_t *in, tran_low_t *out, int stride,
+                TxfmParam *txfm_param) {
+  av1_fht8x4_c(in, out, stride, txfm_param);
 }
 
-void iht8x4_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht8x4_32_add_c(in, out, stride, tx_type);
+void iht8x4_ref(const tran_low_t *in, uint8_t *out, int stride,
+                const TxfmParam *txfm_param) {
+  av1_iht8x4_32_add_c(in, out, stride, txfm_param);
 }
 
 class AV1Trans8x4HT : public libaom_test::TransformTestBase,
@@ -45,7 +47,6 @@ class AV1Trans8x4HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 8;
     height_ = 4;
     fwd_txfm_ref = fht8x4_ref;
@@ -53,16 +54,17 @@ class AV1Trans8x4HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fht8x8_test.cc b/third_party/aom/test/av1_fht8x8_test.cc
index 99cff1014..259557cfb 100644
--- a/third_party/aom/test/av1_fht8x8_test.cc
+++ b/third_party/aom/test/av1_fht8x8_test.cc
@@ -25,18 +25,20 @@ using libaom_test::ACMRandom;
 
 namespace {
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 
 using libaom_test::FhtFunc;
 using std::tr1::tuple;
 typedef tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht8x8Param;
 
-void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht8x8_c(in, out, stride, tx_type);
+void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride,
+                TxfmParam *txfm_param) {
+  av1_fht8x8_c(in, out, stride, txfm_param);
 }
 
-void iht8x8_ref(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_iht8x8_64_add_c(in, out, stride, tx_type);
+void iht8x8_ref(const tran_low_t *in, uint8_t *out, int stride,
+                const TxfmParam *txfm_param) {
+  av1_iht8x8_64_add_c(in, out, stride, txfm_param);
 }
 
 #if CONFIG_HIGHBITDEPTH
@@ -61,7 +63,6 @@ class AV1Trans8x8HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 8;
     height_ = 8;
     fwd_txfm_ref = fht8x8_ref;
@@ -69,16 +70,17 @@ class AV1Trans8x8HT : public libaom_test::TransformTestBase,
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
diff --git a/third_party/aom/test/av1_fwd_txfm1d_test.cc b/third_party/aom/test/av1_fwd_txfm1d_test.cc
index 511e057fa..b10e84d2c 100644
--- a/third_party/aom/test/av1_fwd_txfm1d_test.cc
+++ b/third_party/aom/test/av1_fwd_txfm1d_test.cc
@@ -82,7 +82,7 @@ TEST(av1_fwd_txfm1d, clamp_block) {
   int col = 1;
   int block_size = 3;
   int stride = 5;
-  clamp_block(block[row] + col, block_size, stride, -4, 2);
+  clamp_block(block[row] + col, block_size, block_size, stride, -4, 2);
   for (int r = 0; r < stride; r++) {
     for (int c = 0; c < stride; c++) {
       EXPECT_EQ(block[r][c], ref_block[r][c]);
diff --git a/third_party/aom/test/av1_inv_txfm1d_test.cc b/third_party/aom/test/av1_inv_txfm1d_test.cc
index 9cf33a2fd..b871105eb 100644
--- a/third_party/aom/test/av1_inv_txfm1d_test.cc
+++ b/third_party/aom/test/av1_inv_txfm1d_test.cc
@@ -10,6 +10,7 @@
  */
 
 #include "test/av1_txfm_test.h"
+#include "test/util.h"
 #include "av1/common/av1_fwd_txfm1d.h"
 #include "av1/common/av1_inv_txfm1d.h"
 
@@ -41,14 +42,12 @@ const TxfmFunc inv_txfm_func_ls[][2] = {
 };
 
 // the maximum stage number of fwd/inv 1d dct/adst txfm is 12
-const int8_t cos_bit[12] = { 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14 };
+const int8_t cos_bit[12] = { 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13 };
 const int8_t range_bit[12] = { 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32 };
 
-#define ARRAY_SIZE(x) (int)(sizeof(x) / sizeof(x[0]))
-
 TEST(av1_inv_txfm1d, round_trip) {
   ACMRandom rnd(ACMRandom::DeterministicSeed());
-  for (int si = 0; si < ARRAY_SIZE(fwd_txfm_func_ls); ++si) {
+  for (int si = 0; si < NELEMENTS(fwd_txfm_func_ls); ++si) {
     int txfm_size = txfm_size_ls[si];
 
     for (int ti = 0; ti < txfm_type_num; ++ti) {
@@ -64,7 +63,7 @@ TEST(av1_inv_txfm1d, round_trip) {
         int32_t output[64];
         int32_t round_trip_output[64];
 
-        assert(txfm_size <= ARRAY_SIZE(input));
+        ASSERT_LE(txfm_size, NELEMENTS(input));
 
         for (int ni = 0; ni < txfm_size; ++ni) {
           input[ni] = rnd.Rand16() % input_base - rnd.Rand16() % input_base;
diff --git a/third_party/aom/test/av1_inv_txfm2d_test.cc b/third_party/aom/test/av1_inv_txfm2d_test.cc
index 89ae34b01..5185c1ca8 100644
--- a/third_party/aom/test/av1_inv_txfm2d_test.cc
+++ b/third_party/aom/test/av1_inv_txfm2d_test.cc
@@ -42,14 +42,6 @@ class AV1InvTxfm2d : public ::testing::TestWithParam<AV1InvTxfm2dParam> {
     max_avg_error_ = GET_PARAM(3);
     txfm1d_size_ = libaom_test::get_txfm1d_size(tx_size_);
     txfm2d_size_ = txfm1d_size_ * txfm1d_size_;
-    count_ = 500;
-
-    input_ = reinterpret_cast<int16_t *>(
-        aom_memalign(16, sizeof(int16_t) * txfm2d_size_));
-    ref_input_ = reinterpret_cast<uint16_t *>(
-        aom_memalign(16, sizeof(uint16_t) * txfm2d_size_));
-    output_ = reinterpret_cast<int32_t *>(
-        aom_memalign(16, sizeof(int32_t) * txfm2d_size_));
   }
 
   void RunRoundtripCheck() {
@@ -59,52 +51,51 @@ class AV1InvTxfm2d : public ::testing::TestWithParam<AV1InvTxfm2dParam> {
         libaom_test::inv_txfm_func_ls[tx_size_];
     double avg_abs_error = 0;
     ACMRandom rnd(ACMRandom::DeterministicSeed());
-    for (int ci = 0; ci < count_; ci++) {
+
+    const int count = 500;
+
+    for (int ci = 0; ci < count; ci++) {
+      int16_t expected[64 * 64] = { 0 };
+      ASSERT_LT(txfm2d_size_, NELEMENTS(expected));
+
       for (int ni = 0; ni < txfm2d_size_; ++ni) {
         if (ci == 0) {
           int extreme_input = input_base - 1;
-          input_[ni] = extreme_input;  // extreme case
-          ref_input_[ni] = 0;
+          expected[ni] = extreme_input;  // extreme case
         } else {
-          input_[ni] = rnd.Rand16() % input_base;
-          ref_input_[ni] = 0;
+          expected[ni] = rnd.Rand16() % input_base;
         }
       }
 
-      fwd_txfm_func(input_, output_, txfm1d_size_, tx_type_, bd);
-      inv_txfm_func(output_, ref_input_, txfm1d_size_, tx_type_, bd);
+      int32_t coeffs[64 * 64] = { 0 };
+      ASSERT_LT(txfm2d_size_, NELEMENTS(coeffs));
+      fwd_txfm_func(expected, coeffs, txfm1d_size_, tx_type_, bd);
+
+      uint16_t actual[64 * 64] = { 0 };
+      ASSERT_LT(txfm2d_size_, NELEMENTS(actual));
+      inv_txfm_func(coeffs, actual, txfm1d_size_, tx_type_, bd);
 
       for (int ni = 0; ni < txfm2d_size_; ++ni) {
-        EXPECT_GE(max_error_, abs(input_[ni] - ref_input_[ni]));
+        EXPECT_GE(max_error_, abs(expected[ni] - actual[ni]));
       }
       avg_abs_error += compute_avg_abs_error<int16_t, uint16_t>(
-          input_, ref_input_, txfm2d_size_);
+          expected, actual, txfm2d_size_);
     }
 
-    avg_abs_error /= count_;
+    avg_abs_error /= count;
     // max_abs_avg_error comes from upper bound of
     // printf("txfm1d_size: %d accuracy_avg_abs_error: %f\n",
     // txfm1d_size_, avg_abs_error);
     EXPECT_GE(max_avg_error_, avg_abs_error);
   }
 
-  virtual void TearDown() {
-    aom_free(input_);
-    aom_free(output_);
-    aom_free(ref_input_);
-  }
-
  private:
-  int count_;
   int max_error_;
   double max_avg_error_;
   TX_TYPE tx_type_;
   TX_SIZE tx_size_;
   int txfm1d_size_;
   int txfm2d_size_;
-  int16_t *input_;
-  uint16_t *ref_input_;
-  int32_t *output_;
 };
 
 TEST_P(AV1InvTxfm2d, RunRoundtripCheck) { RunRoundtripCheck(); }
diff --git a/third_party/aom/test/av1_inv_txfm_test.cc b/third_party/aom/test/av1_inv_txfm_test.cc
index af3fee872..34d45e08b 100644
--- a/third_party/aom/test/av1_inv_txfm_test.cc
+++ b/third_party/aom/test/av1_inv_txfm_test.cc
@@ -43,7 +43,6 @@ void reference_idct_1d(const double *in, double *out, int size) {
   }
 }
 
-typedef void (*IdctFuncRef)(const double *in, double *out, int size);
 typedef void (*IdctFunc)(const tran_low_t *in, tran_low_t *out);
 
 class TransTestBase {
@@ -52,10 +51,10 @@ class TransTestBase {
 
  protected:
   void RunInvAccuracyCheck() {
-    tran_low_t *input = new tran_low_t[txfm_size_];
-    tran_low_t *output = new tran_low_t[txfm_size_];
-    double *ref_input = new double[txfm_size_];
-    double *ref_output = new double[txfm_size_];
+    tran_low_t input[64];
+    tran_low_t output[64];
+    double ref_input[64];
+    double ref_output[64];
 
     ACMRandom rnd(ACMRandom::DeterministicSeed());
     const int count_test_block = 5000;
@@ -65,8 +64,8 @@ class TransTestBase {
         ref_input[ni] = static_cast<double>(input[ni]);
       }
 
-      fwd_txfm_(input, output);
-      fwd_txfm_ref_(ref_input, ref_output, txfm_size_);
+      inv_txfm_(input, output);
+      reference_idct_1d(ref_input, ref_output, txfm_size_);
 
       for (int ni = 0; ni < txfm_size_; ++ni) {
         EXPECT_LE(
@@ -74,40 +73,32 @@ class TransTestBase {
             max_error_);
       }
     }
-
-    delete[] input;
-    delete[] output;
-    delete[] ref_input;
-    delete[] ref_output;
   }
 
   double max_error_;
   int txfm_size_;
-  IdctFunc fwd_txfm_;
-  IdctFuncRef fwd_txfm_ref_;
+  IdctFunc inv_txfm_;
 };
 
-typedef std::tr1::tuple<IdctFunc, IdctFuncRef, int, int> IdctParam;
+typedef std::tr1::tuple<IdctFunc, int, int> IdctParam;
 class AV1InvTxfm : public TransTestBase,
                    public ::testing::TestWithParam<IdctParam> {
  public:
   virtual void SetUp() {
-    fwd_txfm_ = GET_PARAM(0);
-    fwd_txfm_ref_ = GET_PARAM(1);
-    txfm_size_ = GET_PARAM(2);
-    max_error_ = GET_PARAM(3);
+    inv_txfm_ = GET_PARAM(0);
+    txfm_size_ = GET_PARAM(1);
+    max_error_ = GET_PARAM(2);
   }
   virtual void TearDown() {}
 };
 
 TEST_P(AV1InvTxfm, RunInvAccuracyCheck) { RunInvAccuracyCheck(); }
 
-INSTANTIATE_TEST_CASE_P(
-    C, AV1InvTxfm,
-    ::testing::Values(IdctParam(&aom_idct4_c, &reference_idct_1d, 4, 1),
-                      IdctParam(&aom_idct8_c, &reference_idct_1d, 8, 2),
-                      IdctParam(&aom_idct16_c, &reference_idct_1d, 16, 4),
-                      IdctParam(&aom_idct32_c, &reference_idct_1d, 32, 6)));
+INSTANTIATE_TEST_CASE_P(C, AV1InvTxfm,
+                        ::testing::Values(IdctParam(&aom_idct4_c, 4, 1),
+                                          IdctParam(&aom_idct8_c, 8, 2),
+                                          IdctParam(&aom_idct16_c, 16, 4),
+                                          IdctParam(&aom_idct32_c, 32, 6)));
 
 #if CONFIG_AV1_ENCODER
 typedef void (*FwdTxfmFunc)(const int16_t *in, tran_low_t *out, int stride);
@@ -140,6 +131,13 @@ class AV1PartialIDctTest
 };
 
 #if !CONFIG_ADAPT_SCAN
+static MB_MODE_INFO get_mbmi() {
+  MB_MODE_INFO mbmi;
+  mbmi.ref_frame[0] = LAST_FRAME;
+  assert(is_inter_block(&mbmi));
+  return mbmi;
+}
+
 TEST_P(AV1PartialIDctTest, RunQuantCheck) {
   int size;
   switch (tx_size_) {
@@ -186,9 +184,10 @@ TEST_P(AV1PartialIDctTest, RunQuantCheck) {
 
       // quantization with maximum allowed step sizes
       test_coef_block1[0] = (output_ref_block[0] / 1336) * 1336;
+      MB_MODE_INFO mbmi = get_mbmi();
       for (int j = 1; j < last_nonzero_; ++j)
         test_coef_block1[get_scan((const AV1_COMMON *)NULL, tx_size_, DCT_DCT,
-                                  0)
+                                  &mbmi)
                              ->scan[j]] = (output_ref_block[j] / 1828) * 1828;
     }
 
@@ -239,7 +238,9 @@ TEST_P(AV1PartialIDctTest, ResultsMatch) {
         max_energy_leftover = 0;
         coef = 0;
       }
-      test_coef_block1[get_scan((const AV1_COMMON *)NULL, tx_size_, DCT_DCT, 0)
+      MB_MODE_INFO mbmi = get_mbmi();
+      test_coef_block1[get_scan((const AV1_COMMON *)NULL, tx_size_, DCT_DCT,
+                                &mbmi)
                            ->scan[j]] = coef;
     }
 
diff --git a/third_party/aom/test/block_error_test.cc b/third_party/aom/test/block_error_test.cc
new file mode 100644
index 000000000..4364af422
--- /dev/null
+++ b/third_party/aom/test/block_error_test.cc
@@ -0,0 +1,136 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include "third_party/googletest/src/googletest/include/gtest/gtest.h"
+
+#include "./aom_config.h"
+#include "./av1_rtcd.h"
+
+#include "test/acm_random.h"
+#include "test/clear_system_state.h"
+#include "test/register_state_check.h"
+#include "test/util.h"
+
+namespace {
+using libaom_test::ACMRandom;
+
+typedef int64_t (*BlockErrorFunc)(const tran_low_t *coeff,
+                                  const tran_low_t *dqcoeff, intptr_t size,
+                                  int64_t *ssz);
+#if CONFIG_HIGHBITDEPTH
+typedef int64_t (*HbdBlockErrorFunc)(const tran_low_t *coeff,
+                                     const tran_low_t *dqcoeff, intptr_t size,
+                                     int64_t *ssz, int bd);
+#endif
+
+typedef std::tr1::tuple<BlockErrorFunc, BlockErrorFunc, TX_SIZE,
+                        aom_bit_depth_t>
+    BlockErrorParam;
+
+const int kTestNum = 10000;
+
+class BlockErrorTest : public ::testing::TestWithParam<BlockErrorParam> {
+ public:
+  BlockErrorTest()
+      : blk_err_ref_(GET_PARAM(0)), blk_err_(GET_PARAM(1)),
+        tx_size_(GET_PARAM(2)), bd_(GET_PARAM(3)) {}
+
+  virtual ~BlockErrorTest() {}
+
+  virtual void SetUp() {
+    const intptr_t block_size = getCoeffNum();
+    coeff_ = reinterpret_cast<tran_low_t *>(
+        aom_memalign(16, 2 * block_size * sizeof(tran_low_t)));
+  }
+
+  virtual void TearDown() {
+    aom_free(coeff_);
+    coeff_ = NULL;
+    libaom_test::ClearSystemState();
+  }
+
+  void BlockErrorRun(int testNum) {
+    int i;
+    int64_t error_ref, error;
+    int64_t sse_ref, sse;
+    const intptr_t block_size = getCoeffNum();
+    tran_low_t *dqcoeff = coeff_ + block_size;
+    for (i = 0; i < testNum; ++i) {
+      FillRandomData();
+
+      error_ref = blk_err_ref_(coeff_, dqcoeff, block_size, &sse_ref);
+      ASM_REGISTER_STATE_CHECK(error =
+                                   blk_err_(coeff_, dqcoeff, block_size, &sse));
+
+      EXPECT_EQ(error_ref, error) << "Error doesn't match on test: " << i;
+      EXPECT_EQ(sse_ref, sse) << "SSE doesn't match on test: " << i;
+    }
+  }
+
+  intptr_t getCoeffNum() { return tx_size_2d[tx_size_]; }
+
+  void FillRandomData() {
+    const intptr_t block_size = getCoeffNum();
+    tran_low_t *dqcoeff = coeff_ + block_size;
+    intptr_t i;
+    int16_t margin = 512;
+    for (i = 0; i < block_size; ++i) {
+      coeff_[i] = GetRandomNumWithRange(INT16_MIN + margin, INT16_MAX - margin);
+      dqcoeff[i] = coeff_[i] + GetRandomDeltaWithRange(margin);
+    }
+  }
+
+  void FillConstantData() {
+    const intptr_t block_size = getCoeffNum();
+    tran_low_t *dqcoeff = coeff_ + block_size;
+    intptr_t i;
+    for (i = 0; i < block_size; ++i) {
+      coeff_[i] = 5;
+      dqcoeff[i] = 7;
+    }
+  }
+
+  tran_low_t GetRandomNumWithRange(int16_t min, int16_t max) {
+    return clamp((int16_t)rnd_.Rand16(), min, max);
+  }
+
+  tran_low_t GetRandomDeltaWithRange(int16_t delta) {
+    tran_low_t value = (int16_t)rnd_.Rand16();
+    value %= delta;
+    return value;
+  }
+
+  BlockErrorFunc blk_err_ref_;
+  BlockErrorFunc blk_err_;
+  TX_SIZE tx_size_;
+  aom_bit_depth_t bd_;
+  ACMRandom rnd_;
+  tran_low_t *coeff_;
+};
+
+TEST_P(BlockErrorTest, BitExact) { BlockErrorRun(kTestNum); }
+
+using std::tr1::make_tuple;
+
+#if !CONFIG_HIGHBITDEPTH && HAVE_SSE2
+const BlockErrorParam kBlkErrParamArraySse2[] = { make_tuple(
+    &av1_block_error_c, &av1_block_error_sse2, TX_32X32, AOM_BITS_8) };
+INSTANTIATE_TEST_CASE_P(SSE2, BlockErrorTest,
+                        ::testing::ValuesIn(kBlkErrParamArraySse2));
+#endif
+
+#if HAVE_AVX2
+const BlockErrorParam kBlkErrParamArrayAvx2[] = { make_tuple(
+    &av1_block_error_c, &av1_block_error_avx2, TX_32X32, AOM_BITS_8) };
+INSTANTIATE_TEST_CASE_P(AVX2, BlockErrorTest,
+                        ::testing::ValuesIn(kBlkErrParamArrayAvx2));
+#endif
+}  // namespace
diff --git a/third_party/aom/test/borders_test.cc b/third_party/aom/test/borders_test.cc
index d4e8dea23..ee771707c 100644
--- a/third_party/aom/test/borders_test.cc
+++ b/third_party/aom/test/borders_test.cc
@@ -20,8 +20,8 @@
 namespace {
 
 class BordersTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   BordersTest() : EncoderTest(GET_PARAM(0)) {}
   virtual ~BordersTest() {}
diff --git a/third_party/aom/test/coding_path_sync.cc b/third_party/aom/test/coding_path_sync.cc
new file mode 100644
index 000000000..68ee160bf
--- /dev/null
+++ b/third_party/aom/test/coding_path_sync.cc
@@ -0,0 +1,145 @@
+/*
+ * Copyright (c) 2016, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include <vector>
+#include "third_party/googletest/src/googletest/include/gtest/gtest.h"
+#include "test/acm_random.h"
+
+#include "./aom_config.h"
+
+#if CONFIG_AV1_ENCODER && CONFIG_AV1_DECODER
+
+#include "aom_ports/mem.h"  // ROUND_POWER_OF_TWO
+#include "aom/aomcx.h"
+#include "aom/aomdx.h"
+#include "aom/aom_encoder.h"
+#include "aom/aom_decoder.h"
+
+using libaom_test::ACMRandom;
+namespace {
+
+struct CompressedSource {
+  explicit CompressedSource(int seed) : rnd_(seed) {
+    frame_count_ = 0;
+    aom_codec_iface_t *algo = &aom_codec_av1_cx_algo;
+
+    aom_codec_enc_cfg_t cfg;
+    aom_codec_enc_config_default(algo, &cfg, 0);
+
+    const int max_q = cfg.rc_max_quantizer;
+
+    cfg.rc_end_usage = AOM_CQ;
+    cfg.rc_max_quantizer = max_q;
+    cfg.rc_min_quantizer = max_q;
+    cfg.g_w = WIDTH;
+    cfg.g_h = HEIGHT;
+    cfg.g_lag_in_frames = 0;
+
+    aom_codec_enc_init(&enc_, algo, &cfg, 0);
+  }
+
+  ~CompressedSource() { aom_codec_destroy(&enc_); }
+
+  const aom_codec_cx_pkt_t *readFrame() {
+    uint8_t buf[WIDTH * HEIGHT * 3 / 2] = { 0 };
+
+    // render regular pattern
+    const int period = rnd_.Rand8() % 32 + 1;
+    const int phase = rnd_.Rand8() % period;
+
+    const int val_a = rnd_.Rand8();
+    const int val_b = rnd_.Rand8();
+    for (int i = 0; i < (int)sizeof buf; ++i)
+      buf[i] = (i + phase) % period < period / 2 ? val_a : val_b;
+
+    aom_image_t img;
+    aom_img_wrap(&img, AOM_IMG_FMT_I420, WIDTH, HEIGHT, 0, buf);
+    aom_codec_encode(&enc_, &img, frame_count_++, 1, 0, 0);
+
+    aom_codec_iter_t iter = NULL;
+    return aom_codec_get_cx_data(&enc_, &iter);
+  }
+
+ private:
+  ACMRandom rnd_;
+  aom_codec_ctx_t enc_;
+  int frame_count_;
+  static const int WIDTH = 32;
+  static const int HEIGHT = 32;
+};
+
+// lowers an aom_image_t to a easily comparable/printable form
+std::vector<int16_t> serialize(const aom_image_t *img) {
+  const int w_uv = ROUND_POWER_OF_TWO(img->d_w, img->x_chroma_shift);
+  const int h_uv = ROUND_POWER_OF_TWO(img->d_h, img->y_chroma_shift);
+  const int w[] = { (int)img->d_w, w_uv, w_uv };
+  const int h[] = { (int)img->d_h, h_uv, h_uv };
+
+  std::vector<int16_t> bytes;
+  bytes.reserve(img->d_w * img->d_h * 3);
+  for (int plane = 0; plane < 3; ++plane)
+    for (int r = 0; r < h[plane]; ++r)
+      for (int c = 0; c < w[plane]; ++c) {
+        const int offset = r * img->stride[plane] + c;
+        if (img->fmt & AOM_IMG_FMT_HIGHBITDEPTH)
+          bytes.push_back(img->planes[plane][offset * 2]);
+        else
+          bytes.push_back(img->planes[plane][offset]);
+      }
+
+  return bytes;
+}
+
+struct Decoder {
+  explicit Decoder(int allowLowbitdepth) {
+    aom_codec_iface_t *algo = &aom_codec_av1_dx_algo;
+
+    aom_codec_dec_cfg cfg = { 0 };
+    cfg.allow_lowbitdepth = allowLowbitdepth;
+
+    aom_codec_dec_init(&dec_, algo, &cfg, 0);
+  }
+
+  ~Decoder() { aom_codec_destroy(&dec_); }
+
+  std::vector<int16_t> decode(const aom_codec_cx_pkt_t *pkt) {
+    aom_codec_decode(&dec_, (uint8_t *)pkt->data.frame.buf, pkt->data.frame.sz,
+                     NULL, 0);
+
+    aom_codec_iter_t iter = NULL;
+    return serialize(aom_codec_get_frame(&dec_, &iter));
+  }
+
+ private:
+  aom_codec_ctx_t dec_;
+};
+
+// Try to reveal a mismatch between LBD and HBD coding paths.
+TEST(CodingPathSync, SearchForHbdLbdMismatch) {
+  // disable test. Re-enable it locally to help diagnosing LBD/HBD mismatches.
+  // And re-enable it once both coding paths match
+  // so they don't diverge anymore.
+  return;
+
+  const int count_tests = 100;
+  for (int i = 0; i < count_tests; ++i) {
+    Decoder dec_HBD(0);
+    Decoder dec_LBD(1);
+
+    CompressedSource enc(i);
+    const aom_codec_cx_pkt_t *frame = enc.readFrame();
+    ASSERT_EQ(dec_LBD.decode(frame), dec_HBD.decode(frame));
+  }
+}
+
+}  // namespace
+
+#endif
diff --git a/third_party/aom/test/convolve_round_test.cc b/third_party/aom/test/convolve_round_test.cc
new file mode 100644
index 000000000..6f77dbb80
--- /dev/null
+++ b/third_party/aom/test/convolve_round_test.cc
@@ -0,0 +1,176 @@
+/*
+ * Copyright (c) 2017, Alliance for Open Media. All rights reserved
+ *
+ * This source code is subject to the terms of the BSD 2 Clause License and
+ * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
+ * was not distributed with this source code in the LICENSE file, you can
+ * obtain it at www.aomedia.org/license/software. If the Alliance for Open
+ * Media Patent License 1.0 was not distributed with this source code in the
+ * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
+ */
+
+#include <assert.h>
+
+#include "./av1_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/googletest/include/gtest/gtest.h"
+#include "aom/aom_integer.h"
+#include "aom_ports/aom_timer.h"
+
+using libaom_test::ACMRandom;
+
+namespace {
+#define CONVOLVE_ROUNDING_PARAM                                            \
+  const int32_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, \
+      int h, int bits
+
+typedef void (*ConvolveRoundFunc)(CONVOLVE_ROUNDING_PARAM);
+
+typedef void (*ConvolveRoundFuncHbd)(CONVOLVE_ROUNDING_PARAM, int bd);
+
+template <ConvolveRoundFuncHbd fn>
+void highbd_convolve_rounding_8(CONVOLVE_ROUNDING_PARAM) {
+  const int bd = 8;
+  fn(src, src_stride, dst, dst_stride, w, h, bits, bd);
+}
+
+template <ConvolveRoundFuncHbd fn>
+void highbd_convolve_rounding_10(CONVOLVE_ROUNDING_PARAM) {
+  const int bd = 10;
+  fn(src, src_stride, dst, dst_stride, w, h, bits, bd);
+}
+
+template <ConvolveRoundFuncHbd fn>
+void highbd_convolve_rounding_12(CONVOLVE_ROUNDING_PARAM) {
+  const int bd = 12;
+  fn(src, src_stride, dst, dst_stride, w, h, bits, bd);
+}
+
+typedef enum { LOWBITDEPTH_TEST, HIGHBITDEPTH_TEST } DataPathType;
+
+using std::tr1::tuple;
+
+typedef tuple<ConvolveRoundFunc, ConvolveRoundFunc, DataPathType>
+    ConvolveRoundParam;
+
+const int kTestNum = 5000;
+
+class ConvolveRoundTest : public ::testing::TestWithParam<ConvolveRoundParam> {
+ protected:
+  ConvolveRoundTest()
+      : func_ref_(GET_PARAM(0)), func_(GET_PARAM(1)), data_path_(GET_PARAM(2)) {
+  }
+  virtual ~ConvolveRoundTest() {}
+
+  virtual void SetUp() {
+    const size_t block_size = 128 * 128;
+    src_ = reinterpret_cast<int32_t *>(
+        aom_memalign(16, 3 * block_size * sizeof(int32_t)));
+    dst_ref_ = reinterpret_cast<uint16_t *>(src_ + block_size);
+    dst_ = dst_ref_ + block_size;
+  }
+
+  virtual void TearDown() { aom_free(src_); }
+
+  void ConvolveRoundingRun() {
+    int test_num = 0;
+    const int src_stride = 128;
+    const int dst_stride = 128;
+    int bits = 13;
+    uint8_t *dst = 0;
+    uint8_t *dst_ref = 0;
+    int diff_wide;
+
+    if (data_path_ == LOWBITDEPTH_TEST) {
+      dst = reinterpret_cast<uint8_t *>(dst_);
+      dst_ref = reinterpret_cast<uint8_t *>(dst_ref_);
+#if CONFIG_HIGHBITDEPTH
+    } else if (data_path_ == HIGHBITDEPTH_TEST) {
+      dst = CONVERT_TO_BYTEPTR(dst_);
+      dst_ref = CONVERT_TO_BYTEPTR(dst_ref_);
+#endif
+    } else {
+      assert(0);
+    }
+
+    while (test_num < kTestNum) {
+      int block_size = test_num % BLOCK_SIZES_ALL;
+      int w = block_size_wide[block_size];
+      int h = block_size_high[block_size];
+
+      if (test_num % 2 == 0)
+        bits -= 1;
+      else
+        bits += 1;
+
+      GenerateBufferWithRandom(src_, src_stride, bits, w, h);
+
+      func_ref_(src_, src_stride, dst_ref, dst_stride, w, h, bits);
+      func_(src_, src_stride, dst, dst_stride, w, h, bits);
+
+      diff_wide = w;
+      if (data_path_ == LOWBITDEPTH_TEST) diff_wide >>= 1;
+      for (int r = 0; r < h; ++r) {
+        for (int c = 0; c < diff_wide; ++c) {
+          ASSERT_EQ(dst_ref_[r * dst_stride + c], dst_[r * dst_stride + c])
+              << "Mismatch at r: " << r << " c: " << c << " test: " << test_num;
+        }
+      }
+
+      test_num++;
+    }
+  }
+
+  void GenerateBufferWithRandom(int32_t *src, int src_stride, int bits, int w,
+                                int h) {
+    int32_t number;
+    for (int r = 0; r < h; ++r) {
+      for (int c = 0; c < w; ++c) {
+        number = static_cast<int32_t>(rand_.Rand31());
+        number %= 1 << (bits + 9);
+        src[r * src_stride + c] = number;
+      }
+    }
+  }
+
+  ACMRandom rand_;
+  int32_t *src_;
+  uint16_t *dst_ref_;
+  uint16_t *dst_;
+
+  ConvolveRoundFunc func_ref_;
+  ConvolveRoundFunc func_;
+  DataPathType data_path_;
+};
+
+TEST_P(ConvolveRoundTest, BitExactCheck) { ConvolveRoundingRun(); }
+
+using std::tr1::make_tuple;
+
+#if HAVE_AVX2
+#if CONFIG_HIGHBITDEPTH
+const ConvolveRoundParam kConvRndParamArray[] = {
+  make_tuple(&av1_convolve_rounding_c, &av1_convolve_rounding_avx2,
+             LOWBITDEPTH_TEST),
+  make_tuple(&highbd_convolve_rounding_8<av1_highbd_convolve_rounding_c>,
+             &highbd_convolve_rounding_8<av1_highbd_convolve_rounding_avx2>,
+             HIGHBITDEPTH_TEST),
+  make_tuple(&highbd_convolve_rounding_10<av1_highbd_convolve_rounding_c>,
+             &highbd_convolve_rounding_10<av1_highbd_convolve_rounding_avx2>,
+             HIGHBITDEPTH_TEST),
+  make_tuple(&highbd_convolve_rounding_12<av1_highbd_convolve_rounding_c>,
+             &highbd_convolve_rounding_12<av1_highbd_convolve_rounding_avx2>,
+             HIGHBITDEPTH_TEST)
+};
+#else
+const ConvolveRoundParam kConvRndParamArray[] = { make_tuple(
+    &av1_convolve_rounding_c, &av1_convolve_rounding_avx2, LOWBITDEPTH_TEST) };
+#endif
+
+INSTANTIATE_TEST_CASE_P(AVX2, ConvolveRoundTest,
+                        ::testing::ValuesIn(kConvRndParamArray));
+#endif  // HAVE_AVX2
+}  // namespace
diff --git a/third_party/aom/test/cpu_speed_test.cc b/third_party/aom/test/cpu_speed_test.cc
index ad0f2a874..bde00472d 100644
--- a/third_party/aom/test/cpu_speed_test.cc
+++ b/third_party/aom/test/cpu_speed_test.cc
@@ -21,8 +21,8 @@ namespace {
 const int kMaxPSNR = 100;
 
 class CpuSpeedTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   CpuSpeedTest()
       : EncoderTest(GET_PARAM(0)), encoding_mode_(GET_PARAM(1)),
diff --git a/third_party/aom/test/datarate_test.cc b/third_party/aom/test/datarate_test.cc
index a4a682681..6a1b4e101 100644
--- a/third_party/aom/test/datarate_test.cc
+++ b/third_party/aom/test/datarate_test.cc
@@ -21,8 +21,8 @@
 namespace {
 
 class DatarateTestLarge
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  public:
   DatarateTestLarge() : EncoderTest(GET_PARAM(0)) {}
 
diff --git a/third_party/aom/test/dct16x16_test.cc b/third_party/aom/test/dct16x16_test.cc
index 89263ce89..c2c072494 100644
--- a/third_party/aom/test/dct16x16_test.cc
+++ b/third_party/aom/test/dct16x16_test.cc
@@ -226,9 +226,9 @@ void reference_16x16_dct_2d(int16_t input[256], double output[256]) {
 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);
+                        TxfmParam *txfm_param);
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 
 typedef std::tr1::tuple<FdctFunc, IdctFunc, int, aom_bit_depth_t> Dct16x16Param;
 typedef std::tr1::tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t> Ht16x16Param;
@@ -236,31 +236,46 @@ typedef std::tr1::tuple<IdctFunc, IdctFunc, int, aom_bit_depth_t>
     Idct16x16Param;
 
 void fdct16x16_ref(const int16_t *in, tran_low_t *out, int stride,
-                   int /*tx_type*/) {
+                   TxfmParam * /*txfm_param*/) {
   aom_fdct16x16_c(in, out, stride);
 }
 
 void idct16x16_ref(const tran_low_t *in, uint8_t *dest, int stride,
-                   int /*tx_type*/) {
+                   const TxfmParam * /*txfm_param*/) {
   aom_idct16x16_256_add_c(in, dest, stride);
 }
 
-void fht16x16_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht16x16_c(in, out, stride, tx_type);
+void fht16x16_ref(const int16_t *in, tran_low_t *out, int stride,
+                  TxfmParam *txfm_param) {
+  av1_fht16x16_c(in, out, stride, txfm_param);
 }
 
 void iht16x16_ref(const tran_low_t *in, uint8_t *dest, int stride,
-                  int tx_type) {
-  av1_iht16x16_256_add_c(in, dest, stride, tx_type);
+                  const TxfmParam *txfm_param) {
+  av1_iht16x16_256_add_c(in, dest, stride, txfm_param);
 }
 
 #if CONFIG_HIGHBITDEPTH
-void iht16x16_10(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_highbd_iht16x16_256_add_c(in, out, stride, tx_type, 10);
+void fht16x16_10(const int16_t *in, tran_low_t *out, int stride,
+                 TxfmParam *txfm_param) {
+  av1_fwd_txfm2d_16x16_c(in, out, stride, txfm_param->tx_type, 10);
 }
 
-void iht16x16_12(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_highbd_iht16x16_256_add_c(in, out, stride, tx_type, 12);
+void fht16x16_12(const int16_t *in, tran_low_t *out, int stride,
+                 TxfmParam *txfm_param) {
+  av1_fwd_txfm2d_16x16_c(in, out, stride, txfm_param->tx_type, 12);
+}
+
+void iht16x16_10(const tran_low_t *in, uint8_t *out, int stride,
+                 const TxfmParam *txfm_param) {
+  av1_inv_txfm2d_add_16x16_c(in, CONVERT_TO_SHORTPTR(out), stride,
+                             txfm_param->tx_type, 10);
+}
+
+void iht16x16_12(const tran_low_t *in, uint8_t *out, int stride,
+                 const TxfmParam *txfm_param) {
+  av1_inv_txfm2d_add_16x16_c(in, CONVERT_TO_SHORTPTR(out), stride,
+                             txfm_param->tx_type, 12);
 }
 #endif  // CONFIG_HIGHBITDEPTH
 
@@ -346,7 +361,7 @@ class Trans16x16TestBase {
       for (int j = 0; j < kNumCoeffs; ++j)
         input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);
 
-      fwd_txfm_ref(input_block, output_ref_block, pitch_, tx_type_);
+      fwd_txfm_ref(input_block, output_ref_block, pitch_, &txfm_param_);
       ASM_REGISTER_STATE_CHECK(RunFwdTxfm(input_block, output_block, pitch_));
 
       // The minimum quant value is 4.
@@ -373,7 +388,7 @@ class Trans16x16TestBase {
         for (int j = 0; j < kNumCoeffs; ++j) input_extreme_block[j] = -mask_;
       }
 
-      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, tx_type_);
+      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, &txfm_param_);
       ASM_REGISTER_STATE_CHECK(
           RunFwdTxfm(input_extreme_block, output_block, pitch_));
 
@@ -409,7 +424,7 @@ class Trans16x16TestBase {
       if (i == 1)
         for (int j = 0; j < kNumCoeffs; ++j) input_extreme_block[j] = -mask_;
 
-      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, tx_type_);
+      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, &txfm_param_);
 
       // clear reconstructed pixel buffers
       memset(dst, 0, kNumCoeffs * sizeof(uint8_t));
@@ -424,12 +439,12 @@ class Trans16x16TestBase {
       for (int j = 1; j < kNumCoeffs; ++j)
         output_ref_block[j] = (output_ref_block[j] / ac_thred) * ac_thred;
       if (bit_depth_ == AOM_BITS_8) {
-        inv_txfm_ref(output_ref_block, ref, pitch_, tx_type_);
+        inv_txfm_ref(output_ref_block, ref, pitch_, &txfm_param_);
         ASM_REGISTER_STATE_CHECK(RunInvTxfm(output_ref_block, dst, pitch_));
 #if CONFIG_HIGHBITDEPTH
       } else {
         inv_txfm_ref(output_ref_block, CONVERT_TO_BYTEPTR(ref16), pitch_,
-                     tx_type_);
+                     &txfm_param_);
         ASM_REGISTER_STATE_CHECK(
             RunInvTxfm(output_ref_block, CONVERT_TO_BYTEPTR(dst16), pitch_));
 #endif
@@ -558,11 +573,11 @@ class Trans16x16TestBase {
   }
 
   int pitch_;
-  int tx_type_;
   aom_bit_depth_t bit_depth_;
   int mask_;
   FhtFunc fwd_txfm_ref;
   IhtFunc inv_txfm_ref;
+  TxfmParam txfm_param_;
 };
 
 class Trans16x16DCT : public Trans16x16TestBase,
@@ -573,13 +588,13 @@ class Trans16x16DCT : public Trans16x16TestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     bit_depth_ = GET_PARAM(3);
     pitch_ = 16;
     fwd_txfm_ref = fdct16x16_ref;
     inv_txfm_ref = idct16x16_ref;
     mask_ = (1 << bit_depth_) - 1;
     inv_txfm_ref = idct16x16_ref;
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
@@ -617,19 +632,27 @@ class Trans16x16HT : public Trans16x16TestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     bit_depth_ = GET_PARAM(3);
     pitch_ = 16;
-    fwd_txfm_ref = fht16x16_ref;
-    inv_txfm_ref = iht16x16_ref;
     mask_ = (1 << bit_depth_) - 1;
+    txfm_param_.tx_type = GET_PARAM(2);
 #if CONFIG_HIGHBITDEPTH
     switch (bit_depth_) {
-      case AOM_BITS_10: inv_txfm_ref = iht16x16_10; break;
-      case AOM_BITS_12: inv_txfm_ref = iht16x16_12; break;
-      default: inv_txfm_ref = iht16x16_ref; break;
+      case AOM_BITS_10:
+        fwd_txfm_ref = fht16x16_10;
+        inv_txfm_ref = iht16x16_10;
+        break;
+      case AOM_BITS_12:
+        fwd_txfm_ref = fht16x16_12;
+        inv_txfm_ref = iht16x16_12;
+        break;
+      default:
+        fwd_txfm_ref = fht16x16_ref;
+        inv_txfm_ref = iht16x16_ref;
+        break;
     }
 #else
+    fwd_txfm_ref = fht16x16_ref;
     inv_txfm_ref = iht16x16_ref;
 #endif
   }
@@ -637,10 +660,10 @@ class Trans16x16HT : public Trans16x16TestBase,
 
  protected:
   void RunFwdTxfm(int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
   void RunInvTxfm(tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
@@ -767,23 +790,18 @@ INSTANTIATE_TEST_CASE_P(C, Trans16x16DCT,
 INSTANTIATE_TEST_CASE_P(
     C, Trans16x16HT,
     ::testing::Values(
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_10, 0, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_10, 1, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_10, 2, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_10, 3, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_12, 0, AOM_BITS_12),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_12, 1, AOM_BITS_12),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_12, 2, AOM_BITS_12),
-        make_tuple(&av1_highbd_fht16x16_c, &iht16x16_12, 3, AOM_BITS_12),
+        make_tuple(&fht16x16_10, &iht16x16_10, 0, AOM_BITS_10),
+        make_tuple(&fht16x16_10, &iht16x16_10, 1, AOM_BITS_10),
+        make_tuple(&fht16x16_10, &iht16x16_10, 2, AOM_BITS_10),
+        make_tuple(&fht16x16_10, &iht16x16_10, 3, AOM_BITS_10),
+        make_tuple(&fht16x16_12, &iht16x16_12, 0, AOM_BITS_12),
+        make_tuple(&fht16x16_12, &iht16x16_12, 1, AOM_BITS_12),
+        make_tuple(&fht16x16_12, &iht16x16_12, 2, AOM_BITS_12),
+        make_tuple(&fht16x16_12, &iht16x16_12, 3, AOM_BITS_12),
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 0, AOM_BITS_8),
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 1, AOM_BITS_8),
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 2, AOM_BITS_8),
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 3, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(
-    C, PartialTrans16x16Test,
-    ::testing::Values(make_tuple(&aom_highbd_fdct16x16_1_c, AOM_BITS_8),
-                      make_tuple(&aom_highbd_fdct16x16_1_c, AOM_BITS_10),
-                      make_tuple(&aom_highbd_fdct16x16_1_c, AOM_BITS_12)));
 #else
 INSTANTIATE_TEST_CASE_P(
     C, Trans16x16HT,
@@ -792,9 +810,6 @@ INSTANTIATE_TEST_CASE_P(
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 1, AOM_BITS_8),
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 2, AOM_BITS_8),
         make_tuple(&av1_fht16x16_c, &av1_iht16x16_256_add_c, 3, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(C, PartialTrans16x16Test,
-                        ::testing::Values(make_tuple(&aom_fdct16x16_1_c,
-                                                     AOM_BITS_8)));
 #endif  // CONFIG_HIGHBITDEPTH
 
 #if HAVE_NEON_ASM && !CONFIG_HIGHBITDEPTH
@@ -819,17 +834,8 @@ INSTANTIATE_TEST_CASE_P(
                                  2, AOM_BITS_8),
                       make_tuple(&av1_fht16x16_sse2, &av1_iht16x16_256_add_sse2,
                                  3, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(SSE2, PartialTrans16x16Test,
-                        ::testing::Values(make_tuple(&aom_fdct16x16_1_sse2,
-                                                     AOM_BITS_8)));
 #endif  // HAVE_SSE2 && !CONFIG_HIGHBITDEPTH
 
-#if HAVE_AVX2 && !CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(AVX2, PartialTrans16x16Test,
-                        ::testing::Values(make_tuple(&aom_fdct16x16_1_avx2,
-                                                     AOM_BITS_8)));
-#endif  // HAVE_AVX2 && !CONFIG_HIGHBITDEPTH
-
 #if HAVE_SSE2 && CONFIG_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(SSE2, Trans16x16DCT,
                         ::testing::Values(make_tuple(&aom_fdct16x16_sse2,
@@ -843,14 +849,6 @@ INSTANTIATE_TEST_CASE_P(
         make_tuple(&av1_fht16x16_sse2, &av1_iht16x16_256_add_c, 2, AOM_BITS_8),
         make_tuple(&av1_fht16x16_sse2, &av1_iht16x16_256_add_c, 3,
                    AOM_BITS_8)));
-// TODO(luoyi):
-// For this test case, we should test function: aom_highbd_fdct16x16_1_sse2.
-// However this function is not available yet. if we mistakely test
-// aom_fdct16x16_1_sse2, it could only pass AOM_BITS_8/AOM_BITS_10 but not
-// AOM_BITS_12.
-INSTANTIATE_TEST_CASE_P(SSE2, PartialTrans16x16Test,
-                        ::testing::Values(make_tuple(&aom_fdct16x16_1_sse2,
-                                                     AOM_BITS_8)));
 #endif  // HAVE_SSE2 && CONFIG_HIGHBITDEPTH
 
 #if HAVE_MSA && !CONFIG_HIGHBITDEPTH
@@ -869,8 +867,5 @@ INSTANTIATE_TEST_CASE_P(
         make_tuple(&av1_fht16x16_msa, &av1_iht16x16_256_add_msa, 3,
                    AOM_BITS_8)));
 #endif  // !CONFIG_EXT_TX
-INSTANTIATE_TEST_CASE_P(MSA, PartialTrans16x16Test,
-                        ::testing::Values(make_tuple(&aom_fdct16x16_1_msa,
-                                                     AOM_BITS_8)));
 #endif  // HAVE_MSA && !CONFIG_HIGHBITDEPTH
 }  // namespace
diff --git a/third_party/aom/test/dct32x32_test.cc b/third_party/aom/test/dct32x32_test.cc
index 7c1db6501..0a30f7f38 100644
--- a/third_party/aom/test/dct32x32_test.cc
+++ b/third_party/aom/test/dct32x32_test.cc
@@ -350,11 +350,6 @@ INSTANTIATE_TEST_CASE_P(
                                  AOM_BITS_8),
                       make_tuple(&aom_fdct32x32_rd_c, &aom_idct32x32_1024_add_c,
                                  1, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(
-    C, PartialTrans32x32Test,
-    ::testing::Values(make_tuple(&aom_highbd_fdct32x32_1_c, AOM_BITS_8),
-                      make_tuple(&aom_highbd_fdct32x32_1_c, AOM_BITS_10),
-                      make_tuple(&aom_highbd_fdct32x32_1_c, AOM_BITS_12)));
 #else
 INSTANTIATE_TEST_CASE_P(
     C, Trans32x32Test,
@@ -362,9 +357,6 @@ INSTANTIATE_TEST_CASE_P(
                                  AOM_BITS_8),
                       make_tuple(&aom_fdct32x32_rd_c, &aom_idct32x32_1024_add_c,
                                  1, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(C, PartialTrans32x32Test,
-                        ::testing::Values(make_tuple(&aom_fdct32x32_1_c,
-                                                     AOM_BITS_8)));
 #endif  // CONFIG_HIGHBITDEPTH
 
 #if HAVE_NEON && !CONFIG_HIGHBITDEPTH
@@ -383,17 +375,8 @@ INSTANTIATE_TEST_CASE_P(
                                  &aom_idct32x32_1024_add_sse2, 0, AOM_BITS_8),
                       make_tuple(&aom_fdct32x32_rd_sse2,
                                  &aom_idct32x32_1024_add_sse2, 1, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(SSE2, PartialTrans32x32Test,
-                        ::testing::Values(make_tuple(&aom_fdct32x32_1_sse2,
-                                                     AOM_BITS_8)));
 #endif  // HAVE_SSE2 && !CONFIG_HIGHBITDEPTH
 
-#if HAVE_AVX2 && !CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(AVX2, PartialTrans32x32Test,
-                        ::testing::Values(make_tuple(&aom_fdct32x32_1_avx2,
-                                                     AOM_BITS_8)));
-#endif  // HAVE_AVX2 && !CONFIG_HIGHBITDEPTH
-
 #if HAVE_SSE2 && CONFIG_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(
     SSE2, Trans32x32Test,
@@ -401,9 +384,6 @@ INSTANTIATE_TEST_CASE_P(
                                  0, AOM_BITS_8),
                       make_tuple(&aom_fdct32x32_rd_sse2,
                                  &aom_idct32x32_1024_add_c, 1, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(SSE2, PartialTrans32x32Test,
-                        ::testing::Values(make_tuple(&aom_fdct32x32_1_sse2,
-                                                     AOM_BITS_8)));
 #endif  // HAVE_SSE2 && CONFIG_HIGHBITDEPTH
 
 #if HAVE_AVX2 && !CONFIG_HIGHBITDEPTH
@@ -431,8 +411,5 @@ INSTANTIATE_TEST_CASE_P(
                                  &aom_idct32x32_1024_add_msa, 0, AOM_BITS_8),
                       make_tuple(&aom_fdct32x32_rd_msa,
                                  &aom_idct32x32_1024_add_msa, 1, AOM_BITS_8)));
-INSTANTIATE_TEST_CASE_P(MSA, PartialTrans32x32Test,
-                        ::testing::Values(make_tuple(&aom_fdct32x32_1_msa,
-                                                     AOM_BITS_8)));
 #endif  // HAVE_MSA && !CONFIG_HIGHBITDEPTH
 }  // namespace
diff --git a/third_party/aom/test/decode_api_test.cc b/third_party/aom/test/decode_api_test.cc
index 6bd72a45d..8b1c9d268 100644
--- a/third_party/aom/test/decode_api_test.cc
+++ b/third_party/aom/test/decode_api_test.cc
@@ -13,13 +13,12 @@
 
 #include "./aom_config.h"
 #include "test/ivf_video_source.h"
+#include "test/util.h"
 #include "aom/aomdx.h"
 #include "aom/aom_decoder.h"
 
 namespace {
 
-#define NELEMENTS(x) static_cast<int>(sizeof(x) / sizeof(x[0]))
-
 TEST(DecodeAPI, InvalidParams) {
   static const aom_codec_iface_t *kCodecs[] = {
 #if CONFIG_AV1_DECODER
diff --git a/third_party/aom/test/decode_perf_test.cc b/third_party/aom/test/decode_perf_test.cc
index ede4f8849..a24d02a6c 100644
--- a/third_party/aom/test/decode_perf_test.cc
+++ b/third_party/aom/test/decode_perf_test.cc
@@ -63,6 +63,7 @@ TEST_P(DecodePerfTest, PerfTest) {
 
   aom_codec_dec_cfg_t cfg = aom_codec_dec_cfg_t();
   cfg.threads = threads;
+  cfg.allow_lowbitdepth = 1;
   libaom_test::AV1Decoder decoder(cfg, 0);
 
   aom_usec_timer t;
@@ -93,8 +94,8 @@ TEST_P(DecodePerfTest, PerfTest) {
 //                        ::testing::ValuesIn(kAV1DecodePerfVectors));
 
 class AV1NewEncodeDecodePerfTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   AV1NewEncodeDecodePerfTest()
       : EncoderTest(GET_PARAM(0)), encoding_mode_(GET_PARAM(1)), speed_(0),
@@ -115,7 +116,6 @@ class AV1NewEncodeDecodePerfTest
     cfg_.rc_buf_sz = 1000;
     cfg_.rc_buf_initial_sz = 500;
     cfg_.rc_buf_optimal_sz = 600;
-    cfg_.rc_resize_allowed = 0;
     cfg_.rc_end_usage = AOM_VBR;
   }
 
@@ -211,6 +211,7 @@ TEST_P(AV1NewEncodeDecodePerfTest, PerfTest) {
 
   aom_codec_dec_cfg_t cfg = aom_codec_dec_cfg_t();
   cfg.threads = threads;
+  cfg.allow_lowbitdepth = 1;
   libaom_test::AV1Decoder decoder(cfg, 0);
 
   aom_usec_timer t;
diff --git a/third_party/aom/test/encode_api_test.cc b/third_party/aom/test/encode_api_test.cc
index 14e43c847..80c42fee4 100644
--- a/third_party/aom/test/encode_api_test.cc
+++ b/third_party/aom/test/encode_api_test.cc
@@ -12,13 +12,12 @@
 #include "third_party/googletest/src/googletest/include/gtest/gtest.h"
 
 #include "./aom_config.h"
+#include "test/util.h"
 #include "aom/aomcx.h"
 #include "aom/aom_encoder.h"
 
 namespace {
 
-#define NELEMENTS(x) static_cast<int>(sizeof(x) / sizeof(x[0]))
-
 TEST(EncodeAPI, InvalidParams) {
   static const aom_codec_iface_t *kCodecs[] = {
 #if CONFIG_AV1_ENCODER
diff --git a/third_party/aom/test/encode_perf_test.cc b/third_party/aom/test/encode_perf_test.cc
index e2a4f2b71..5a37b480b 100644
--- a/third_party/aom/test/encode_perf_test.cc
+++ b/third_party/aom/test/encode_perf_test.cc
@@ -53,11 +53,9 @@ const EncodePerfTestVideo kAV1EncodePerfTestVectors[] = {
 const int kEncodePerfTestSpeeds[] = { 5, 6, 7, 8 };
 const int kEncodePerfTestThreads[] = { 1, 2, 4 };
 
-#define NELEMENTS(x) (sizeof((x)) / sizeof((x)[0]))
-
 class AV1EncodePerfTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   AV1EncodePerfTest()
       : EncoderTest(GET_PARAM(0)), min_psnr_(kMaxPsnr), nframes_(0),
@@ -78,7 +76,6 @@ class AV1EncodePerfTest
     cfg_.rc_buf_sz = 1000;
     cfg_.rc_buf_initial_sz = 500;
     cfg_.rc_buf_optimal_sz = 600;
-    cfg_.rc_resize_allowed = 0;
     cfg_.rc_end_usage = AOM_CBR;
     cfg_.g_error_resilient = 1;
     cfg_.g_threads = threads_;
diff --git a/third_party/aom/test/encode_test_driver.cc b/third_party/aom/test/encode_test_driver.cc
index 80f155ab2..ec168e969 100644
--- a/third_party/aom/test/encode_test_driver.cc
+++ b/third_party/aom/test/encode_test_driver.cc
@@ -99,7 +99,6 @@ void Encoder::Flush() {
 
 void EncoderTest::InitializeConfig() {
   const aom_codec_err_t res = codec_->DefaultEncoderConfig(&cfg_, 0);
-  dec_cfg_ = aom_codec_dec_cfg_t();
   ASSERT_EQ(AOM_CODEC_OK, res);
 }
 
@@ -211,6 +210,7 @@ void EncoderTest::MismatchHook(const aom_image_t *img_enc,
 
 void EncoderTest::RunLoop(VideoSource *video) {
   aom_codec_dec_cfg_t dec_cfg = aom_codec_dec_cfg_t();
+  dec_cfg.allow_lowbitdepth = 1;
 
   stats_.Reset();
 
diff --git a/third_party/aom/test/encode_test_driver.h b/third_party/aom/test/encode_test_driver.h
index 91027b4f6..97c1bf860 100644
--- a/third_party/aom/test/encode_test_driver.h
+++ b/third_party/aom/test/encode_test_driver.h
@@ -232,7 +232,6 @@ class EncoderTest {
 
   bool abort_;
   aom_codec_enc_cfg_t cfg_;
-  aom_codec_dec_cfg_t dec_cfg_;
   unsigned int passes_;
   unsigned long deadline_;
   TwopassStatsStore stats_;
diff --git a/third_party/aom/test/encoder_parms_get_to_decoder.cc b/third_party/aom/test/encoder_parms_get_to_decoder.cc
index ca6a24ebe..227ee8246 100644
--- a/third_party/aom/test/encoder_parms_get_to_decoder.cc
+++ b/third_party/aom/test/encoder_parms_get_to_decoder.cc
@@ -41,6 +41,10 @@ struct EncodeParameters {
   int32_t frame_parallel;
   aom_color_range_t color_range;
   aom_color_space_t cs;
+#if CONFIG_COLORSPACE_HEADERS
+  aom_transfer_function_t tf;
+  aom_chroma_sample_position_t csp;
+#endif
   int render_size[2];
   // TODO(JBB): quantizers / bitrate
 };
@@ -48,15 +52,20 @@ struct EncodeParameters {
 const EncodeParameters kAV1EncodeParameterSet[] = {
   { 0, 0, 0, 1, 0, AOM_CR_STUDIO_RANGE, AOM_CS_BT_601, { 0, 0 } },
   { 0, 0, 0, 0, 0, AOM_CR_FULL_RANGE, AOM_CS_BT_709, { 0, 0 } },
+#if CONFIG_COLORSPACE_HEADERS
+  { 0, 0, 1, 0, 0, AOM_CR_FULL_RANGE, AOM_CS_BT_2020_NCL, { 0, 0 } },
+#else
   { 0, 0, 1, 0, 0, AOM_CR_FULL_RANGE, AOM_CS_BT_2020, { 0, 0 } },
+#endif
   { 0, 2, 0, 0, 1, AOM_CR_STUDIO_RANGE, AOM_CS_UNKNOWN, { 640, 480 } },
   // TODO(JBB): Test profiles (requires more work).
 };
 
 class AvxEncoderParmsGetToDecoder
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<EncodeParameters,
-                                                 EncodePerfTestVideo> {
+    : public ::libaom_test::CodecTestWith2Params<EncodeParameters,
+                                                 EncodePerfTestVideo>,
+      public ::libaom_test::EncoderTest,
+{
  protected:
   AvxEncoderParmsGetToDecoder()
       : EncoderTest(GET_PARAM(0)), encode_parms(GET_PARAM(1)) {}
@@ -77,6 +86,10 @@ class AvxEncoderParmsGetToDecoder
                                   ::libaom_test::Encoder *encoder) {
     if (video->frame() == 1) {
       encoder->Control(AV1E_SET_COLOR_SPACE, encode_parms.cs);
+#if CONFIG_COLORSPACE_HEADERS
+      encoder->Control(AV1E_SET_TRANSFER_FUNCTION, encode_parms.tf);
+      encoder->Control(AV1E_SET_CHROMA_SAMPLE_POSITION, encode_parms.csp);
+#endif
       encoder->Control(AV1E_SET_COLOR_RANGE, encode_parms.color_range);
       encoder->Control(AV1E_SET_LOSSLESS, encode_parms.lossless);
       encoder->Control(AV1E_SET_FRAME_PARALLEL_DECODING,
@@ -114,6 +127,10 @@ class AvxEncoderParmsGetToDecoder
     }
     EXPECT_EQ(encode_parms.color_range, common->color_range);
     EXPECT_EQ(encode_parms.cs, common->color_space);
+#if CONFIG_COLORSPACE_HEADERS
+    EXPECT_EQ(encode_parms.tf, common->transfer_function);
+    EXPECT_EQ(encode_parms.csp, common->chroma_sample_position);
+#endif
     if (encode_parms.render_size[0] > 0 && encode_parms.render_size[1] > 0) {
       EXPECT_EQ(encode_parms.render_size[0], common->render_width);
       EXPECT_EQ(encode_parms.render_size[1], common->render_height);
diff --git a/third_party/aom/test/end_to_end_test.cc b/third_party/aom/test/end_to_end_test.cc
index 0c8cbe274..93bc1625e 100644
--- a/third_party/aom/test/end_to_end_test.cc
+++ b/third_party/aom/test/end_to_end_test.cc
@@ -86,9 +86,9 @@ int is_extension_y4m(const char *filename) {
 }
 
 class EndToEndTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith3Params<libaom_test::TestMode,
-                                                 TestVideoParam, int> {
+    : public ::libaom_test::CodecTestWith3Params<libaom_test::TestMode,
+                                                 TestVideoParam, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   EndToEndTest()
       : EncoderTest(GET_PARAM(0)), test_video_param_(GET_PARAM(2)),
@@ -110,7 +110,6 @@ class EndToEndTest
       cfg_.rc_buf_initial_sz = 500;
       cfg_.rc_buf_optimal_sz = 600;
     }
-    dec_cfg_.threads = 4;
   }
 
   virtual void BeginPassHook(unsigned int) {
diff --git a/third_party/aom/test/error_resilience_test.cc b/third_party/aom/test/error_resilience_test.cc
index b1d93a01f..e9abdde6d 100644
--- a/third_party/aom/test/error_resilience_test.cc
+++ b/third_party/aom/test/error_resilience_test.cc
@@ -21,8 +21,8 @@ const int kMaxErrorFrames = 12;
 const int kMaxDroppableFrames = 12;
 
 class ErrorResilienceTestLarge
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   ErrorResilienceTestLarge()
       : EncoderTest(GET_PARAM(0)), psnr_(0.0), nframes_(0), mismatch_psnr_(0.0),
diff --git a/third_party/aom/test/ethread_test.cc b/third_party/aom/test/ethread_test.cc
index 5b519f8fe..86eb3228e 100644
--- a/third_party/aom/test/ethread_test.cc
+++ b/third_party/aom/test/ethread_test.cc
@@ -20,8 +20,8 @@
 
 namespace {
 class AVxEncoderThreadTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   AVxEncoderThreadTest()
       : EncoderTest(GET_PARAM(0)), encoder_initialized_(false),
@@ -30,8 +30,9 @@ class AVxEncoderThreadTest
     aom_codec_dec_cfg_t cfg = aom_codec_dec_cfg_t();
     cfg.w = 1280;
     cfg.h = 720;
+    cfg.allow_lowbitdepth = 1;
     decoder_ = codec_->CreateDecoder(cfg, 0);
-#if CONFIG_AV1 && CONFIG_EXT_TILE
+#if CONFIG_AV1
     if (decoder_->IsAV1()) {
       decoder_->Control(AV1_SET_DECODE_TILE_ROW, -1);
       decoder_->Control(AV1_SET_DECODE_TILE_COL, -1);
@@ -69,20 +70,7 @@ class AVxEncoderThreadTest
   virtual void PreEncodeFrameHook(::libaom_test::VideoSource * /*video*/,
                                   ::libaom_test::Encoder *encoder) {
     if (!encoder_initialized_) {
-#if CONFIG_AV1 && CONFIG_EXT_TILE
-      encoder->Control(AV1E_SET_TILE_COLUMNS, 1);
-      if (codec_ == &libaom_test::kAV1) {
-        // TODO(geza): Start using multiple tile rows when the multi-threaded
-        // encoder can handle them
-        encoder->Control(AV1E_SET_TILE_ROWS, 32);
-      } else {
-        encoder->Control(AV1E_SET_TILE_ROWS, 0);
-      }
-#else
-      // Encode 4 tile columns.
-      encoder->Control(AV1E_SET_TILE_COLUMNS, 2);
-      encoder->Control(AV1E_SET_TILE_ROWS, 0);
-#endif  // CONFIG_AV1 && CONFIG_EXT_TILE
+      SetTileSize(encoder);
 #if CONFIG_LOOPFILTERING_ACROSS_TILES
       encoder->Control(AV1E_SET_TILE_LOOPFILTER, 0);
 #endif  // CONFIG_LOOPFILTERING_ACROSS_TILES
@@ -100,6 +88,12 @@ class AVxEncoderThreadTest
     }
   }
 
+  virtual void SetTileSize(libaom_test::Encoder *encoder) {
+    // Encode 4 tile columns.
+    encoder->Control(AV1E_SET_TILE_COLUMNS, 2);
+    encoder->Control(AV1E_SET_TILE_ROWS, 0);
+  }
+
   virtual void FramePktHook(const aom_codec_cx_pkt_t *pkt) {
     size_enc_.push_back(pkt->data.frame.sz);
 
@@ -169,11 +163,21 @@ class AVxEncoderThreadTest
   std::vector<std::string> md5_dec_;
 };
 
-TEST_P(AVxEncoderThreadTest, EncoderResultTest) { DoTest(); }
+TEST_P(AVxEncoderThreadTest, EncoderResultTest) {
+#if CONFIG_AV1 && CONFIG_EXT_TILE
+  cfg_.large_scale_tile = 0;
+#endif  // CONFIG_AV1 && CONFIG_EXT_TILE
+  DoTest();
+}
 
 class AVxEncoderThreadTestLarge : public AVxEncoderThreadTest {};
 
-TEST_P(AVxEncoderThreadTestLarge, EncoderResultTest) { DoTest(); }
+TEST_P(AVxEncoderThreadTestLarge, EncoderResultTest) {
+#if CONFIG_AV1 && CONFIG_EXT_TILE
+  cfg_.large_scale_tile = 0;
+#endif  // CONFIG_AV1 && CONFIG_EXT_TILE
+  DoTest();
+}
 
 // For AV1, only test speed 0 to 3.
 AV1_INSTANTIATE_TEST_CASE(AVxEncoderThreadTest,
@@ -185,4 +189,36 @@ AV1_INSTANTIATE_TEST_CASE(AVxEncoderThreadTestLarge,
                           ::testing::Values(::libaom_test::kTwoPassGood,
                                             ::libaom_test::kOnePassGood),
                           ::testing::Range(0, 2));
+
+#if CONFIG_AV1 && CONFIG_EXT_TILE
+class AVxEncoderThreadLSTest : public AVxEncoderThreadTest {
+  virtual void SetTileSize(libaom_test::Encoder *encoder) {
+    encoder->Control(AV1E_SET_TILE_COLUMNS, 1);
+    // TODO(geza): Start using multiple tile rows when the multi-threaded
+    // encoder can handle them
+    encoder->Control(AV1E_SET_TILE_ROWS, 32);
+  }
+};
+
+TEST_P(AVxEncoderThreadLSTest, EncoderResultTest) {
+  cfg_.large_scale_tile = 1;
+  DoTest();
+}
+
+class AVxEncoderThreadLSTestLarge : public AVxEncoderThreadLSTest {};
+
+TEST_P(AVxEncoderThreadLSTestLarge, EncoderResultTest) {
+  cfg_.large_scale_tile = 1;
+  DoTest();
+}
+
+AV1_INSTANTIATE_TEST_CASE(AVxEncoderThreadLSTest,
+                          ::testing::Values(::libaom_test::kTwoPassGood,
+                                            ::libaom_test::kOnePassGood),
+                          ::testing::Range(2, 4));
+AV1_INSTANTIATE_TEST_CASE(AVxEncoderThreadLSTestLarge,
+                          ::testing::Values(::libaom_test::kTwoPassGood,
+                                            ::libaom_test::kOnePassGood),
+                          ::testing::Range(0, 2));
+#endif  // CONFIG_AV1 && CONFIG_EXT_TILE
 }  // namespace
diff --git a/third_party/aom/test/fdct4x4_test.cc b/third_party/aom/test/fdct4x4_test.cc
index ed265e84f..e1bd61254 100644
--- a/third_party/aom/test/fdct4x4_test.cc
+++ b/third_party/aom/test/fdct4x4_test.cc
@@ -33,7 +33,7 @@ namespace {
 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 (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 using libaom_test::FhtFunc;
 
 typedef std::tr1::tuple<FdctFunc, IdctFunc, int, aom_bit_depth_t, int>
@@ -41,34 +41,41 @@ typedef std::tr1::tuple<FdctFunc, IdctFunc, int, aom_bit_depth_t, int>
 typedef std::tr1::tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t, int> Ht4x4Param;
 
 void fdct4x4_ref(const int16_t *in, tran_low_t *out, int stride,
-                 int /*tx_type*/) {
+                 TxfmParam * /*txfm_param*/) {
   aom_fdct4x4_c(in, out, stride);
 }
 
-void fht4x4_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht4x4_c(in, out, stride, tx_type);
+void fht4x4_ref(const int16_t *in, tran_low_t *out, int stride,
+                TxfmParam *txfm_param) {
+  av1_fht4x4_c(in, out, stride, txfm_param);
 }
 
 void fwht4x4_ref(const int16_t *in, tran_low_t *out, int stride,
-                 int /*tx_type*/) {
+                 TxfmParam * /*txfm_param*/) {
   av1_fwht4x4_c(in, out, stride);
 }
 
 #if CONFIG_HIGHBITDEPTH
-void idct4x4_10(const tran_low_t *in, uint8_t *out, int stride) {
-  aom_highbd_idct4x4_16_add_c(in, out, stride, 10);
+void fht4x4_10(const int16_t *in, tran_low_t *out, int stride,
+               TxfmParam *txfm_param) {
+  av1_fwd_txfm2d_4x4_c(in, out, stride, txfm_param->tx_type, 10);
 }
 
-void idct4x4_12(const tran_low_t *in, uint8_t *out, int stride) {
-  aom_highbd_idct4x4_16_add_c(in, out, stride, 12);
+void fht4x4_12(const int16_t *in, tran_low_t *out, int stride,
+               TxfmParam *txfm_param) {
+  av1_fwd_txfm2d_4x4_c(in, out, stride, txfm_param->tx_type, 12);
 }
 
-void iht4x4_10(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_highbd_iht4x4_16_add_c(in, out, stride, tx_type, 10);
+void iht4x4_10(const tran_low_t *in, uint8_t *out, int stride,
+               const TxfmParam *txfm_param) {
+  av1_inv_txfm2d_add_4x4_c(in, CONVERT_TO_SHORTPTR(out), stride,
+                           txfm_param->tx_type, 10);
 }
 
-void iht4x4_12(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_highbd_iht4x4_16_add_c(in, out, stride, tx_type, 12);
+void iht4x4_12(const tran_low_t *in, uint8_t *out, int stride,
+               const TxfmParam *txfm_param) {
+  av1_inv_txfm2d_add_4x4_c(in, CONVERT_TO_SHORTPTR(out), stride,
+                           txfm_param->tx_type, 12);
 }
 
 void iwht4x4_10(const tran_low_t *in, uint8_t *out, int stride) {
@@ -78,16 +85,6 @@ void iwht4x4_10(const tran_low_t *in, uint8_t *out, int stride) {
 void iwht4x4_12(const tran_low_t *in, uint8_t *out, int stride) {
   aom_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) {
-  aom_highbd_idct4x4_16_add_sse2(in, out, stride, 10);
-}
-
-void idct4x4_12_sse2(const tran_low_t *in, uint8_t *out, int stride) {
-  aom_highbd_idct4x4_16_add_sse2(in, out, stride, 12);
-}
-#endif  // HAVE_SSE2
 #endif  // CONFIG_HIGHBITDEPTH
 
 class Trans4x4DCT : public libaom_test::TransformTestBase,
@@ -98,13 +95,13 @@ class Trans4x4DCT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 4;
     height_ = 4;
     fwd_txfm_ref = fdct4x4_ref;
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
@@ -136,23 +133,30 @@ class Trans4x4HT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 4;
     height_ = 4;
     fwd_txfm_ref = fht4x4_ref;
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
     num_coeffs_ = GET_PARAM(4);
+    txfm_param_.tx_type = GET_PARAM(2);
+#if CONFIG_HIGHBITDEPTH
+    switch (bit_depth_) {
+      case AOM_BITS_10: fwd_txfm_ref = fht4x4_10; break;
+      case AOM_BITS_12: fwd_txfm_ref = fht4x4_12; break;
+      default: fwd_txfm_ref = fht4x4_ref; break;
+    }
+#endif
   }
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(const int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
 
   void RunInvTxfm(const tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
@@ -175,7 +179,6 @@ class Trans4x4WHT : public libaom_test::TransformTestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 4;
     height_ = 4;
     fwd_txfm_ref = fwht4x4_ref;
@@ -206,32 +209,26 @@ TEST_P(Trans4x4WHT, MemCheck) { RunMemCheck(); }
 TEST_P(Trans4x4WHT, InvAccuracyCheck) { RunInvAccuracyCheck(0); }
 using std::tr1::make_tuple;
 
-#if CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(
-    C, Trans4x4DCT,
-    ::testing::Values(
-        make_tuple(&aom_highbd_fdct4x4_c, &idct4x4_10, 0, AOM_BITS_10, 16),
-        make_tuple(&aom_highbd_fdct4x4_c, &idct4x4_12, 0, AOM_BITS_12, 16),
-        make_tuple(&aom_fdct4x4_c, &aom_idct4x4_16_add_c, 0, AOM_BITS_8, 16)));
-#else
 INSTANTIATE_TEST_CASE_P(C, Trans4x4DCT,
                         ::testing::Values(make_tuple(&aom_fdct4x4_c,
                                                      &aom_idct4x4_16_add_c, 0,
                                                      AOM_BITS_8, 16)));
-#endif  // CONFIG_HIGHBITDEPTH
 
 #if CONFIG_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(
+    DISABLED_C, Trans4x4HT,
+    ::testing::Values(make_tuple(&fht4x4_12, &iht4x4_12, 0, AOM_BITS_12, 16),
+                      make_tuple(&fht4x4_12, &iht4x4_12, 1, AOM_BITS_12, 16),
+                      make_tuple(&fht4x4_12, &iht4x4_12, 2, AOM_BITS_12, 16),
+                      make_tuple(&fht4x4_12, &iht4x4_12, 3, AOM_BITS_12, 16)));
+
+INSTANTIATE_TEST_CASE_P(
     C, Trans4x4HT,
     ::testing::Values(
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_10, 0, AOM_BITS_10, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_10, 1, AOM_BITS_10, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_10, 2, AOM_BITS_10, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_10, 3, AOM_BITS_10, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_12, 0, AOM_BITS_12, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_12, 1, AOM_BITS_12, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_12, 2, AOM_BITS_12, 16),
-        make_tuple(&av1_highbd_fht4x4_c, &iht4x4_12, 3, AOM_BITS_12, 16),
+        make_tuple(&fht4x4_10, &iht4x4_10, 0, AOM_BITS_10, 16),
+        make_tuple(&fht4x4_10, &iht4x4_10, 1, AOM_BITS_10, 16),
+        make_tuple(&fht4x4_10, &iht4x4_10, 2, AOM_BITS_10, 16),
+        make_tuple(&fht4x4_10, &iht4x4_10, 3, AOM_BITS_10, 16),
         make_tuple(&av1_fht4x4_c, &av1_iht4x4_16_add_c, 0, AOM_BITS_8, 16),
         make_tuple(&av1_fht4x4_c, &av1_iht4x4_16_add_c, 1, AOM_BITS_8, 16),
         make_tuple(&av1_fht4x4_c, &av1_iht4x4_16_add_c, 2, AOM_BITS_8, 16),
@@ -305,18 +302,6 @@ INSTANTIATE_TEST_CASE_P(
 
 #if HAVE_SSE2 && CONFIG_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(
-    SSE2, Trans4x4DCT,
-    ::testing::Values(
-        make_tuple(&aom_highbd_fdct4x4_c, &idct4x4_10_sse2, 0, AOM_BITS_10, 16),
-        make_tuple(&aom_highbd_fdct4x4_sse2, &idct4x4_10_sse2, 0, AOM_BITS_10,
-                   16),
-        make_tuple(&aom_highbd_fdct4x4_c, &idct4x4_12_sse2, 0, AOM_BITS_12, 16),
-        make_tuple(&aom_highbd_fdct4x4_sse2, &idct4x4_12_sse2, 0, AOM_BITS_12,
-                   16),
-        make_tuple(&aom_fdct4x4_sse2, &aom_idct4x4_16_add_c, 0, AOM_BITS_8,
-                   16)));
-
-INSTANTIATE_TEST_CASE_P(
     SSE2, Trans4x4HT,
     ::testing::Values(
         make_tuple(&av1_fht4x4_sse2, &av1_iht4x4_16_add_c, 0, AOM_BITS_8, 16),
diff --git a/third_party/aom/test/fdct8x8_test.cc b/third_party/aom/test/fdct8x8_test.cc
index 0e86c70aa..62cdf6229 100644
--- a/third_party/aom/test/fdct8x8_test.cc
+++ b/third_party/aom/test/fdct8x8_test.cc
@@ -7,7 +7,7 @@
  * obtain it at www.aomedia.org/license/software. If the Alliance for Open
  * Media Patent License 1.0 was not distributed with this source code in the
  * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
-*/
+ */
 
 #include <math.h>
 #include <stdlib.h>
@@ -40,9 +40,9 @@ 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);
+                        TxfmParam *txfm_param);
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 
 typedef std::tr1::tuple<FdctFunc, IdctFunc, int, aom_bit_depth_t> Dct8x8Param;
 typedef std::tr1::tuple<FhtFunc, IhtFunc, int, aom_bit_depth_t> Ht8x8Param;
@@ -78,21 +78,36 @@ void reference_8x8_dct_2d(const int16_t input[kNumCoeffs],
 }
 
 void fdct8x8_ref(const int16_t *in, tran_low_t *out, int stride,
-                 int /*tx_type*/) {
+                 TxfmParam * /*txfm_param*/) {
   aom_fdct8x8_c(in, out, stride);
 }
 
-void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
-  av1_fht8x8_c(in, out, stride, tx_type);
+void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride,
+                TxfmParam *txfm_param) {
+  av1_fht8x8_c(in, out, stride, txfm_param);
 }
 
 #if CONFIG_HIGHBITDEPTH
-void iht8x8_10(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_highbd_iht8x8_64_add_c(in, out, stride, tx_type, 10);
+void fht8x8_10(const int16_t *in, tran_low_t *out, int stride,
+               TxfmParam *txfm_param) {
+  av1_fwd_txfm2d_8x8_c(in, out, stride, txfm_param->tx_type, 10);
 }
 
-void iht8x8_12(const tran_low_t *in, uint8_t *out, int stride, int tx_type) {
-  av1_highbd_iht8x8_64_add_c(in, out, stride, tx_type, 12);
+void fht8x8_12(const int16_t *in, tran_low_t *out, int stride,
+               TxfmParam *txfm_param) {
+  av1_fwd_txfm2d_8x8_c(in, out, stride, txfm_param->tx_type, 12);
+}
+
+void iht8x8_10(const tran_low_t *in, uint8_t *out, int stride,
+               const TxfmParam *txfm_param) {
+  av1_inv_txfm2d_add_8x8_c(in, CONVERT_TO_SHORTPTR(out), stride,
+                           txfm_param->tx_type, 10);
+}
+
+void iht8x8_12(const tran_low_t *in, uint8_t *out, int stride,
+               const TxfmParam *txfm_param) {
+  av1_inv_txfm2d_add_8x8_c(in, CONVERT_TO_SHORTPTR(out), stride,
+                           txfm_param->tx_type, 12);
 }
 
 #endif  // CONFIG_HIGHBITDEPTH
@@ -296,7 +311,7 @@ class FwdTrans8x8TestBase {
       ASM_REGISTER_STATE_CHECK(
           RunFwdTxfm(test_input_block, test_temp_block, pitch_));
       ASM_REGISTER_STATE_CHECK(
-          fwd_txfm_ref(test_input_block, ref_temp_block, pitch_, tx_type_));
+          fwd_txfm_ref(test_input_block, ref_temp_block, pitch_, &txfm_param_));
       if (bit_depth_ == AOM_BITS_8) {
         ASM_REGISTER_STATE_CHECK(RunInvTxfm(test_temp_block, dst, pitch_));
 #if CONFIG_HIGHBITDEPTH
@@ -476,10 +491,10 @@ class FwdTrans8x8TestBase {
     }
   }
   int pitch_;
-  int tx_type_;
   FhtFunc fwd_txfm_ref;
   aom_bit_depth_t bit_depth_;
   int mask_;
+  TxfmParam txfm_param_;
 };
 
 class FwdTrans8x8DCT : public FwdTrans8x8TestBase,
@@ -490,11 +505,11 @@ class FwdTrans8x8DCT : public FwdTrans8x8TestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 8;
     fwd_txfm_ref = fdct8x8_ref;
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
+    txfm_param_.tx_type = GET_PARAM(2);
   }
 
   virtual void TearDown() { libaom_test::ClearSystemState(); }
@@ -529,21 +544,28 @@ class FwdTrans8x8HT : public FwdTrans8x8TestBase,
   virtual void SetUp() {
     fwd_txfm_ = GET_PARAM(0);
     inv_txfm_ = GET_PARAM(1);
-    tx_type_ = GET_PARAM(2);
     pitch_ = 8;
     fwd_txfm_ref = fht8x8_ref;
     bit_depth_ = GET_PARAM(3);
     mask_ = (1 << bit_depth_) - 1;
+    txfm_param_.tx_type = GET_PARAM(2);
+#if CONFIG_HIGHBITDEPTH
+    switch (bit_depth_) {
+      case AOM_BITS_10: fwd_txfm_ref = fht8x8_10; break;
+      case AOM_BITS_12: fwd_txfm_ref = fht8x8_12; break;
+      default: fwd_txfm_ref = fht8x8_ref; break;
+    }
+#endif
   }
 
   virtual void TearDown() { libaom_test::ClearSystemState(); }
 
  protected:
   void RunFwdTxfm(int16_t *in, tran_low_t *out, int stride) {
-    fwd_txfm_(in, out, stride, tx_type_);
+    fwd_txfm_(in, out, stride, &txfm_param_);
   }
   void RunInvTxfm(tran_low_t *out, uint8_t *dst, int stride) {
-    inv_txfm_(out, dst, stride, tx_type_);
+    inv_txfm_(out, dst, stride, &txfm_param_);
   }
 
   FhtFunc fwd_txfm_;
@@ -606,14 +628,14 @@ INSTANTIATE_TEST_CASE_P(
     C, FwdTrans8x8HT,
     ::testing::Values(
         make_tuple(&av1_fht8x8_c, &av1_iht8x8_64_add_c, 0, AOM_BITS_8),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_10, 0, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_10, 1, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_10, 2, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_10, 3, AOM_BITS_10),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_12, 0, AOM_BITS_12),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_12, 1, AOM_BITS_12),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_12, 2, AOM_BITS_12),
-        make_tuple(&av1_highbd_fht8x8_c, &iht8x8_12, 3, AOM_BITS_12),
+        make_tuple(&fht8x8_10, &iht8x8_10, 0, AOM_BITS_10),
+        make_tuple(&fht8x8_10, &iht8x8_10, 1, AOM_BITS_10),
+        make_tuple(&fht8x8_10, &iht8x8_10, 2, AOM_BITS_10),
+        make_tuple(&fht8x8_10, &iht8x8_10, 3, AOM_BITS_10),
+        make_tuple(&fht8x8_12, &iht8x8_12, 0, AOM_BITS_12),
+        make_tuple(&fht8x8_12, &iht8x8_12, 1, AOM_BITS_12),
+        make_tuple(&fht8x8_12, &iht8x8_12, 2, AOM_BITS_12),
+        make_tuple(&fht8x8_12, &iht8x8_12, 3, AOM_BITS_12),
         make_tuple(&av1_fht8x8_c, &av1_iht8x8_64_add_c, 1, AOM_BITS_8),
         make_tuple(&av1_fht8x8_c, &av1_iht8x8_64_add_c, 2, AOM_BITS_8),
         make_tuple(&av1_fht8x8_c, &av1_iht8x8_64_add_c, 3, AOM_BITS_8)));
diff --git a/third_party/aom/test/frame_size_tests.cc b/third_party/aom/test/frame_size_tests.cc
index d2e762ff9..442f2523d 100644
--- a/third_party/aom/test/frame_size_tests.cc
+++ b/third_party/aom/test/frame_size_tests.cc
@@ -15,8 +15,8 @@
 
 namespace {
 
-class AV1FrameSizeTests : public ::libaom_test::EncoderTest,
-                          public ::testing::Test {
+class AV1FrameSizeTests : public ::testing::Test,
+                          public ::libaom_test::EncoderTest {
  protected:
   AV1FrameSizeTests()
       : EncoderTest(&::libaom_test::kAV1), expected_res_(AOM_CODEC_OK) {}
diff --git a/third_party/aom/test/hiprec_convolve_test_util.cc b/third_party/aom/test/hiprec_convolve_test_util.cc
index d53384c5b..f5661ec07 100644
--- a/third_party/aom/test/hiprec_convolve_test_util.cc
+++ b/third_party/aom/test/hiprec_convolve_test_util.cc
@@ -92,8 +92,8 @@ void AV1HiprecConvolveTest::RunCheckOutput(hiprec_convolve_func test_impl) {
 
   for (i = 0; i < num_iters; ++i) {
     // Choose random locations within the source block
-    int offset_r = 3 + rnd_.PseudoUniform(w - out_w - 7);
-    int offset_c = 3 + rnd_.PseudoUniform(h - out_h - 7);
+    int offset_r = 3 + rnd_.PseudoUniform(h - out_h - 7);
+    int offset_c = 3 + rnd_.PseudoUniform(w - out_w - 7);
     aom_convolve8_add_src_hip_c(input + offset_r * w + offset_c, w, output,
                                 out_w, hkernel, 16, vkernel, 16, out_w, out_h);
     test_impl(input + offset_r * w + offset_c, w, output2, out_w, hkernel, 16,
@@ -166,8 +166,8 @@ void AV1HighbdHiprecConvolveTest::RunCheckOutput(
 
   for (i = 0; i < num_iters; ++i) {
     // Choose random locations within the source block
-    int offset_r = 3 + rnd_.PseudoUniform(w - out_w - 7);
-    int offset_c = 3 + rnd_.PseudoUniform(h - out_h - 7);
+    int offset_r = 3 + rnd_.PseudoUniform(h - out_h - 7);
+    int offset_c = 3 + rnd_.PseudoUniform(w - out_w - 7);
     aom_highbd_convolve8_add_src_hip_c(input_ptr + offset_r * w + offset_c, w,
                                        output_ptr, out_w, hkernel, 16, vkernel,
                                        16, out_w, out_h, bd);
diff --git a/third_party/aom/test/idct_test.cc b/third_party/aom/test/idct_test.cc
deleted file mode 100644
index a880a9182..000000000
--- a/third_party/aom/test/idct_test.cc
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * Copyright (c) 2016, Alliance for Open Media. All rights reserved
- *
- * This source code is subject to the terms of the BSD 2 Clause License and
- * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
- * was not distributed with this source code in the LICENSE file, you can
- * obtain it at www.aomedia.org/license/software. If the Alliance for Open
- * Media Patent License 1.0 was not distributed with this source code in the
- * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
-*/
-
-#include "./aom_config.h"
-#include "./aom_rtcd.h"
-
-#include "third_party/googletest/src/googletest/include/gtest/gtest.h"
-
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
-#include "aom/aom_integer.h"
-
-typedef void (*IdctFunc)(int16_t *input, unsigned char *pred_ptr,
-                         int pred_stride, unsigned char *dst_ptr,
-                         int dst_stride);
-namespace {
-class IDCTTest : public ::testing::TestWithParam<IdctFunc> {
- protected:
-  virtual void SetUp() {
-    int i;
-
-    UUT = GetParam();
-    memset(input, 0, sizeof(input));
-    /* Set up guard blocks */
-    for (i = 0; i < 256; i++) output[i] = ((i & 0xF) < 4 && (i < 64)) ? 0 : -1;
-  }
-
-  virtual void TearDown() { libaom_test::ClearSystemState(); }
-
-  IdctFunc UUT;
-  int16_t input[16];
-  unsigned char output[256];
-  unsigned char predict[256];
-};
-
-TEST_P(IDCTTest, TestGuardBlocks) {
-  int i;
-
-  for (i = 0; i < 256; i++)
-    if ((i & 0xF) < 4 && i < 64)
-      EXPECT_EQ(0, output[i]) << i;
-    else
-      EXPECT_EQ(255, output[i]);
-}
-
-TEST_P(IDCTTest, TestAllZeros) {
-  int i;
-
-  ASM_REGISTER_STATE_CHECK(UUT(input, output, 16, output, 16));
-
-  for (i = 0; i < 256; i++)
-    if ((i & 0xF) < 4 && i < 64)
-      EXPECT_EQ(0, output[i]) << "i==" << i;
-    else
-      EXPECT_EQ(255, output[i]) << "i==" << i;
-}
-
-TEST_P(IDCTTest, TestAllOnes) {
-  int i;
-
-  input[0] = 4;
-  ASM_REGISTER_STATE_CHECK(UUT(input, output, 16, output, 16));
-
-  for (i = 0; i < 256; i++)
-    if ((i & 0xF) < 4 && i < 64)
-      EXPECT_EQ(1, output[i]) << "i==" << i;
-    else
-      EXPECT_EQ(255, output[i]) << "i==" << i;
-}
-
-TEST_P(IDCTTest, TestAddOne) {
-  int i;
-
-  for (i = 0; i < 256; i++) predict[i] = i;
-  input[0] = 4;
-  ASM_REGISTER_STATE_CHECK(UUT(input, predict, 16, output, 16));
-
-  for (i = 0; i < 256; i++)
-    if ((i & 0xF) < 4 && i < 64)
-      EXPECT_EQ(i + 1, output[i]) << "i==" << i;
-    else
-      EXPECT_EQ(255, output[i]) << "i==" << i;
-}
-
-TEST_P(IDCTTest, TestWithData) {
-  int i;
-
-  for (i = 0; i < 16; i++) input[i] = i;
-
-  ASM_REGISTER_STATE_CHECK(UUT(input, output, 16, output, 16));
-
-  for (i = 0; i < 256; i++)
-    if ((i & 0xF) > 3 || i > 63)
-      EXPECT_EQ(255, output[i]) << "i==" << i;
-    else if (i == 0)
-      EXPECT_EQ(11, output[i]) << "i==" << i;
-    else if (i == 34)
-      EXPECT_EQ(1, output[i]) << "i==" << i;
-    else if (i == 2 || i == 17 || i == 32)
-      EXPECT_EQ(3, output[i]) << "i==" << i;
-    else
-      EXPECT_EQ(0, output[i]) << "i==" << i;
-}
-
-INSTANTIATE_TEST_CASE_P(C, IDCTTest, ::testing::Values(aom_short_idct4x4llm_c));
-#if HAVE_MMX
-INSTANTIATE_TEST_CASE_P(MMX, IDCTTest,
-                        ::testing::Values(aom_short_idct4x4llm_mmx));
-#endif
-#if HAVE_MSA
-INSTANTIATE_TEST_CASE_P(MSA, IDCTTest,
-                        ::testing::Values(aom_short_idct4x4llm_msa));
-#endif
-}
diff --git a/third_party/aom/test/level_test.cc b/third_party/aom/test/level_test.cc
index 1049d4901..12f391817 100644
--- a/third_party/aom/test/level_test.cc
+++ b/third_party/aom/test/level_test.cc
@@ -17,8 +17,8 @@
 
 namespace {
 class LevelTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   LevelTest()
       : EncoderTest(GET_PARAM(0)), encoding_mode_(GET_PARAM(1)),
diff --git a/third_party/aom/test/lossless_test.cc b/third_party/aom/test/lossless_test.cc
index 5c5b32d93..f4978fe21 100644
--- a/third_party/aom/test/lossless_test.cc
+++ b/third_party/aom/test/lossless_test.cc
@@ -23,8 +23,8 @@ namespace {
 const int kMaxPsnr = 100;
 
 class LosslessTestLarge
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   LosslessTestLarge()
       : EncoderTest(GET_PARAM(0)), psnr_(kMaxPsnr), nframes_(0),
diff --git a/third_party/aom/test/lpf_8_test.cc b/third_party/aom/test/lpf_8_test.cc
index cee0d3b81..f050718bb 100644
--- a/third_party/aom/test/lpf_8_test.cc
+++ b/third_party/aom/test/lpf_8_test.cc
@@ -410,77 +410,75 @@ TEST_P(Loop8Test9Param, ValueCheck) {
 
 using std::tr1::make_tuple;
 
-#if HAVE_SSE2
+#if HAVE_SSE2 && (!CONFIG_PARALLEL_DEBLOCKING)
 #if CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(
-    SSE2, Loop8Test6Param,
-    ::testing::Values(make_tuple(&aom_highbd_lpf_horizontal_4_sse2,
-                                 &aom_highbd_lpf_horizontal_4_c, 8),
-                      make_tuple(&aom_highbd_lpf_vertical_4_sse2,
-                                 &aom_highbd_lpf_vertical_4_c, 8),
-                      make_tuple(&aom_highbd_lpf_horizontal_8_sse2,
-                                 &aom_highbd_lpf_horizontal_8_c, 8),
-                      make_tuple(&aom_highbd_lpf_horizontal_edge_8_sse2,
-                                 &aom_highbd_lpf_horizontal_edge_8_c, 8),
-                      make_tuple(&aom_highbd_lpf_horizontal_edge_16_sse2,
-                                 &aom_highbd_lpf_horizontal_edge_16_c, 8),
-                      make_tuple(&aom_highbd_lpf_vertical_8_sse2,
-                                 &aom_highbd_lpf_vertical_8_c, 8),
-                      make_tuple(&aom_highbd_lpf_vertical_16_sse2,
-                                 &aom_highbd_lpf_vertical_16_c, 8),
-                      make_tuple(&aom_highbd_lpf_horizontal_4_sse2,
-                                 &aom_highbd_lpf_horizontal_4_c, 10),
-                      make_tuple(&aom_highbd_lpf_vertical_4_sse2,
-                                 &aom_highbd_lpf_vertical_4_c, 10),
-                      make_tuple(&aom_highbd_lpf_horizontal_8_sse2,
-                                 &aom_highbd_lpf_horizontal_8_c, 10),
-                      make_tuple(&aom_highbd_lpf_horizontal_edge_8_sse2,
-                                 &aom_highbd_lpf_horizontal_edge_8_c, 10),
-                      make_tuple(&aom_highbd_lpf_horizontal_edge_16_sse2,
-                                 &aom_highbd_lpf_horizontal_edge_16_c, 10),
-                      make_tuple(&aom_highbd_lpf_vertical_8_sse2,
-                                 &aom_highbd_lpf_vertical_8_c, 10),
-                      make_tuple(&aom_highbd_lpf_vertical_16_sse2,
-                                 &aom_highbd_lpf_vertical_16_c, 10),
-                      make_tuple(&aom_highbd_lpf_horizontal_4_sse2,
-                                 &aom_highbd_lpf_horizontal_4_c, 12),
-                      make_tuple(&aom_highbd_lpf_vertical_4_sse2,
-                                 &aom_highbd_lpf_vertical_4_c, 12),
-                      make_tuple(&aom_highbd_lpf_horizontal_8_sse2,
-                                 &aom_highbd_lpf_horizontal_8_c, 12),
-                      make_tuple(&aom_highbd_lpf_horizontal_edge_8_sse2,
-                                 &aom_highbd_lpf_horizontal_edge_8_c, 12),
-                      make_tuple(&aom_highbd_lpf_horizontal_edge_16_sse2,
-                                 &aom_highbd_lpf_horizontal_edge_16_c, 12),
-                      make_tuple(&aom_highbd_lpf_vertical_8_sse2,
-                                 &aom_highbd_lpf_vertical_8_c, 12),
-                      make_tuple(&aom_highbd_lpf_vertical_16_sse2,
-                                 &aom_highbd_lpf_vertical_16_c, 12),
-                      make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2,
-                                 &aom_highbd_lpf_vertical_16_dual_c, 8),
-                      make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2,
-                                 &aom_highbd_lpf_vertical_16_dual_c, 10),
-                      make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2,
-                                 &aom_highbd_lpf_vertical_16_dual_c, 12)));
+
+const loop8_param_t kHbdLoop8Test6[] = {
+  make_tuple(&aom_highbd_lpf_horizontal_4_sse2, &aom_highbd_lpf_horizontal_4_c,
+             8),
+  make_tuple(&aom_highbd_lpf_vertical_4_sse2, &aom_highbd_lpf_vertical_4_c, 8),
+  make_tuple(&aom_highbd_lpf_horizontal_8_sse2, &aom_highbd_lpf_horizontal_8_c,
+             8),
+  make_tuple(&aom_highbd_lpf_horizontal_edge_8_sse2,
+             &aom_highbd_lpf_horizontal_edge_8_c, 8),
+  make_tuple(&aom_highbd_lpf_horizontal_edge_16_sse2,
+             &aom_highbd_lpf_horizontal_edge_16_c, 8),
+  make_tuple(&aom_highbd_lpf_vertical_8_sse2, &aom_highbd_lpf_vertical_8_c, 8),
+  make_tuple(&aom_highbd_lpf_vertical_16_sse2, &aom_highbd_lpf_vertical_16_c,
+             8),
+  make_tuple(&aom_highbd_lpf_horizontal_4_sse2, &aom_highbd_lpf_horizontal_4_c,
+             10),
+  make_tuple(&aom_highbd_lpf_vertical_4_sse2, &aom_highbd_lpf_vertical_4_c, 10),
+  make_tuple(&aom_highbd_lpf_horizontal_8_sse2, &aom_highbd_lpf_horizontal_8_c,
+             10),
+  make_tuple(&aom_highbd_lpf_horizontal_edge_8_sse2,
+             &aom_highbd_lpf_horizontal_edge_8_c, 10),
+  make_tuple(&aom_highbd_lpf_horizontal_edge_16_sse2,
+             &aom_highbd_lpf_horizontal_edge_16_c, 10),
+  make_tuple(&aom_highbd_lpf_vertical_8_sse2, &aom_highbd_lpf_vertical_8_c, 10),
+  make_tuple(&aom_highbd_lpf_vertical_16_sse2, &aom_highbd_lpf_vertical_16_c,
+             10),
+  make_tuple(&aom_highbd_lpf_horizontal_4_sse2, &aom_highbd_lpf_horizontal_4_c,
+             12),
+  make_tuple(&aom_highbd_lpf_vertical_4_sse2, &aom_highbd_lpf_vertical_4_c, 12),
+  make_tuple(&aom_highbd_lpf_horizontal_8_sse2, &aom_highbd_lpf_horizontal_8_c,
+             12),
+  make_tuple(&aom_highbd_lpf_horizontal_edge_8_sse2,
+             &aom_highbd_lpf_horizontal_edge_8_c, 12),
+  make_tuple(&aom_highbd_lpf_horizontal_edge_16_sse2,
+             &aom_highbd_lpf_horizontal_edge_16_c, 12),
+  make_tuple(&aom_highbd_lpf_vertical_8_sse2, &aom_highbd_lpf_vertical_8_c, 12),
+  make_tuple(&aom_highbd_lpf_vertical_16_sse2, &aom_highbd_lpf_vertical_16_c,
+             12),
+  make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2,
+             &aom_highbd_lpf_vertical_16_dual_c, 8),
+  make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2,
+             &aom_highbd_lpf_vertical_16_dual_c, 10),
+  make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2,
+             &aom_highbd_lpf_vertical_16_dual_c, 12)
+};
+
+INSTANTIATE_TEST_CASE_P(SSE2, Loop8Test6Param,
+                        ::testing::ValuesIn(kHbdLoop8Test6));
 #else
-INSTANTIATE_TEST_CASE_P(
-    SSE2, Loop8Test6Param,
-    ::testing::Values(
-        make_tuple(&aom_lpf_horizontal_4_sse2, &aom_lpf_horizontal_4_c, 8),
-        make_tuple(&aom_lpf_horizontal_8_sse2, &aom_lpf_horizontal_8_c, 8),
-        make_tuple(&aom_lpf_horizontal_edge_8_sse2,
-                   &aom_lpf_horizontal_edge_8_c, 8),
-        make_tuple(&aom_lpf_horizontal_edge_16_sse2,
-                   &aom_lpf_horizontal_edge_16_c, 8),
-        make_tuple(&aom_lpf_vertical_4_sse2, &aom_lpf_vertical_4_c, 8),
-        make_tuple(&aom_lpf_vertical_8_sse2, &aom_lpf_vertical_8_c, 8),
-        make_tuple(&aom_lpf_vertical_16_sse2, &aom_lpf_vertical_16_c, 8),
-        make_tuple(&aom_lpf_vertical_16_dual_sse2, &aom_lpf_vertical_16_dual_c,
-                   8)));
+const loop8_param_t kLoop8Test6[] = {
+  make_tuple(&aom_lpf_horizontal_4_sse2, &aom_lpf_horizontal_4_c, 8),
+  make_tuple(&aom_lpf_horizontal_8_sse2, &aom_lpf_horizontal_8_c, 8),
+  make_tuple(&aom_lpf_horizontal_edge_8_sse2, &aom_lpf_horizontal_edge_8_c, 8),
+  make_tuple(&aom_lpf_horizontal_edge_16_sse2, &aom_lpf_horizontal_edge_16_c,
+             8),
+  make_tuple(&aom_lpf_vertical_4_sse2, &aom_lpf_vertical_4_c, 8),
+  make_tuple(&aom_lpf_vertical_8_sse2, &aom_lpf_vertical_8_c, 8),
+  make_tuple(&aom_lpf_vertical_16_sse2, &aom_lpf_vertical_16_c, 8),
+  make_tuple(&aom_lpf_vertical_16_dual_sse2, &aom_lpf_vertical_16_dual_c, 8)
+};
+
+INSTANTIATE_TEST_CASE_P(SSE2, Loop8Test6Param,
+                        ::testing::ValuesIn(kLoop8Test6));
 #endif  // CONFIG_HIGHBITDEPTH
 #endif
 
-#if HAVE_AVX2 && (!CONFIG_HIGHBITDEPTH)
+#if HAVE_AVX2 && (!CONFIG_HIGHBITDEPTH) && (!CONFIG_PARALLEL_DEBLOCKING)
 INSTANTIATE_TEST_CASE_P(
     AVX2, Loop8Test6Param,
     ::testing::Values(make_tuple(&aom_lpf_horizontal_edge_8_avx2,
@@ -489,49 +487,51 @@ INSTANTIATE_TEST_CASE_P(
                                  &aom_lpf_horizontal_edge_16_c, 8)));
 #endif
 
-#if HAVE_SSE2
+#if HAVE_SSE2 && (!CONFIG_PARALLEL_DEBLOCKING)
 #if CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(
-    SSE2, Loop8Test9Param,
-    ::testing::Values(make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,
-                                 &aom_highbd_lpf_horizontal_4_dual_c, 8),
-                      make_tuple(&aom_highbd_lpf_horizontal_8_dual_sse2,
-                                 &aom_highbd_lpf_horizontal_8_dual_c, 8),
-                      make_tuple(&aom_highbd_lpf_vertical_4_dual_sse2,
-                                 &aom_highbd_lpf_vertical_4_dual_c, 8),
-                      make_tuple(&aom_highbd_lpf_vertical_8_dual_sse2,
-                                 &aom_highbd_lpf_vertical_8_dual_c, 8),
-                      make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,
-                                 &aom_highbd_lpf_horizontal_4_dual_c, 10),
-                      make_tuple(&aom_highbd_lpf_horizontal_8_dual_sse2,
-                                 &aom_highbd_lpf_horizontal_8_dual_c, 10),
-                      make_tuple(&aom_highbd_lpf_vertical_4_dual_sse2,
-                                 &aom_highbd_lpf_vertical_4_dual_c, 10),
-                      make_tuple(&aom_highbd_lpf_vertical_8_dual_sse2,
-                                 &aom_highbd_lpf_vertical_8_dual_c, 10),
-                      make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,
-                                 &aom_highbd_lpf_horizontal_4_dual_c, 12),
-                      make_tuple(&aom_highbd_lpf_horizontal_8_dual_sse2,
-                                 &aom_highbd_lpf_horizontal_8_dual_c, 12),
-                      make_tuple(&aom_highbd_lpf_vertical_4_dual_sse2,
-                                 &aom_highbd_lpf_vertical_4_dual_c, 12),
-                      make_tuple(&aom_highbd_lpf_vertical_8_dual_sse2,
-                                 &aom_highbd_lpf_vertical_8_dual_c, 12)));
+const dualloop8_param_t kHbdLoop8Test9[] = {
+  make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,
+             &aom_highbd_lpf_horizontal_4_dual_c, 8),
+  make_tuple(&aom_highbd_lpf_horizontal_8_dual_sse2,
+             &aom_highbd_lpf_horizontal_8_dual_c, 8),
+  make_tuple(&aom_highbd_lpf_vertical_4_dual_sse2,
+             &aom_highbd_lpf_vertical_4_dual_c, 8),
+  make_tuple(&aom_highbd_lpf_vertical_8_dual_sse2,
+             &aom_highbd_lpf_vertical_8_dual_c, 8),
+  make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,
+             &aom_highbd_lpf_horizontal_4_dual_c, 10),
+  make_tuple(&aom_highbd_lpf_horizontal_8_dual_sse2,
+             &aom_highbd_lpf_horizontal_8_dual_c, 10),
+  make_tuple(&aom_highbd_lpf_vertical_4_dual_sse2,
+             &aom_highbd_lpf_vertical_4_dual_c, 10),
+  make_tuple(&aom_highbd_lpf_vertical_8_dual_sse2,
+             &aom_highbd_lpf_vertical_8_dual_c, 10),
+  make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,
+             &aom_highbd_lpf_horizontal_4_dual_c, 12),
+  make_tuple(&aom_highbd_lpf_horizontal_8_dual_sse2,
+             &aom_highbd_lpf_horizontal_8_dual_c, 12),
+  make_tuple(&aom_highbd_lpf_vertical_4_dual_sse2,
+             &aom_highbd_lpf_vertical_4_dual_c, 12),
+  make_tuple(&aom_highbd_lpf_vertical_8_dual_sse2,
+             &aom_highbd_lpf_vertical_8_dual_c, 12)
+};
+
+INSTANTIATE_TEST_CASE_P(SSE2, Loop8Test9Param,
+                        ::testing::ValuesIn(kHbdLoop8Test9));
 #else
-INSTANTIATE_TEST_CASE_P(
-    SSE2, Loop8Test9Param,
-    ::testing::Values(make_tuple(&aom_lpf_horizontal_4_dual_sse2,
-                                 &aom_lpf_horizontal_4_dual_c, 8),
-                      make_tuple(&aom_lpf_horizontal_8_dual_sse2,
-                                 &aom_lpf_horizontal_8_dual_c, 8),
-                      make_tuple(&aom_lpf_vertical_4_dual_sse2,
-                                 &aom_lpf_vertical_4_dual_c, 8),
-                      make_tuple(&aom_lpf_vertical_8_dual_sse2,
-                                 &aom_lpf_vertical_8_dual_c, 8)));
+const dualloop8_param_t kLoop8Test9[] = {
+  make_tuple(&aom_lpf_horizontal_4_dual_sse2, &aom_lpf_horizontal_4_dual_c, 8),
+  make_tuple(&aom_lpf_horizontal_8_dual_sse2, &aom_lpf_horizontal_8_dual_c, 8),
+  make_tuple(&aom_lpf_vertical_4_dual_sse2, &aom_lpf_vertical_4_dual_c, 8),
+  make_tuple(&aom_lpf_vertical_8_dual_sse2, &aom_lpf_vertical_8_dual_c, 8)
+};
+
+INSTANTIATE_TEST_CASE_P(SSE2, Loop8Test9Param,
+                        ::testing::ValuesIn(kLoop8Test9));
 #endif  // CONFIG_HIGHBITDEPTH
 #endif
 
-#if HAVE_NEON
+#if HAVE_NEON && (!CONFIG_PARALLEL_DEBLOCKING)
 #if CONFIG_HIGHBITDEPTH
 // No neon high bitdepth functions.
 #else
@@ -567,9 +567,9 @@ INSTANTIATE_TEST_CASE_P(NEON, Loop8Test9Param,
                             make_tuple(&aom_lpf_vertical_4_dual_neon,
                                        &aom_lpf_vertical_4_dual_c, 8)));
 #endif  // CONFIG_HIGHBITDEPTH
-#endif  // HAVE_NEON
+#endif  // HAVE_NEON && (!CONFIG_PARALLEL_DEBLOCKING)
 
-#if HAVE_DSPR2 && !CONFIG_HIGHBITDEPTH
+#if HAVE_DSPR2 && !CONFIG_HIGHBITDEPTH && (!CONFIG_PARALLEL_DEBLOCKING)
 INSTANTIATE_TEST_CASE_P(
     DSPR2, Loop8Test6Param,
     ::testing::Values(
@@ -593,9 +593,9 @@ INSTANTIATE_TEST_CASE_P(
                                  &aom_lpf_vertical_4_dual_c, 8),
                       make_tuple(&aom_lpf_vertical_8_dual_dspr2,
                                  &aom_lpf_vertical_8_dual_c, 8)));
-#endif  // HAVE_DSPR2 && !CONFIG_HIGHBITDEPTH
+#endif  // HAVE_DSPR2 && !CONFIG_HIGHBITDEPTH && (!CONFIG_PARALLEL_DEBLOCKING)
 
-#if HAVE_MSA && (!CONFIG_HIGHBITDEPTH)
+#if HAVE_MSA && (!CONFIG_HIGHBITDEPTH) && (!CONFIG_PARALLEL_DEBLOCKING)
 INSTANTIATE_TEST_CASE_P(
     MSA, Loop8Test6Param,
     ::testing::Values(
@@ -619,6 +619,6 @@ INSTANTIATE_TEST_CASE_P(
                                  &aom_lpf_vertical_4_dual_c, 8),
                       make_tuple(&aom_lpf_vertical_8_dual_msa,
                                  &aom_lpf_vertical_8_dual_c, 8)));
-#endif  // HAVE_MSA && (!CONFIG_HIGHBITDEPTH)
+#endif  // HAVE_MSA && (!CONFIG_HIGHBITDEPTH) && (!CONFIG_PARALLEL_DEBLOCKING)
 
 }  // namespace
diff --git a/third_party/aom/test/masked_sad_test.cc b/third_party/aom/test/masked_sad_test.cc
index 2dde3c537..19f97718d 100644
--- a/third_party/aom/test/masked_sad_test.cc
+++ b/third_party/aom/test/masked_sad_test.cc
@@ -160,64 +160,56 @@ TEST_P(HighbdMaskedSADTest, OperationCheck) {
 using std::tr1::make_tuple;
 
 #if HAVE_SSSE3
-INSTANTIATE_TEST_CASE_P(
-    SSSE3_C_COMPARE, MaskedSADTest,
-    ::testing::Values(
+const MaskedSADParam msad_test[] = {
 #if CONFIG_EXT_PARTITION
-        make_tuple(&aom_masked_sad128x128_ssse3, &aom_masked_sad128x128_c),
-        make_tuple(&aom_masked_sad128x64_ssse3, &aom_masked_sad128x64_c),
-        make_tuple(&aom_masked_sad64x128_ssse3, &aom_masked_sad64x128_c),
+  make_tuple(&aom_masked_sad128x128_ssse3, &aom_masked_sad128x128_c),
+  make_tuple(&aom_masked_sad128x64_ssse3, &aom_masked_sad128x64_c),
+  make_tuple(&aom_masked_sad64x128_ssse3, &aom_masked_sad64x128_c),
 #endif  // CONFIG_EXT_PARTITION
-        make_tuple(&aom_masked_sad64x64_ssse3, &aom_masked_sad64x64_c),
-        make_tuple(&aom_masked_sad64x32_ssse3, &aom_masked_sad64x32_c),
-        make_tuple(&aom_masked_sad32x64_ssse3, &aom_masked_sad32x64_c),
-        make_tuple(&aom_masked_sad32x32_ssse3, &aom_masked_sad32x32_c),
-        make_tuple(&aom_masked_sad32x16_ssse3, &aom_masked_sad32x16_c),
-        make_tuple(&aom_masked_sad16x32_ssse3, &aom_masked_sad16x32_c),
-        make_tuple(&aom_masked_sad16x16_ssse3, &aom_masked_sad16x16_c),
-        make_tuple(&aom_masked_sad16x8_ssse3, &aom_masked_sad16x8_c),
-        make_tuple(&aom_masked_sad8x16_ssse3, &aom_masked_sad8x16_c),
-        make_tuple(&aom_masked_sad8x8_ssse3, &aom_masked_sad8x8_c),
-        make_tuple(&aom_masked_sad8x4_ssse3, &aom_masked_sad8x4_c),
-        make_tuple(&aom_masked_sad4x8_ssse3, &aom_masked_sad4x8_c),
-        make_tuple(&aom_masked_sad4x4_ssse3, &aom_masked_sad4x4_c)));
+  make_tuple(&aom_masked_sad64x64_ssse3, &aom_masked_sad64x64_c),
+  make_tuple(&aom_masked_sad64x32_ssse3, &aom_masked_sad64x32_c),
+  make_tuple(&aom_masked_sad32x64_ssse3, &aom_masked_sad32x64_c),
+  make_tuple(&aom_masked_sad32x32_ssse3, &aom_masked_sad32x32_c),
+  make_tuple(&aom_masked_sad32x16_ssse3, &aom_masked_sad32x16_c),
+  make_tuple(&aom_masked_sad16x32_ssse3, &aom_masked_sad16x32_c),
+  make_tuple(&aom_masked_sad16x16_ssse3, &aom_masked_sad16x16_c),
+  make_tuple(&aom_masked_sad16x8_ssse3, &aom_masked_sad16x8_c),
+  make_tuple(&aom_masked_sad8x16_ssse3, &aom_masked_sad8x16_c),
+  make_tuple(&aom_masked_sad8x8_ssse3, &aom_masked_sad8x8_c),
+  make_tuple(&aom_masked_sad8x4_ssse3, &aom_masked_sad8x4_c),
+  make_tuple(&aom_masked_sad4x8_ssse3, &aom_masked_sad4x8_c),
+  make_tuple(&aom_masked_sad4x4_ssse3, &aom_masked_sad4x4_c)
+};
+
+INSTANTIATE_TEST_CASE_P(SSSE3_C_COMPARE, MaskedSADTest,
+                        ::testing::ValuesIn(msad_test));
 #if CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(SSSE3_C_COMPARE, HighbdMaskedSADTest,
-                        ::testing::Values(
+const HighbdMaskedSADParam hbd_msad_test[] = {
 #if CONFIG_EXT_PARTITION
-                            make_tuple(&aom_highbd_masked_sad128x128_ssse3,
-                                       &aom_highbd_masked_sad128x128_c),
-                            make_tuple(&aom_highbd_masked_sad128x64_ssse3,
-                                       &aom_highbd_masked_sad128x64_c),
-                            make_tuple(&aom_highbd_masked_sad64x128_ssse3,
-                                       &aom_highbd_masked_sad64x128_c),
+  make_tuple(&aom_highbd_masked_sad128x128_ssse3,
+             &aom_highbd_masked_sad128x128_c),
+  make_tuple(&aom_highbd_masked_sad128x64_ssse3,
+             &aom_highbd_masked_sad128x64_c),
+  make_tuple(&aom_highbd_masked_sad64x128_ssse3,
+             &aom_highbd_masked_sad64x128_c),
 #endif  // CONFIG_EXT_PARTITION
-                            make_tuple(&aom_highbd_masked_sad64x64_ssse3,
-                                       &aom_highbd_masked_sad64x64_c),
-                            make_tuple(&aom_highbd_masked_sad64x32_ssse3,
-                                       &aom_highbd_masked_sad64x32_c),
-                            make_tuple(&aom_highbd_masked_sad32x64_ssse3,
-                                       &aom_highbd_masked_sad32x64_c),
-                            make_tuple(&aom_highbd_masked_sad32x32_ssse3,
-                                       &aom_highbd_masked_sad32x32_c),
-                            make_tuple(&aom_highbd_masked_sad32x16_ssse3,
-                                       &aom_highbd_masked_sad32x16_c),
-                            make_tuple(&aom_highbd_masked_sad16x32_ssse3,
-                                       &aom_highbd_masked_sad16x32_c),
-                            make_tuple(&aom_highbd_masked_sad16x16_ssse3,
-                                       &aom_highbd_masked_sad16x16_c),
-                            make_tuple(&aom_highbd_masked_sad16x8_ssse3,
-                                       &aom_highbd_masked_sad16x8_c),
-                            make_tuple(&aom_highbd_masked_sad8x16_ssse3,
-                                       &aom_highbd_masked_sad8x16_c),
-                            make_tuple(&aom_highbd_masked_sad8x8_ssse3,
-                                       &aom_highbd_masked_sad8x8_c),
-                            make_tuple(&aom_highbd_masked_sad8x4_ssse3,
-                                       &aom_highbd_masked_sad8x4_c),
-                            make_tuple(&aom_highbd_masked_sad4x8_ssse3,
-                                       &aom_highbd_masked_sad4x8_c),
-                            make_tuple(&aom_highbd_masked_sad4x4_ssse3,
-                                       &aom_highbd_masked_sad4x4_c)));
+  make_tuple(&aom_highbd_masked_sad64x64_ssse3, &aom_highbd_masked_sad64x64_c),
+  make_tuple(&aom_highbd_masked_sad64x32_ssse3, &aom_highbd_masked_sad64x32_c),
+  make_tuple(&aom_highbd_masked_sad32x64_ssse3, &aom_highbd_masked_sad32x64_c),
+  make_tuple(&aom_highbd_masked_sad32x32_ssse3, &aom_highbd_masked_sad32x32_c),
+  make_tuple(&aom_highbd_masked_sad32x16_ssse3, &aom_highbd_masked_sad32x16_c),
+  make_tuple(&aom_highbd_masked_sad16x32_ssse3, &aom_highbd_masked_sad16x32_c),
+  make_tuple(&aom_highbd_masked_sad16x16_ssse3, &aom_highbd_masked_sad16x16_c),
+  make_tuple(&aom_highbd_masked_sad16x8_ssse3, &aom_highbd_masked_sad16x8_c),
+  make_tuple(&aom_highbd_masked_sad8x16_ssse3, &aom_highbd_masked_sad8x16_c),
+  make_tuple(&aom_highbd_masked_sad8x8_ssse3, &aom_highbd_masked_sad8x8_c),
+  make_tuple(&aom_highbd_masked_sad8x4_ssse3, &aom_highbd_masked_sad8x4_c),
+  make_tuple(&aom_highbd_masked_sad4x8_ssse3, &aom_highbd_masked_sad4x8_c),
+  make_tuple(&aom_highbd_masked_sad4x4_ssse3, &aom_highbd_masked_sad4x4_c)
+};
+
+INSTANTIATE_TEST_CASE_P(SSSE3_C_COMPARE, HighbdMaskedSADTest,
+                        ::testing::ValuesIn(hbd_msad_test));
 #endif  // CONFIG_HIGHBITDEPTH
 #endif  // HAVE_SSSE3
 }  // namespace
diff --git a/third_party/aom/test/masked_variance_test.cc b/third_party/aom/test/masked_variance_test.cc
index bf113c69e..a9cbdc80d 100644
--- a/third_party/aom/test/masked_variance_test.cc
+++ b/third_party/aom/test/masked_variance_test.cc
@@ -59,30 +59,25 @@ TEST_P(MaskedSubPixelVarianceTest, OperationCheck) {
   unsigned int ref_ret, opt_ret;
   unsigned int ref_sse, opt_sse;
   ACMRandom rnd(ACMRandom::DeterministicSeed());
-  // Note: We pad the input arrays out with 15 extra elements, since the SSE
-  // implementations can read up to 15 elements off the end of the main data.
-  // The extra data is never actually used, but it simplifies the code
-  // if we can do this.
+  // Note: We pad out the input array to a multiple of 16 bytes wide, so that
+  // consecutive rows keep the 16-byte alignment.
+  DECLARE_ALIGNED(16, uint8_t, src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
+  DECLARE_ALIGNED(16, uint8_t, ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
   DECLARE_ALIGNED(16, uint8_t,
-                  src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
+                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
+  DECLARE_ALIGNED(16, uint8_t, msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
   int err_count = 0;
   int first_failure = -1;
-  int src_stride = (MAX_SB_SIZE + 1);
-  int ref_stride = (MAX_SB_SIZE + 1);
-  int msk_stride = (MAX_SB_SIZE + 1);
+  int src_stride = (MAX_SB_SIZE + 16);
+  int ref_stride = (MAX_SB_SIZE + 16);
+  int msk_stride = (MAX_SB_SIZE + 16);
   int xoffset;
   int yoffset;
 
   for (int i = 0; i < number_of_iterations; ++i) {
     int xoffsets[] = { 0, 4, rnd(BIL_SUBPEL_SHIFTS) };
     int yoffsets[] = { 0, 4, rnd(BIL_SUBPEL_SHIFTS) };
-    for (int j = 0; j < (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1); j++) {
+    for (int j = 0; j < (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16); j++) {
       src_ptr[j] = rnd.Rand8();
       ref_ptr[j] = rnd.Rand8();
       second_pred_ptr[j] = rnd.Rand8();
@@ -120,33 +115,30 @@ TEST_P(MaskedSubPixelVarianceTest, ExtremeValues) {
   unsigned int ref_ret, opt_ret;
   unsigned int ref_sse, opt_sse;
   ACMRandom rnd(ACMRandom::DeterministicSeed());
+  DECLARE_ALIGNED(16, uint8_t, src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
+  DECLARE_ALIGNED(16, uint8_t, ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
   DECLARE_ALIGNED(16, uint8_t,
-                  src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 15]);
+                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
+  DECLARE_ALIGNED(16, uint8_t, msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16)]);
   int first_failure_x = -1;
   int first_failure_y = -1;
   int err_count = 0;
   int first_failure = -1;
-  int src_stride = (MAX_SB_SIZE + 1);
-  int ref_stride = (MAX_SB_SIZE + 1);
-  int msk_stride = (MAX_SB_SIZE + 1);
+  int src_stride = (MAX_SB_SIZE + 16);
+  int ref_stride = (MAX_SB_SIZE + 16);
+  int msk_stride = (MAX_SB_SIZE + 16);
 
   for (int xoffset = 0; xoffset < BIL_SUBPEL_SHIFTS; xoffset++) {
     for (int yoffset = 0; yoffset < BIL_SUBPEL_SHIFTS; yoffset++) {
       for (int i = 0; i < 16; ++i) {
         memset(src_ptr, (i & 0x1) ? 255 : 0,
-               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16));
         memset(ref_ptr, (i & 0x2) ? 255 : 0,
-               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16));
         memset(second_pred_ptr, (i & 0x4) ? 255 : 0,
-               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16));
         memset(msk_ptr, (i & 0x8) ? 64 : 0,
-               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 16));
 
         for (int invert_mask = 0; invert_mask < 2; ++invert_mask) {
           ref_ret = ref_func_(src_ptr, src_stride, xoffset, yoffset, ref_ptr,
@@ -204,18 +196,11 @@ TEST_P(HighbdMaskedSubPixelVarianceTest, OperationCheck) {
   unsigned int ref_ret, opt_ret;
   unsigned int ref_sse, opt_sse;
   ACMRandom rnd(ACMRandom::DeterministicSeed());
-  // Note: We pad the input arrays out with 7 extra elements, since the SSE
-  // implementations can read up to 7 elements off the end of the main data.
-  // The extra data is never actually used, but it simplifies the code
-  // if we can do this.
-  DECLARE_ALIGNED(16, uint16_t,
-                  src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
+  DECLARE_ALIGNED(16, uint16_t, src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
+  DECLARE_ALIGNED(16, uint16_t, ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
   DECLARE_ALIGNED(16, uint16_t,
-                  ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
-  DECLARE_ALIGNED(16, uint16_t,
-                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
+                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
+  DECLARE_ALIGNED(16, uint8_t, msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
   uint8_t *src8_ptr = CONVERT_TO_BYTEPTR(src_ptr);
   uint8_t *ref8_ptr = CONVERT_TO_BYTEPTR(ref_ptr);
   uint8_t *second_pred8_ptr = CONVERT_TO_BYTEPTR(second_pred_ptr);
@@ -223,13 +208,13 @@ TEST_P(HighbdMaskedSubPixelVarianceTest, OperationCheck) {
   int first_failure = -1;
   int first_failure_x = -1;
   int first_failure_y = -1;
-  int src_stride = (MAX_SB_SIZE + 1);
-  int ref_stride = (MAX_SB_SIZE + 1);
-  int msk_stride = (MAX_SB_SIZE + 1);
+  int src_stride = (MAX_SB_SIZE + 8);
+  int ref_stride = (MAX_SB_SIZE + 8);
+  int msk_stride = (MAX_SB_SIZE + 8);
   int xoffset, yoffset;
 
   for (int i = 0; i < number_of_iterations; ++i) {
-    for (int j = 0; j < (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1); j++) {
+    for (int j = 0; j < (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8); j++) {
       src_ptr[j] = rnd.Rand16() & ((1 << bit_depth_) - 1);
       ref_ptr[j] = rnd.Rand16() & ((1 << bit_depth_) - 1);
       second_pred_ptr[j] = rnd.Rand16() & ((1 << bit_depth_) - 1);
@@ -270,14 +255,11 @@ TEST_P(HighbdMaskedSubPixelVarianceTest, ExtremeValues) {
   unsigned int ref_ret, opt_ret;
   unsigned int ref_sse, opt_sse;
   ACMRandom rnd(ACMRandom::DeterministicSeed());
+  DECLARE_ALIGNED(16, uint16_t, src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
+  DECLARE_ALIGNED(16, uint16_t, ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
+  DECLARE_ALIGNED(16, uint8_t, msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
   DECLARE_ALIGNED(16, uint16_t,
-                  src_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
-  DECLARE_ALIGNED(16, uint16_t,
-                  ref_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
-  DECLARE_ALIGNED(16, uint8_t,
-                  msk_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
-  DECLARE_ALIGNED(16, uint16_t,
-                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1) + 7]);
+                  second_pred_ptr[(MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8)]);
   uint8_t *src8_ptr = CONVERT_TO_BYTEPTR(src_ptr);
   uint8_t *ref8_ptr = CONVERT_TO_BYTEPTR(ref_ptr);
   uint8_t *second_pred8_ptr = CONVERT_TO_BYTEPTR(second_pred_ptr);
@@ -285,21 +267,21 @@ TEST_P(HighbdMaskedSubPixelVarianceTest, ExtremeValues) {
   int first_failure_y = -1;
   int err_count = 0;
   int first_failure = -1;
-  int src_stride = (MAX_SB_SIZE + 1);
-  int ref_stride = (MAX_SB_SIZE + 1);
-  int msk_stride = (MAX_SB_SIZE + 1);
+  int src_stride = (MAX_SB_SIZE + 8);
+  int ref_stride = (MAX_SB_SIZE + 8);
+  int msk_stride = (MAX_SB_SIZE + 8);
 
   for (int xoffset = 0; xoffset < BIL_SUBPEL_SHIFTS; xoffset++) {
     for (int yoffset = 0; yoffset < BIL_SUBPEL_SHIFTS; yoffset++) {
       for (int i = 0; i < 16; ++i) {
         aom_memset16(src_ptr, (i & 0x1) ? ((1 << bit_depth_) - 1) : 0,
-                     (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+                     (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8));
         aom_memset16(ref_ptr, (i & 0x2) ? ((1 << bit_depth_) - 1) : 0,
-                     (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+                     (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8));
         aom_memset16(second_pred_ptr, (i & 0x4) ? ((1 << bit_depth_) - 1) : 0,
-                     (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+                     (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8));
         memset(msk_ptr, (i & 0x8) ? 64 : 0,
-               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 1));
+               (MAX_SB_SIZE + 1) * (MAX_SB_SIZE + 8));
 
         for (int invert_mask = 0; invert_mask < 2; ++invert_mask) {
           ref_ret = ref_func_(src8_ptr, src_stride, xoffset, yoffset, ref8_ptr,
@@ -334,172 +316,155 @@ TEST_P(HighbdMaskedSubPixelVarianceTest, ExtremeValues) {
 using std::tr1::make_tuple;
 
 #if HAVE_SSSE3
-INSTANTIATE_TEST_CASE_P(
-    SSSE3_C_COMPARE, MaskedSubPixelVarianceTest,
-    ::testing::Values(
+
+const MaskedSubPixelVarianceParam sub_pel_var_test[] = {
 #if CONFIG_EXT_PARTITION
-        make_tuple(&aom_masked_sub_pixel_variance128x128_ssse3,
-                   &aom_masked_sub_pixel_variance128x128_c),
-        make_tuple(&aom_masked_sub_pixel_variance128x64_ssse3,
-                   &aom_masked_sub_pixel_variance128x64_c),
-        make_tuple(&aom_masked_sub_pixel_variance64x128_ssse3,
-                   &aom_masked_sub_pixel_variance64x128_c),
+  make_tuple(&aom_masked_sub_pixel_variance128x128_ssse3,
+             &aom_masked_sub_pixel_variance128x128_c),
+  make_tuple(&aom_masked_sub_pixel_variance128x64_ssse3,
+             &aom_masked_sub_pixel_variance128x64_c),
+  make_tuple(&aom_masked_sub_pixel_variance64x128_ssse3,
+             &aom_masked_sub_pixel_variance64x128_c),
 #endif  // CONFIG_EXT_PARTITION
-        make_tuple(&aom_masked_sub_pixel_variance64x64_ssse3,
-                   &aom_masked_sub_pixel_variance64x64_c),
-        make_tuple(&aom_masked_sub_pixel_variance64x32_ssse3,
-                   &aom_masked_sub_pixel_variance64x32_c),
-        make_tuple(&aom_masked_sub_pixel_variance32x64_ssse3,
-                   &aom_masked_sub_pixel_variance32x64_c),
-        make_tuple(&aom_masked_sub_pixel_variance32x32_ssse3,
-                   &aom_masked_sub_pixel_variance32x32_c),
-        make_tuple(&aom_masked_sub_pixel_variance32x16_ssse3,
-                   &aom_masked_sub_pixel_variance32x16_c),
-        make_tuple(&aom_masked_sub_pixel_variance16x32_ssse3,
-                   &aom_masked_sub_pixel_variance16x32_c),
-        make_tuple(&aom_masked_sub_pixel_variance16x16_ssse3,
-                   &aom_masked_sub_pixel_variance16x16_c),
-        make_tuple(&aom_masked_sub_pixel_variance16x8_ssse3,
-                   &aom_masked_sub_pixel_variance16x8_c),
-        make_tuple(&aom_masked_sub_pixel_variance8x16_ssse3,
-                   &aom_masked_sub_pixel_variance8x16_c),
-        make_tuple(&aom_masked_sub_pixel_variance8x8_ssse3,
-                   &aom_masked_sub_pixel_variance8x8_c),
-        make_tuple(&aom_masked_sub_pixel_variance8x4_ssse3,
-                   &aom_masked_sub_pixel_variance8x4_c),
-        make_tuple(&aom_masked_sub_pixel_variance4x8_ssse3,
-                   &aom_masked_sub_pixel_variance4x8_c),
-        make_tuple(&aom_masked_sub_pixel_variance4x4_ssse3,
-                   &aom_masked_sub_pixel_variance4x4_c)));
+  make_tuple(&aom_masked_sub_pixel_variance64x64_ssse3,
+             &aom_masked_sub_pixel_variance64x64_c),
+  make_tuple(&aom_masked_sub_pixel_variance64x32_ssse3,
+             &aom_masked_sub_pixel_variance64x32_c),
+  make_tuple(&aom_masked_sub_pixel_variance32x64_ssse3,
+             &aom_masked_sub_pixel_variance32x64_c),
+  make_tuple(&aom_masked_sub_pixel_variance32x32_ssse3,
+             &aom_masked_sub_pixel_variance32x32_c),
+  make_tuple(&aom_masked_sub_pixel_variance32x16_ssse3,
+             &aom_masked_sub_pixel_variance32x16_c),
+  make_tuple(&aom_masked_sub_pixel_variance16x32_ssse3,
+             &aom_masked_sub_pixel_variance16x32_c),
+  make_tuple(&aom_masked_sub_pixel_variance16x16_ssse3,
+             &aom_masked_sub_pixel_variance16x16_c),
+  make_tuple(&aom_masked_sub_pixel_variance16x8_ssse3,
+             &aom_masked_sub_pixel_variance16x8_c),
+  make_tuple(&aom_masked_sub_pixel_variance8x16_ssse3,
+             &aom_masked_sub_pixel_variance8x16_c),
+  make_tuple(&aom_masked_sub_pixel_variance8x8_ssse3,
+             &aom_masked_sub_pixel_variance8x8_c),
+  make_tuple(&aom_masked_sub_pixel_variance8x4_ssse3,
+             &aom_masked_sub_pixel_variance8x4_c),
+  make_tuple(&aom_masked_sub_pixel_variance4x8_ssse3,
+             &aom_masked_sub_pixel_variance4x8_c),
+  make_tuple(&aom_masked_sub_pixel_variance4x4_ssse3,
+             &aom_masked_sub_pixel_variance4x4_c)
+};
+
+INSTANTIATE_TEST_CASE_P(SSSE3_C_COMPARE, MaskedSubPixelVarianceTest,
+                        ::testing::ValuesIn(sub_pel_var_test));
 
 #if CONFIG_HIGHBITDEPTH
-INSTANTIATE_TEST_CASE_P(
-    SSSE3_C_COMPARE, HighbdMaskedSubPixelVarianceTest,
-    ::testing::Values(
+const HighbdMaskedSubPixelVarianceParam hbd_sub_pel_var_test[] = {
 #if CONFIG_EXT_PARTITION
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance128x128_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance128x128_c,
-                   AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance128x64_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance128x64_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance64x128_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance64x128_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance128x128_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance128x128_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance128x64_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance128x64_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance64x128_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance64x128_c, AOM_BITS_8),
 #endif  // CONFIG_EXT_PARTITION
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance64x64_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance64x64_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance64x32_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance64x32_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance32x64_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance32x64_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance32x32_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance32x32_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance32x16_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance32x16_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance16x32_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance16x32_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance16x16_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance16x16_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance16x8_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance16x8_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance8x16_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance8x16_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance8x8_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance8x8_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance8x4_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance8x4_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance4x8_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance4x8_c, AOM_BITS_8),
-        make_tuple(&aom_highbd_8_masked_sub_pixel_variance4x4_ssse3,
-                   &aom_highbd_8_masked_sub_pixel_variance4x4_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance64x64_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance64x64_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance64x32_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance64x32_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance32x64_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance32x64_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance32x32_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance32x32_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance32x16_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance32x16_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance16x32_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance16x32_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance16x16_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance16x16_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance16x8_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance16x8_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance8x16_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance8x16_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance8x8_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance8x8_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance8x4_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance8x4_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance4x8_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance4x8_c, AOM_BITS_8),
+  make_tuple(&aom_highbd_8_masked_sub_pixel_variance4x4_ssse3,
+             &aom_highbd_8_masked_sub_pixel_variance4x4_c, AOM_BITS_8),
 #if CONFIG_EXT_PARTITION
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance128x128_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance128x128_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance128x64_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance128x64_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance64x128_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance64x128_c,
-                   AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance128x128_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance128x128_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance128x64_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance128x64_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance64x128_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance64x128_c, AOM_BITS_10),
 #endif  // CONFIG_EXT_PARTITION
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance64x64_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance64x64_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance64x32_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance64x32_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance32x64_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance32x64_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance32x32_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance32x32_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance32x16_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance32x16_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance16x32_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance16x32_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance16x16_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance16x16_c,
-                   AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance16x8_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance16x8_c, AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance8x16_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance8x16_c, AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance8x8_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance8x8_c, AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance8x4_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance8x4_c, AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance4x8_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance4x8_c, AOM_BITS_10),
-        make_tuple(&aom_highbd_10_masked_sub_pixel_variance4x4_ssse3,
-                   &aom_highbd_10_masked_sub_pixel_variance4x4_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance64x64_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance64x64_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance64x32_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance64x32_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance32x64_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance32x64_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance32x32_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance32x32_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance32x16_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance32x16_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance16x32_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance16x32_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance16x16_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance16x16_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance16x8_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance16x8_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance8x16_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance8x16_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance8x8_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance8x8_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance8x4_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance8x4_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance4x8_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance4x8_c, AOM_BITS_10),
+  make_tuple(&aom_highbd_10_masked_sub_pixel_variance4x4_ssse3,
+             &aom_highbd_10_masked_sub_pixel_variance4x4_c, AOM_BITS_10),
 #if CONFIG_EXT_PARTITION
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance128x128_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance128x128_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance128x64_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance128x64_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance64x128_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance64x128_c,
-                   AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance128x128_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance128x128_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance128x64_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance128x64_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance64x128_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance64x128_c, AOM_BITS_12),
 #endif  // CONFIG_EXT_PARTITION
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance64x64_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance64x64_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance64x32_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance64x32_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance32x64_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance32x64_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance32x32_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance32x32_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance32x16_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance32x16_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance16x32_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance16x32_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance16x16_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance16x16_c,
-                   AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance16x8_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance16x8_c, AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance8x16_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance8x16_c, AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance8x8_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance8x8_c, AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance8x4_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance8x4_c, AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance4x8_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance4x8_c, AOM_BITS_12),
-        make_tuple(&aom_highbd_12_masked_sub_pixel_variance4x4_ssse3,
-                   &aom_highbd_12_masked_sub_pixel_variance4x4_c,
-                   AOM_BITS_12)));
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance64x64_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance64x64_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance64x32_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance64x32_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance32x64_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance32x64_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance32x32_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance32x32_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance32x16_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance32x16_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance16x32_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance16x32_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance16x16_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance16x16_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance16x8_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance16x8_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance8x16_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance8x16_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance8x8_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance8x8_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance8x4_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance8x4_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance4x8_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance4x8_c, AOM_BITS_12),
+  make_tuple(&aom_highbd_12_masked_sub_pixel_variance4x4_ssse3,
+             &aom_highbd_12_masked_sub_pixel_variance4x4_c, AOM_BITS_12)
+};
+
+INSTANTIATE_TEST_CASE_P(SSSE3_C_COMPARE, HighbdMaskedSubPixelVarianceTest,
+                        ::testing::ValuesIn(hbd_sub_pel_var_test));
 #endif  // CONFIG_HIGHBITDEPTH
 
 #endif  // HAVE_SSSE3
diff --git a/third_party/aom/test/motion_vector_test.cc b/third_party/aom/test/motion_vector_test.cc
index fa47494e8..fe20fd10a 100644
--- a/third_party/aom/test/motion_vector_test.cc
+++ b/third_party/aom/test/motion_vector_test.cc
@@ -32,9 +32,9 @@ const int kCpuUsedVectors[] = { 1, 5 };
 const int kMVTestModes[] = { MAX_EXTREME_MV, MIN_EXTREME_MV };
 
 class MotionVectorTestLarge
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith3Params<libaom_test::TestMode, int,
-                                                 int> {
+    : public ::libaom_test::CodecTestWith3Params<libaom_test::TestMode, int,
+                                                 int>,
+      public ::libaom_test::EncoderTest {
  protected:
   MotionVectorTestLarge()
       : EncoderTest(GET_PARAM(0)), encoding_mode_(GET_PARAM(1)),
diff --git a/third_party/aom/test/partial_idct_test.cc b/third_party/aom/test/partial_idct_test.cc
index 033f18294..b2ea176e8 100644
--- a/third_party/aom/test/partial_idct_test.cc
+++ b/third_party/aom/test/partial_idct_test.cc
@@ -41,13 +41,6 @@ void wrapper(const tran_low_t *in, uint8_t *out, int stride, int bd) {
   fn(in, out, stride);
 }
 
-#if CONFIG_HIGHBITDEPTH
-template <InvTxfmWithBdFunc fn>
-void highbd_wrapper(const tran_low_t *in, uint8_t *out, int stride, int bd) {
-  fn(in, CONVERT_TO_BYTEPTR(out), stride, bd);
-}
-#endif
-
 typedef std::tr1::tuple<FwdTxfmFunc, InvTxfmWithBdFunc, InvTxfmWithBdFunc,
                         TX_SIZE, int, int, int>
     PartialInvTxfmParam;
@@ -285,26 +278,6 @@ TEST_P(PartialIDctTest, DISABLED_Speed) {
 using std::tr1::make_tuple;
 
 const PartialInvTxfmParam c_partial_idct_tests[] = {
-#if CONFIG_HIGHBITDEPTH
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>, TX_4X4, 16, 8, 2),
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>, TX_4X4, 16, 10, 2),
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>, TX_4X4, 16, 12, 2),
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_1_add_c>, TX_4X4, 1, 8, 2),
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_1_add_c>, TX_4X4, 1, 10, 2),
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_1_add_c>, TX_4X4, 1, 12, 2),
-#endif  // CONFIG_HIGHBITDEPTH
   make_tuple(&aom_fdct32x32_c, &wrapper<aom_idct32x32_1024_add_c>,
              &wrapper<aom_idct32x32_1024_add_c>, TX_32X32, 1024, 8, 1),
   make_tuple(&aom_fdct32x32_c, &wrapper<aom_idct32x32_1024_add_c>,
@@ -358,17 +331,6 @@ INSTANTIATE_TEST_CASE_P(NEON, PartialIDctTest,
 
 #if HAVE_SSE2
 const PartialInvTxfmParam sse2_partial_idct_tests[] = {
-#if CONFIG_HIGHBITDEPTH
-  make_tuple(&aom_highbd_fdct4x4_c,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-             &highbd_wrapper<aom_highbd_idct4x4_16_add_sse2>, TX_4X4, 16, 8, 2),
-  make_tuple(
-      &aom_highbd_fdct4x4_c, &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-      &highbd_wrapper<aom_highbd_idct4x4_16_add_sse2>, TX_4X4, 16, 10, 2),
-  make_tuple(
-      &aom_highbd_fdct4x4_c, &highbd_wrapper<aom_highbd_idct4x4_16_add_c>,
-      &highbd_wrapper<aom_highbd_idct4x4_16_add_sse2>, TX_4X4, 16, 12, 2),
-#endif  // CONFIG_HIGHBITDEPTH
   make_tuple(&aom_fdct32x32_c, &wrapper<aom_idct32x32_1024_add_c>,
              &wrapper<aom_idct32x32_1024_add_sse2>, TX_32X32, 1024, 8, 1),
   make_tuple(&aom_fdct32x32_c, &wrapper<aom_idct32x32_1024_add_c>,
diff --git a/third_party/aom/test/quantize_func_test.cc b/third_party/aom/test/quantize_func_test.cc
index 32b1d5139..94dd056b4 100644
--- a/third_party/aom/test/quantize_func_test.cc
+++ b/third_party/aom/test/quantize_func_test.cc
@@ -12,8 +12,10 @@
 #include "third_party/googletest/src/googletest/include/gtest/gtest.h"
 
 #include "./aom_config.h"
+#include "./aom_dsp_rtcd.h"
 #include "./av1_rtcd.h"
 #include "aom/aom_codec.h"
+#include "aom_ports/aom_timer.h"
 #include "av1/encoder/encoder.h"
 #include "av1/encoder/av1_quantize.h"
 #include "test/acm_random.h"
@@ -24,26 +26,44 @@
 namespace {
 using libaom_test::ACMRandom;
 
-#if !CONFIG_AOM_QM
-typedef void (*QuantizeFunc)(const tran_low_t *coeff_ptr, intptr_t n_coeffs,
-                             int skip_block, const int16_t *zbin_ptr,
-                             const int16_t *round_ptr, const int16_t *quant_ptr,
-                             const int16_t *quant_shift_ptr,
-                             tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr,
-                             const int16_t *dequant_ptr, uint16_t *eob_ptr,
-                             const int16_t *scan, const int16_t *iscan);
-#else
-typedef void (*QuantizeFunc)(const tran_low_t *coeff_ptr, intptr_t n_coeffs,
-                             int skip_block, const int16_t *zbin_ptr,
-                             const int16_t *round_ptr, const int16_t *quant_ptr,
-                             const int16_t *quant_shift_ptr,
-                             tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr,
-                             const int16_t *dequant_ptr, uint16_t *eob_ptr,
-                             const int16_t *scan, const int16_t *iscan,
-                             const qm_val_t *qm_ptr, const qm_val_t *iqm_ptr);
-#endif
+#define QUAN_PARAM_LIST                                                   \
+  const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block,         \
+      const int16_t *zbin_ptr, const int16_t *round_ptr,                  \
+      const int16_t *quant_ptr, const int16_t *quant_shift_ptr,           \
+      tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr,                    \
+      const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, \
+      const int16_t *iscan
+
+typedef void (*QuantizeFunc)(QUAN_PARAM_LIST);
+typedef void (*QuantizeFuncHbd)(QUAN_PARAM_LIST, int log_scale);
+
+#define HBD_QUAN_FUNC                                                      \
+  fn(coeff_ptr, n_coeffs, skip_block, zbin_ptr, round_ptr, quant_ptr,      \
+     quant_shift_ptr, qcoeff_ptr, dqcoeff_ptr, dequant_ptr, eob_ptr, scan, \
+     iscan, log_scale)
+
+template <QuantizeFuncHbd fn>
+void highbd_quan16x16_wrapper(QUAN_PARAM_LIST) {
+  const int log_scale = 0;
+  HBD_QUAN_FUNC;
+}
 
-typedef std::tr1::tuple<QuantizeFunc, QuantizeFunc, TX_SIZE, aom_bit_depth_t>
+template <QuantizeFuncHbd fn>
+void highbd_quan32x32_wrapper(QUAN_PARAM_LIST) {
+  const int log_scale = 1;
+  HBD_QUAN_FUNC;
+}
+
+template <QuantizeFuncHbd fn>
+void highbd_quan64x64_wrapper(QUAN_PARAM_LIST) {
+  const int log_scale = 2;
+  HBD_QUAN_FUNC;
+}
+
+typedef enum { TYPE_B, TYPE_DC, TYPE_FP } QuantType;
+
+typedef std::tr1::tuple<QuantizeFunc, QuantizeFunc, TX_SIZE, QuantType,
+                        aom_bit_depth_t>
     QuantizeParam;
 
 typedef struct {
@@ -57,15 +77,15 @@ class QuantizeTest : public ::testing::TestWithParam<QuantizeParam> {
  protected:
   QuantizeTest()
       : quant_ref_(GET_PARAM(0)), quant_(GET_PARAM(1)), tx_size_(GET_PARAM(2)),
-        bd_(GET_PARAM(3)) {}
+        type_(GET_PARAM(3)), bd_(GET_PARAM(4)) {}
 
   virtual ~QuantizeTest() {}
 
   virtual void SetUp() {
-    qtab_ = reinterpret_cast<QuanTable *>(aom_memalign(16, sizeof(*qtab_)));
-    const int n_coeffs = getCoeffNum();
+    qtab_ = reinterpret_cast<QuanTable *>(aom_memalign(32, sizeof(*qtab_)));
+    const int n_coeffs = coeff_num();
     coeff_ = reinterpret_cast<tran_low_t *>(
-        aom_memalign(16, 6 * n_coeffs * sizeof(tran_low_t)));
+        aom_memalign(32, 6 * n_coeffs * sizeof(tran_low_t)));
     InitQuantizer();
   }
 
@@ -81,9 +101,9 @@ class QuantizeTest : public ::testing::TestWithParam<QuantizeParam> {
     av1_build_quantizer(bd_, 0, 0, 0, &qtab_->quant, &qtab_->dequant);
   }
 
-  void QuantizeRun(bool isLoop, int q = 0, int testNum = 1) {
+  void QuantizeRun(bool is_loop, int q = 0, int test_num = 1) {
     tran_low_t *coeff_ptr = coeff_;
-    const intptr_t n_coeffs = getCoeffNum();
+    const intptr_t n_coeffs = coeff_num();
     const int skip_block = 0;
 
     tran_low_t *qcoeff_ref = coeff_ptr + n_coeffs;
@@ -98,31 +118,47 @@ class QuantizeTest : public ::testing::TestWithParam<QuantizeParam> {
 
     // Testing uses luminance quantization table
     const int16_t *zbin = qtab_->quant.y_zbin[q];
-    const int16_t *round_fp = qtab_->quant.y_round_fp[q];
-    const int16_t *quant_fp = qtab_->quant.y_quant_fp[q];
+
+    const int16_t *round = 0;
+    const int16_t *quant = 0;
+    if (type_ == TYPE_B) {
+      round = qtab_->quant.y_round[q];
+      quant = qtab_->quant.y_quant[q];
+    } else if (type_ == TYPE_FP) {
+      round = qtab_->quant.y_round_fp[q];
+      quant = qtab_->quant.y_quant_fp[q];
+    }
+
     const int16_t *quant_shift = qtab_->quant.y_quant_shift[q];
     const int16_t *dequant = qtab_->dequant.y_dequant[q];
-    const size_t bufferSize = n_coeffs;
 
-    int i = 0;
-    while (i < testNum) {
-      if (isLoop) FillCoeffRandom();
+    for (int i = 0; i < test_num; ++i) {
+      if (is_loop) FillCoeffRandom();
 
       memset(qcoeff_ref, 0, 5 * n_coeffs * sizeof(*qcoeff_ref));
 
-      quant_ref_(coeff_ptr, n_coeffs, skip_block, zbin, round_fp, quant_fp,
+      quant_ref_(coeff_ptr, n_coeffs, skip_block, zbin, round, quant,
                  quant_shift, qcoeff_ref, dqcoeff_ref, dequant, &eob[0],
                  sc->scan, sc->iscan);
 
       ASM_REGISTER_STATE_CHECK(quant_(
-          coeff_ptr, n_coeffs, skip_block, zbin, round_fp, quant_fp,
-          quant_shift, qcoeff, dqcoeff, dequant, &eob[1], sc->scan, sc->iscan));
+          coeff_ptr, n_coeffs, skip_block, zbin, round, quant, quant_shift,
+          qcoeff, dqcoeff, dequant, &eob[1], sc->scan, sc->iscan));
 
-      CompareResults(qcoeff_ref, qcoeff, bufferSize, "Qcoeff", q, i);
-      CompareResults(dqcoeff_ref, dqcoeff, bufferSize, "Dqcoeff", q, i);
-      ASSERT_EQ(eob[0], eob[1]) << "eobs mismatch on test: " << i;
+      for (int j = 0; j < n_coeffs; ++j) {
+        ASSERT_EQ(qcoeff_ref[j], qcoeff[j])
+            << "Q mismatch on test: " << i << " at position: " << j
+            << " Q: " << q << " coeff: " << coeff_ptr[j];
+      }
 
-      i++;
+      for (int j = 0; j < n_coeffs; ++j) {
+        ASSERT_EQ(dqcoeff_ref[j], dqcoeff[j])
+            << "Dq mismatch on test: " << i << " at position: " << j
+            << " Q: " << q << " coeff: " << coeff_ptr[j];
+      }
+
+      ASSERT_EQ(eob[0], eob[1]) << "eobs mismatch on test: " << i
+                                << " Q: " << q;
     }
   }
 
@@ -135,29 +171,29 @@ class QuantizeTest : public ::testing::TestWithParam<QuantizeParam> {
     }
   }
 
-  int getCoeffNum() { return tx_size_2d[tx_size_]; }
+  int coeff_num() const { return tx_size_2d[tx_size_]; }
 
-  void FillCoeffGeneric(bool isConstant, tran_low_t c = 0) {
-    const int n_coeffs = getCoeffNum();
-    int i;
-    if (isConstant) {
-      for (i = 0; i < n_coeffs; ++i) {
-        coeff_[i] = c;
-      }
-    } else {
-      FillCoeffZero();
-      int num = rnd_.Rand16() % n_coeffs;
-      for (i = 0; i < num; ++i) {
-        coeff_[i] = GetRandomCoeff();
-      }
+  void FillCoeff(tran_low_t c) {
+    const int n_coeffs = coeff_num();
+    for (int i = 0; i < n_coeffs; ++i) {
+      coeff_[i] = c;
     }
   }
 
-  void FillCoeffZero() { FillCoeffGeneric(true); }
+  void FillCoeffRandom() {
+    const int n_coeffs = coeff_num();
+    FillCoeffZero();
+    int num = rnd_.Rand16() % n_coeffs;
+    for (int i = 0; i < num; ++i) {
+      coeff_[i] = GetRandomCoeff();
+    }
+  }
+
+  void FillCoeffZero() { FillCoeff(0); }
 
   void FillCoeffConstant() {
     tran_low_t c = GetRandomCoeff();
-    FillCoeffGeneric(true, c);
+    FillCoeff(c);
   }
 
   void FillDcOnly() {
@@ -173,10 +209,17 @@ class QuantizeTest : public ::testing::TestWithParam<QuantizeParam> {
     coeff_[0] = -8191;
   }
 
-  void FillCoeffRandom() { FillCoeffGeneric(false); }
-
   tran_low_t GetRandomCoeff() {
-    return clamp((int16_t)rnd_.Rand16(), INT16_MIN + 1, INT16_MAX);
+    tran_low_t coeff;
+    if (bd_ == AOM_BITS_8) {
+      coeff =
+          clamp(static_cast<int16_t>(rnd_.Rand16()), INT16_MIN + 1, INT16_MAX);
+    } else {
+      tran_low_t min = -(1 << (7 + bd_));
+      tran_low_t max = -min - 1;
+      coeff = clamp(static_cast<tran_low_t>(rnd_.Rand31()), min, max);
+    }
+    return coeff;
   }
 
   ACMRandom rnd_;
@@ -185,6 +228,7 @@ class QuantizeTest : public ::testing::TestWithParam<QuantizeParam> {
   QuantizeFunc quant_ref_;
   QuantizeFunc quant_;
   TX_SIZE tx_size_;
+  QuantType type_;
   aom_bit_depth_t bd_;
 };
 
@@ -195,12 +239,12 @@ TEST_P(QuantizeTest, ZeroInput) {
 
 TEST_P(QuantizeTest, LargeNegativeInput) {
   FillDcLargeNegative();
-  QuantizeRun(false);
+  QuantizeRun(false, 0, 1);
 }
 
 TEST_P(QuantizeTest, DcOnlyInput) {
   FillDcOnly();
-  QuantizeRun(false);
+  QuantizeRun(false, 0, 1);
 }
 
 TEST_P(QuantizeTest, RandomInput) { QuantizeRun(true, 0, kTestNum); }
@@ -211,26 +255,134 @@ TEST_P(QuantizeTest, MultipleQ) {
   }
 }
 
+TEST_P(QuantizeTest, DISABLED_Speed) {
+  tran_low_t *coeff_ptr = coeff_;
+  const intptr_t n_coeffs = coeff_num();
+  const int skip_block = 0;
+
+  tran_low_t *qcoeff_ref = coeff_ptr + n_coeffs;
+  tran_low_t *dqcoeff_ref = qcoeff_ref + n_coeffs;
+
+  tran_low_t *qcoeff = dqcoeff_ref + n_coeffs;
+  tran_low_t *dqcoeff = qcoeff + n_coeffs;
+  uint16_t *eob = (uint16_t *)(dqcoeff + n_coeffs);
+
+  // Testing uses 2-D DCT scan order table
+  const SCAN_ORDER *const sc = get_default_scan(tx_size_, DCT_DCT, 0);
+
+  // Testing uses luminance quantization table
+  const int q = 22;
+  const int16_t *zbin = qtab_->quant.y_zbin[q];
+  const int16_t *round_fp = qtab_->quant.y_round_fp[q];
+  const int16_t *quant_fp = qtab_->quant.y_quant_fp[q];
+  const int16_t *quant_shift = qtab_->quant.y_quant_shift[q];
+  const int16_t *dequant = qtab_->dequant.y_dequant[q];
+  const int kNumTests = 5000000;
+  aom_usec_timer timer;
+
+  FillCoeffRandom();
+
+  aom_usec_timer_start(&timer);
+  for (int n = 0; n < kNumTests; ++n) {
+    quant_(coeff_ptr, n_coeffs, skip_block, zbin, round_fp, quant_fp,
+           quant_shift, qcoeff, dqcoeff, dequant, eob, sc->scan, sc->iscan);
+  }
+  aom_usec_timer_mark(&timer);
+
+  const int elapsed_time = static_cast<int>(aom_usec_timer_elapsed(&timer));
+  printf("Elapsed time: %d us\n", elapsed_time);
+}
+
 using std::tr1::make_tuple;
 
+#if HAVE_AVX2
+const QuantizeParam kQParamArrayAvx2[] = {
+  make_tuple(&av1_quantize_fp_c, &av1_quantize_fp_avx2, TX_16X16, TYPE_FP,
+             AOM_BITS_8),
+  make_tuple(&av1_quantize_fp_32x32_c, &av1_quantize_fp_32x32_avx2, TX_32X32,
+             TYPE_FP, AOM_BITS_8),
+#if CONFIG_HIGHBITDEPTH
+  make_tuple(&highbd_quan16x16_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan16x16_wrapper<av1_highbd_quantize_fp_avx2>, TX_16X16,
+             TYPE_FP, AOM_BITS_8),
+  make_tuple(&highbd_quan16x16_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan16x16_wrapper<av1_highbd_quantize_fp_avx2>, TX_16X16,
+             TYPE_FP, AOM_BITS_10),
+  make_tuple(&highbd_quan16x16_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan16x16_wrapper<av1_highbd_quantize_fp_avx2>, TX_16X16,
+             TYPE_FP, AOM_BITS_12),
+  make_tuple(&highbd_quan32x32_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan32x32_wrapper<av1_highbd_quantize_fp_avx2>, TX_32X32,
+             TYPE_FP, AOM_BITS_8),
+  make_tuple(&highbd_quan32x32_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan32x32_wrapper<av1_highbd_quantize_fp_avx2>, TX_32X32,
+             TYPE_FP, AOM_BITS_10),
+  make_tuple(&highbd_quan32x32_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan32x32_wrapper<av1_highbd_quantize_fp_avx2>, TX_32X32,
+             TYPE_FP, AOM_BITS_12),
+#if CONFIG_TX64X64
+  make_tuple(&highbd_quan64x64_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan64x64_wrapper<av1_highbd_quantize_fp_avx2>, TX_64X64,
+             TYPE_FP, AOM_BITS_8),
+  make_tuple(&highbd_quan64x64_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan64x64_wrapper<av1_highbd_quantize_fp_avx2>, TX_64X64,
+             TYPE_FP, AOM_BITS_10),
+  make_tuple(&highbd_quan64x64_wrapper<av1_highbd_quantize_fp_c>,
+             &highbd_quan64x64_wrapper<av1_highbd_quantize_fp_avx2>, TX_64X64,
+             TYPE_FP, AOM_BITS_12),
+#endif  // CONFIG_TX64X64
+  make_tuple(&aom_highbd_quantize_b_c, &aom_highbd_quantize_b_avx2, TX_16X16,
+             TYPE_B, AOM_BITS_8),
+  make_tuple(&aom_highbd_quantize_b_c, &aom_highbd_quantize_b_avx2, TX_16X16,
+             TYPE_B, AOM_BITS_10),
+  make_tuple(&aom_highbd_quantize_b_c, &aom_highbd_quantize_b_avx2, TX_16X16,
+             TYPE_B, AOM_BITS_12),
+#endif  // CONFIG_HIGHBITDEPTH
+};
+
+INSTANTIATE_TEST_CASE_P(AVX2, QuantizeTest,
+                        ::testing::ValuesIn(kQParamArrayAvx2));
+#endif  // HAVE_AVX2
+
 #if HAVE_SSE2
-const QuantizeParam kQParamArraySSE2[] = { make_tuple(
-    &av1_quantize_fp_c, &av1_quantize_fp_sse2, TX_16X16, AOM_BITS_8) };
+const QuantizeParam kQParamArraySSE2[] = {
+  make_tuple(&av1_quantize_fp_c, &av1_quantize_fp_sse2, TX_16X16, TYPE_FP,
+             AOM_BITS_8),
+#if CONFIG_HIGHBITDEPTH
+  make_tuple(&aom_highbd_quantize_b_c, &aom_highbd_quantize_b_sse2, TX_16X16,
+             TYPE_B, AOM_BITS_8),
+  make_tuple(&aom_highbd_quantize_b_c, &aom_highbd_quantize_b_sse2, TX_16X16,
+             TYPE_B, AOM_BITS_10),
+  make_tuple(&aom_highbd_quantize_b_c, &aom_highbd_quantize_b_sse2, TX_16X16,
+             TYPE_B, AOM_BITS_12),
+  make_tuple(&aom_highbd_quantize_b_32x32_c, &aom_highbd_quantize_b_32x32_sse2,
+             TX_32X32, TYPE_B, AOM_BITS_8),
+  make_tuple(&aom_highbd_quantize_b_32x32_c, &aom_highbd_quantize_b_32x32_sse2,
+             TX_32X32, TYPE_B, AOM_BITS_10),
+  make_tuple(&aom_highbd_quantize_b_32x32_c, &aom_highbd_quantize_b_32x32_sse2,
+             TX_32X32, TYPE_B, AOM_BITS_12),
+#endif
+};
 
 INSTANTIATE_TEST_CASE_P(SSE2, QuantizeTest,
                         ::testing::ValuesIn(kQParamArraySSE2));
 #endif
 
 #if !CONFIG_HIGHBITDEPTH && HAVE_SSSE3 && ARCH_X86_64
-const QuantizeParam kQParamArraySSSE3[] = {
-  make_tuple(&av1_quantize_fp_c, &av1_quantize_fp_ssse3, TX_16X16, AOM_BITS_8),
-  // TODO(any):
-  //  The following test couldn't pass yet
-  // make_tuple(av1_quantize_fp_c, av1_quantize_fp_32x32_ssse3, TX_32X32,
-  // AOM_BITS_8)
+const QuantizeParam kQ16x16ParamArraySSSE3[] = {
+  make_tuple(&av1_quantize_fp_c, &av1_quantize_fp_ssse3, TX_16X16, TYPE_FP,
+             AOM_BITS_8),
 };
 INSTANTIATE_TEST_CASE_P(SSSE3, QuantizeTest,
-                        ::testing::ValuesIn(kQParamArraySSSE3));
+                        ::testing::ValuesIn(kQ16x16ParamArraySSSE3));
+
+// TODO(any):
+//  The following test does not pass yet
+const QuantizeParam kQ32x32ParamArraySSSE3[] = { make_tuple(
+    &av1_quantize_fp_32x32_c, &av1_quantize_fp_32x32_ssse3, TX_32X32, TYPE_FP,
+    AOM_BITS_8) };
+INSTANTIATE_TEST_CASE_P(DISABLED_SSSE3, QuantizeTest,
+                        ::testing::ValuesIn(kQ32x32ParamArraySSSE3));
 #endif
 
 }  // namespace
diff --git a/third_party/aom/test/quantize_test.cc b/third_party/aom/test/quantize_test.cc
deleted file mode 100644
index 4f61484a2..000000000
--- a/third_party/aom/test/quantize_test.cc
+++ /dev/null
@@ -1,205 +0,0 @@
-/*
- * Copyright (c) 2016, Alliance for Open Media. All rights reserved
- *
- * This source code is subject to the terms of the BSD 2 Clause License and
- * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
- * was not distributed with this source code in the LICENSE file, you can
- * obtain it at www.aomedia.org/license/software. If the Alliance for Open
- * Media Patent License 1.0 was not distributed with this source code in the
- * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
-*/
-
-#include <string.h>
-
-#include "third_party/googletest/src/googletest/include/gtest/gtest.h"
-
-#include "./aom_config.h"
-#include "test/acm_random.h"
-#include "test/clear_system_state.h"
-#include "test/register_state_check.h"
-#include "test/util.h"
-#include "vp8/common/blockd.h"
-#include "vp8/common/onyx.h"
-#include "vp8/encoder/block.h"
-#include "vp8/encoder/onyx_int.h"
-#include "vp8/encoder/quantize.h"
-#include "aom/aom_integer.h"
-#include "aom_mem/aom_mem.h"
-
-namespace {
-#if !CONFIG_AOM_QM
-
-const int kNumBlocks = 25;
-const int kNumBlockEntries = 16;
-
-typedef void (*VP8Quantize)(BLOCK *b, BLOCKD *d);
-
-typedef std::tr1::tuple<VP8Quantize, VP8Quantize> VP8QuantizeParam;
-
-using libaom_test::ACMRandom;
-using std::tr1::make_tuple;
-
-// Create and populate a VP8_COMP instance which has a complete set of
-// quantization inputs as well as a second MACROBLOCKD for output.
-class QuantizeTestBase {
- public:
-  virtual ~QuantizeTestBase() {
-    vp8_remove_compressor(&vp8_comp_);
-    vp8_comp_ = NULL;
-    aom_free(macroblockd_dst_);
-    macroblockd_dst_ = NULL;
-    libaom_test::ClearSystemState();
-  }
-
- protected:
-  void SetupCompressor() {
-    rnd_.Reset(ACMRandom::DeterministicSeed());
-
-    // The full configuration is necessary to generate the quantization tables.
-    VP8_CONFIG vp8_config;
-    memset(&vp8_config, 0, sizeof(vp8_config));
-
-    vp8_comp_ = vp8_create_compressor(&vp8_config);
-
-    // Set the tables based on a quantizer of 0.
-    vp8_set_quantizer(vp8_comp_, 0);
-
-    // Set up all the block/blockd pointers for the mb in vp8_comp_.
-    vp8cx_frame_init_quantizer(vp8_comp_);
-
-    // Copy macroblockd from the reference to get pre-set-up dequant values.
-    macroblockd_dst_ = reinterpret_cast<MACROBLOCKD *>(
-        aom_memalign(32, 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_);
-  }
-
-  void UpdateQuantizer(int q) {
-    vp8_set_quantizer(vp8_comp_, q);
-
-    memcpy(macroblockd_dst_, &vp8_comp_->mb.e_mbd, sizeof(*macroblockd_dst_));
-    vp8_setup_block_dptrs(macroblockd_dst_);
-  }
-
-  void FillCoeffConstant(int16_t c) {
-    for (int i = 0; i < kNumBlocks * kNumBlockEntries; ++i) {
-      vp8_comp_->mb.coeff[i] = c;
-    }
-  }
-
-  void FillCoeffRandom() {
-    for (int i = 0; i < kNumBlocks * kNumBlockEntries; ++i) {
-      vp8_comp_->mb.coeff[i] = rnd_.Rand8();
-    }
-  }
-
-  void CheckOutput() {
-    EXPECT_EQ(0, memcmp(vp8_comp_->mb.e_mbd.qcoeff, macroblockd_dst_->qcoeff,
-                        sizeof(*macroblockd_dst_->qcoeff) * kNumBlocks *
-                            kNumBlockEntries))
-        << "qcoeff mismatch";
-    EXPECT_EQ(0, memcmp(vp8_comp_->mb.e_mbd.dqcoeff, macroblockd_dst_->dqcoeff,
-                        sizeof(*macroblockd_dst_->dqcoeff) * kNumBlocks *
-                            kNumBlockEntries))
-        << "dqcoeff mismatch";
-    EXPECT_EQ(0, memcmp(vp8_comp_->mb.e_mbd.eobs, macroblockd_dst_->eobs,
-                        sizeof(*macroblockd_dst_->eobs) * kNumBlocks))
-        << "eobs mismatch";
-  }
-
-  VP8_COMP *vp8_comp_;
-  MACROBLOCKD *macroblockd_dst_;
-
- private:
-  ACMRandom rnd_;
-};
-
-class QuantizeTest : public QuantizeTestBase,
-                     public ::testing::TestWithParam<VP8QuantizeParam> {
- protected:
-  virtual void SetUp() {
-    SetupCompressor();
-    asm_quant_ = GET_PARAM(0);
-    c_quant_ = GET_PARAM(1);
-  }
-
-  void RunComparison() {
-    for (int i = 0; i < kNumBlocks; ++i) {
-      ASM_REGISTER_STATE_CHECK(
-          c_quant_(&vp8_comp_->mb.block[i], &vp8_comp_->mb.e_mbd.block[i]));
-      ASM_REGISTER_STATE_CHECK(
-          asm_quant_(&vp8_comp_->mb.block[i], &macroblockd_dst_->block[i]));
-    }
-
-    CheckOutput();
-  }
-
- private:
-  VP8Quantize asm_quant_;
-  VP8Quantize c_quant_;
-};
-
-TEST_P(QuantizeTest, TestZeroInput) {
-  FillCoeffConstant(0);
-  RunComparison();
-}
-
-TEST_P(QuantizeTest, TestLargeNegativeInput) {
-  FillCoeffConstant(0);
-  // Generate a qcoeff which contains 512/-512 (0x0100/0xFE00) to catch issues
-  // like BUG=883 where the constant being compared was incorrectly initialized.
-  vp8_comp_->mb.coeff[0] = -8191;
-  RunComparison();
-}
-
-TEST_P(QuantizeTest, TestRandomInput) {
-  FillCoeffRandom();
-  RunComparison();
-}
-
-TEST_P(QuantizeTest, TestMultipleQ) {
-  for (int q = 0; q < QINDEX_RANGE; ++q) {
-    UpdateQuantizer(q);
-    FillCoeffRandom();
-    RunComparison();
-  }
-}
-
-#if HAVE_SSE2
-INSTANTIATE_TEST_CASE_P(
-    SSE2, QuantizeTest,
-    ::testing::Values(
-        make_tuple(&vp8_fast_quantize_b_sse2, &vp8_fast_quantize_b_c),
-        make_tuple(&vp8_regular_quantize_b_sse2, &vp8_regular_quantize_b_c)));
-#endif  // HAVE_SSE2
-
-#if HAVE_SSSE3
-INSTANTIATE_TEST_CASE_P(SSSE3, QuantizeTest,
-                        ::testing::Values(make_tuple(&vp8_fast_quantize_b_ssse3,
-                                                     &vp8_fast_quantize_b_c)));
-#endif  // HAVE_SSSE3
-
-#if HAVE_SSE4_1
-INSTANTIATE_TEST_CASE_P(
-    SSE4_1, QuantizeTest,
-    ::testing::Values(make_tuple(&vp8_regular_quantize_b_sse4_1,
-                                 &vp8_regular_quantize_b_c)));
-#endif  // HAVE_SSE4_1
-
-#if HAVE_NEON
-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
-#endif  // CONFIG_AOM_QM
-}  // namespace
diff --git a/third_party/aom/test/realtime_test.cc b/third_party/aom/test/realtime_test.cc
index ffe4a3146..11d2a3241 100644
--- a/third_party/aom/test/realtime_test.cc
+++ b/third_party/aom/test/realtime_test.cc
@@ -22,8 +22,8 @@ const int kVideoSourceHeight = 240;
 const int kFramesToEncode = 2;
 
 class RealtimeTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   RealtimeTest() : EncoderTest(GET_PARAM(0)), frame_packets_(0) {}
   virtual ~RealtimeTest() {}
diff --git a/third_party/aom/test/resize_test.cc b/third_party/aom/test/resize_test.cc
index 994b30117..802713d32 100644
--- a/third_party/aom/test/resize_test.cc
+++ b/third_party/aom/test/resize_test.cc
@@ -265,8 +265,8 @@ class ResizingVideoSource : public ::libaom_test::DummyVideoSource {
 };
 
 class ResizeTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<libaom_test::TestMode> {
+    : public ::libaom_test::CodecTestWithParam<libaom_test::TestMode>,
+      public ::libaom_test::EncoderTest {
  protected:
   ResizeTest() : EncoderTest(GET_PARAM(0)) {}
 
@@ -427,8 +427,8 @@ TEST_P(ResizeInternalTest, TestInternalResizeChangeConfig) {
 }
 
 class ResizeRealtimeTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int> {
+    : public ::libaom_test::CodecTestWith2Params<libaom_test::TestMode, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   ResizeRealtimeTest() : EncoderTest(GET_PARAM(0)) {}
   virtual ~ResizeRealtimeTest() {}
@@ -482,8 +482,6 @@ class ResizeRealtimeTest
     cfg_.rc_dropframe_thresh = 1;
     // Enable error_resilience mode.
     cfg_.g_error_resilient = 1;
-    // Enable dynamic resizing.
-    cfg_.rc_resize_allowed = 1;
     // Run at low bitrate.
     cfg_.rc_target_bitrate = 200;
   }
@@ -499,8 +497,6 @@ TEST_P(ResizeRealtimeTest, TestExternalResizeWorks) {
   ResizingVideoSource video;
   video.flag_codec_ = 1;
   DefaultConfig();
-  // Disable internal resize for this test.
-  cfg_.rc_resize_allowed = 0;
   change_bitrate_ = false;
   mismatch_psnr_ = 0.0;
   mismatch_nframes_ = 0;
diff --git a/third_party/aom/test/sad_test.cc b/third_party/aom/test/sad_test.cc
index 7564a6760..9ac58e653 100644
--- a/third_party/aom/test/sad_test.cc
+++ b/third_party/aom/test/sad_test.cc
@@ -231,8 +231,8 @@ class SADTestBase : public ::testing::Test {
   ACMRandom rnd_;
 };
 
-class SADx4Test : public SADTestBase,
-                  public ::testing::WithParamInterface<SadMxNx4Param> {
+class SADx4Test : public ::testing::WithParamInterface<SadMxNx4Param>,
+                  public SADTestBase {
  public:
   SADx4Test() : SADTestBase(GET_PARAM(0), GET_PARAM(1), GET_PARAM(3)) {}
 
@@ -257,8 +257,8 @@ class SADx4Test : public SADTestBase,
   }
 };
 
-class SADTest : public SADTestBase,
-                public ::testing::WithParamInterface<SadMxNParam> {
+class SADTest : public ::testing::WithParamInterface<SadMxNParam>,
+                public SADTestBase {
  public:
   SADTest() : SADTestBase(GET_PARAM(0), GET_PARAM(1), GET_PARAM(3)) {}
 
@@ -288,8 +288,8 @@ class SADTest : public SADTestBase,
   }
 };
 
-class SADavgTest : public SADTestBase,
-                   public ::testing::WithParamInterface<SadMxNAvgParam> {
+class SADavgTest : public ::testing::WithParamInterface<SadMxNAvgParam>,
+                   public SADTestBase {
  public:
   SADavgTest() : SADTestBase(GET_PARAM(0), GET_PARAM(1), GET_PARAM(3)) {}
 
diff --git a/third_party/aom/test/selfguided_filter_test.cc b/third_party/aom/test/selfguided_filter_test.cc
index e87fe339a..736e3f4c0 100644
--- a/third_party/aom/test/selfguided_filter_test.cc
+++ b/third_party/aom/test/selfguided_filter_test.cc
@@ -47,7 +47,6 @@ class AV1SelfguidedFilterTest
     uint8_t *input = (uint8_t *)aom_memalign(16, w * h * sizeof(uint8_t));
     uint8_t *output = (uint8_t *)aom_memalign(16, w * h * sizeof(uint8_t));
     int32_t *tmpbuf = (int32_t *)aom_memalign(16, RESTORATION_TMPBUF_SIZE);
-    memset(tmpbuf, 0, RESTORATION_TMPBUF_SIZE);
 
     ACMRandom rnd(ACMRandom::DeterministicSeed());
 
@@ -96,7 +95,6 @@ class AV1SelfguidedFilterTest
     uint8_t *output2 =
         (uint8_t *)aom_memalign(16, out_stride * max_h * sizeof(uint8_t));
     int32_t *tmpbuf = (int32_t *)aom_memalign(16, RESTORATION_TMPBUF_SIZE);
-    memset(tmpbuf, 0, RESTORATION_TMPBUF_SIZE);
 
     ACMRandom rnd(ACMRandom::DeterministicSeed());
 
@@ -167,7 +165,6 @@ class AV1HighbdSelfguidedFilterTest
     uint16_t *input = (uint16_t *)aom_memalign(16, w * h * sizeof(uint16_t));
     uint16_t *output = (uint16_t *)aom_memalign(16, w * h * sizeof(uint16_t));
     int32_t *tmpbuf = (int32_t *)aom_memalign(16, RESTORATION_TMPBUF_SIZE);
-    memset(tmpbuf, 0, RESTORATION_TMPBUF_SIZE);
 
     ACMRandom rnd(ACMRandom::DeterministicSeed());
 
@@ -219,7 +216,6 @@ class AV1HighbdSelfguidedFilterTest
     uint16_t *output2 =
         (uint16_t *)aom_memalign(16, out_stride * max_h * sizeof(uint16_t));
     int32_t *tmpbuf = (int32_t *)aom_memalign(16, RESTORATION_TMPBUF_SIZE);
-    memset(tmpbuf, 0, RESTORATION_TMPBUF_SIZE);
 
     ACMRandom rnd(ACMRandom::DeterministicSeed());
 
diff --git a/third_party/aom/test/simd_cmp_impl.h b/third_party/aom/test/simd_cmp_impl.h
index 4a9c1f7be..46f46d751 100644
--- a/third_party/aom/test/simd_cmp_impl.h
+++ b/third_party/aom/test/simd_cmp_impl.h
@@ -1590,10 +1590,10 @@ void TestSimd2Args(uint32_t iterations, uint32_t mask, uint32_t maskwidth,
                typeid(CArg2) == typeid(c_v256)) {
       // S64_V256V256
       error = CompareSimd2Args<int64_t, v256, v256, CRet, CArg1, CArg2>(
-          reinterpret_cast<fptr>(u64_store_aligned),
+          reinterpret_cast<fptr>(s64_store_aligned),
           reinterpret_cast<fptr>(v256_load_aligned),
           reinterpret_cast<fptr>(v256_load_aligned), simd, d,
-          reinterpret_cast<fptr>(c_u64_store_aligned),
+          reinterpret_cast<fptr>(c_s64_store_aligned),
           reinterpret_cast<fptr>(c_v256_load_aligned),
           reinterpret_cast<fptr>(c_v256_load_aligned),
           reinterpret_cast<fptr>(ref_simd), ref_d, s1, s2);
diff --git a/third_party/aom/test/subtract_test.cc b/third_party/aom/test/subtract_test.cc
index c90ca8d56..ad39f56b3 100644
--- a/third_party/aom/test/subtract_test.cc
+++ b/third_party/aom/test/subtract_test.cc
@@ -213,40 +213,44 @@ TEST_P(AV1HBDSubtractBlockTest, CheckSpeed) { RunForSpeed(); }
 #endif  // USE_SPEED_TEST
 
 #if HAVE_SSE2
-INSTANTIATE_TEST_CASE_P(
-    SSE2, AV1HBDSubtractBlockTest,
-    ::testing::Values(make_tuple(4, 4, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(4, 4, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(4, 8, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(4, 8, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(8, 4, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(8, 4, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(8, 8, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(8, 8, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(8, 16, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(8, 16, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(16, 8, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(16, 8, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(16, 16, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(16, 16, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(16, 32, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(16, 32, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(32, 16, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(32, 16, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(32, 32, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(32, 32, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(32, 64, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(32, 64, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(64, 32, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(64, 32, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(64, 64, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(64, 64, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(64, 128, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(64, 128, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(128, 64, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(128, 64, 12, &aom_highbd_subtract_block_c),
-                      make_tuple(128, 128, 12, &aom_highbd_subtract_block_sse2),
-                      make_tuple(128, 128, 12, &aom_highbd_subtract_block_c)));
+
+const Params kAV1HBDSubtractBlock_sse2[] = {
+  make_tuple(4, 4, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(4, 4, 12, &aom_highbd_subtract_block_c),
+  make_tuple(4, 8, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(4, 8, 12, &aom_highbd_subtract_block_c),
+  make_tuple(8, 4, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(8, 4, 12, &aom_highbd_subtract_block_c),
+  make_tuple(8, 8, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(8, 8, 12, &aom_highbd_subtract_block_c),
+  make_tuple(8, 16, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(8, 16, 12, &aom_highbd_subtract_block_c),
+  make_tuple(16, 8, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(16, 8, 12, &aom_highbd_subtract_block_c),
+  make_tuple(16, 16, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(16, 16, 12, &aom_highbd_subtract_block_c),
+  make_tuple(16, 32, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(16, 32, 12, &aom_highbd_subtract_block_c),
+  make_tuple(32, 16, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(32, 16, 12, &aom_highbd_subtract_block_c),
+  make_tuple(32, 32, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(32, 32, 12, &aom_highbd_subtract_block_c),
+  make_tuple(32, 64, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(32, 64, 12, &aom_highbd_subtract_block_c),
+  make_tuple(64, 32, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(64, 32, 12, &aom_highbd_subtract_block_c),
+  make_tuple(64, 64, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(64, 64, 12, &aom_highbd_subtract_block_c),
+  make_tuple(64, 128, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(64, 128, 12, &aom_highbd_subtract_block_c),
+  make_tuple(128, 64, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(128, 64, 12, &aom_highbd_subtract_block_c),
+  make_tuple(128, 128, 12, &aom_highbd_subtract_block_sse2),
+  make_tuple(128, 128, 12, &aom_highbd_subtract_block_c)
+};
+
+INSTANTIATE_TEST_CASE_P(SSE2, AV1HBDSubtractBlockTest,
+                        ::testing::ValuesIn(kAV1HBDSubtractBlock_sse2));
 #endif  // HAVE_SSE2
 #endif  // CONFIG_HIGHBITDEPTH
 }  // namespace
diff --git a/third_party/aom/test/superframe_test.cc b/third_party/aom/test/superframe_test.cc
index a28d35d5f..c8f663e5c 100644
--- a/third_party/aom/test/superframe_test.cc
+++ b/third_party/aom/test/superframe_test.cc
@@ -10,6 +10,7 @@
 */
 
 #include <climits>
+#include <vector>
 #include "third_party/googletest/src/googletest/include/gtest/gtest.h"
 #include "test/codec_factory.h"
 #include "test/encode_test_driver.h"
@@ -25,11 +26,10 @@ const int kTileRows = 2;
 typedef std::tr1::tuple<libaom_test::TestMode, int, int> SuperframeTestParam;
 
 class SuperframeTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWithParam<SuperframeTestParam> {
+    : public ::libaom_test::CodecTestWithParam<SuperframeTestParam>,
+      public ::libaom_test::EncoderTest {
  protected:
-  SuperframeTest()
-      : EncoderTest(GET_PARAM(0)), modified_buf_(NULL), last_sf_pts_(0) {}
+  SuperframeTest() : EncoderTest(GET_PARAM(0)), last_sf_pts_(0) {}
   virtual ~SuperframeTest() {}
 
   virtual void SetUp() {
@@ -43,8 +43,6 @@ class SuperframeTest
     n_tile_rows_ = std::tr1::get<kTileRows>(input);
   }
 
-  virtual void TearDown() { delete[] modified_buf_; }
-
   virtual void PreEncodeFrameHook(libaom_test::VideoSource *video,
                                   libaom_test::Encoder *encoder) {
     if (video->frame() == 1) {
@@ -63,18 +61,18 @@ class SuperframeTest
     if (pkt->kind != AOM_CODEC_CX_FRAME_PKT) return pkt;
 
     const uint8_t *buffer = reinterpret_cast<uint8_t *>(pkt->data.frame.buf);
-    const uint8_t marker = buffer[pkt->data.frame.sz - 1];
+    const uint8_t marker = buffer[0];
     const int frames = (marker & 0x7) + 1;
     const int mag = ((marker >> 3) & 3) + 1;
     const unsigned int index_sz = 2 + mag * (frames - 1);
     if ((marker & 0xe0) == 0xc0 && pkt->data.frame.sz >= index_sz &&
-        buffer[pkt->data.frame.sz - index_sz] == marker) {
+        buffer[index_sz - 1] == marker) {
       // frame is a superframe. strip off the index.
-      if (modified_buf_) delete[] modified_buf_;
-      modified_buf_ = new uint8_t[pkt->data.frame.sz - index_sz];
-      memcpy(modified_buf_, pkt->data.frame.buf, pkt->data.frame.sz - index_sz);
+      modified_buf_.resize(pkt->data.frame.sz - index_sz);
+      memcpy(&modified_buf_[0], (uint8_t *)pkt->data.frame.buf + index_sz,
+             pkt->data.frame.sz - index_sz);
       modified_pkt_ = *pkt;
-      modified_pkt_.data.frame.buf = modified_buf_;
+      modified_pkt_.data.frame.buf = &modified_buf_[0];
       modified_pkt_.data.frame.sz -= index_sz;
 
       sf_count_++;
@@ -91,7 +89,7 @@ class SuperframeTest
   int sf_count_;
   int sf_count_max_;
   aom_codec_cx_pkt_t modified_pkt_;
-  uint8_t *modified_buf_;
+  std::vector<uint8_t> modified_buf_;
   aom_codec_pts_t last_sf_pts_;
 
  private:
@@ -102,7 +100,9 @@ class SuperframeTest
 TEST_P(SuperframeTest, TestSuperframeIndexIsOptional) {
   sf_count_max_ = 0;  // early exit on successful test.
   cfg_.g_lag_in_frames = 25;
-
+#if CONFIG_EXT_TILE
+  cfg_.large_scale_tile = 1;
+#endif  // CONFIG_EXT_TILE
   ::libaom_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
                                        30, 1, 0, 40);
   ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
diff --git a/third_party/aom/test/test.cmake b/third_party/aom/test/test.cmake
index d72a784ad..a02f9203f 100644
--- a/third_party/aom/test/test.cmake
+++ b/third_party/aom/test/test.cmake
@@ -11,6 +11,7 @@
 if (NOT AOM_TEST_TEST_CMAKE_)
 set(AOM_TEST_TEST_CMAKE_ 1)
 
+include(FindPythonInterp)
 include(ProcessorCount)
 
 include("${AOM_ROOT}/test/test_data_util.cmake")
@@ -25,7 +26,6 @@ set(AOM_UNIT_TEST_COMMON_SOURCES
     "${AOM_ROOT}/test/acm_random.h"
     "${AOM_ROOT}/test/clear_system_state.h"
     "${AOM_ROOT}/test/codec_factory.h"
-    "${AOM_ROOT}/test/convolve_test.cc"
     "${AOM_ROOT}/test/decode_test_driver.cc"
     "${AOM_ROOT}/test/decode_test_driver.h"
     "${AOM_ROOT}/test/function_equivalence_test.h"
@@ -35,25 +35,58 @@ set(AOM_UNIT_TEST_COMMON_SOURCES
     "${AOM_ROOT}/test/util.h"
     "${AOM_ROOT}/test/video_source.h")
 
-if (CONFIG_ACCOUNTING)
+if (NOT BUILD_SHARED_LIBS)
   set(AOM_UNIT_TEST_COMMON_SOURCES
       ${AOM_UNIT_TEST_COMMON_SOURCES}
-      "${AOM_ROOT}/test/accounting_test.cc")
-endif ()
+      "${AOM_ROOT}/test/convolve_test.cc"
+      "${AOM_ROOT}/test/simd_impl.h")
 
-if (CONFIG_ADAPT_SCAN)
-  set(AOM_UNIT_TEST_COMMON_SOURCES
-      ${AOM_UNIT_TEST_COMMON_SOURCES}
-      "${AOM_ROOT}/test/scan_test.cc")
-endif ()
-
-if (CONFIG_GLOBAL_MOTION OR CONFIG_WARPED_MOTION)
+  if (HAVE_NEON)
+    set(AOM_UNIT_TEST_COMMON_SOURCES
+        ${AOM_UNIT_TEST_COMMON_SOURCES}
+        "${AOM_ROOT}/test/simd_neon_test.cc")
+  endif ()
   if (HAVE_SSE2)
     set(AOM_UNIT_TEST_COMMON_SOURCES
         ${AOM_UNIT_TEST_COMMON_SOURCES}
-        "${AOM_ROOT}/test/warp_filter_test.cc"
-        "${AOM_ROOT}/test/warp_filter_test_util.cc"
-        "${AOM_ROOT}/test/warp_filter_test_util.h")
+        "${AOM_ROOT}/test/simd_sse2_test.cc")
+  endif ()
+  if (HAVE_SSSE3)
+    set(AOM_UNIT_TEST_COMMON_SOURCES
+        ${AOM_UNIT_TEST_COMMON_SOURCES}
+        "${AOM_ROOT}/test/simd_ssse3_test.cc")
+  endif ()
+  if (HAVE_SSE4)
+    set(AOM_UNIT_TEST_COMMON_SOURCES
+        ${AOM_UNIT_TEST_COMMON_SOURCES}
+        "${AOM_ROOT}/test/simd_sse4_test.cc")
+  endif ()
+  if (HAVE_AVX2)
+    set(AOM_UNIT_TEST_COMMON_SOURCES
+        ${AOM_UNIT_TEST_COMMON_SOURCES}
+        "${AOM_ROOT}/test/simd_avx2_test.cc")
+  endif ()
+
+  if (CONFIG_ACCOUNTING)
+    set(AOM_UNIT_TEST_COMMON_SOURCES
+        ${AOM_UNIT_TEST_COMMON_SOURCES}
+        "${AOM_ROOT}/test/accounting_test.cc")
+  endif ()
+
+  if (CONFIG_ADAPT_SCAN)
+    set(AOM_UNIT_TEST_COMMON_SOURCES
+        ${AOM_UNIT_TEST_COMMON_SOURCES}
+        "${AOM_ROOT}/test/scan_test.cc")
+  endif ()
+
+  if (CONFIG_GLOBAL_MOTION OR CONFIG_WARPED_MOTION)
+    if (HAVE_SSE2)
+      set(AOM_UNIT_TEST_COMMON_SOURCES
+          ${AOM_UNIT_TEST_COMMON_SOURCES}
+          "${AOM_ROOT}/test/warp_filter_test.cc"
+          "${AOM_ROOT}/test/warp_filter_test_util.cc"
+          "${AOM_ROOT}/test/warp_filter_test_util.h")
+    endif ()
   endif ()
 endif ()
 
@@ -65,18 +98,23 @@ set(AOM_UNIT_TEST_ENCODER_SOURCES
     "${AOM_ROOT}/test/altref_test.cc"
     "${AOM_ROOT}/test/aq_segment_test.cc"
     "${AOM_ROOT}/test/datarate_test.cc"
-    "${AOM_ROOT}/test/dct16x16_test.cc"
-    "${AOM_ROOT}/test/dct32x32_test.cc"
     "${AOM_ROOT}/test/encode_api_test.cc"
     "${AOM_ROOT}/test/encode_test_driver.cc"
     "${AOM_ROOT}/test/encode_test_driver.h"
     "${AOM_ROOT}/test/error_resilience_test.cc"
     "${AOM_ROOT}/test/i420_video_source.h"
-    "${AOM_ROOT}/test/sad_test.cc"
     "${AOM_ROOT}/test/y4m_test.cc"
     "${AOM_ROOT}/test/y4m_video_source.h"
     "${AOM_ROOT}/test/yuv_video_source.h")
 
+if (NOT BUILD_SHARED_LIBS)
+  set(AOM_UNIT_TEST_ENCODER_SOURCES
+      ${AOM_UNIT_TEST_ENCODER_SOURCES}
+      "${AOM_ROOT}/test/dct16x16_test.cc"
+      "${AOM_ROOT}/test/dct32x32_test.cc"
+      "${AOM_ROOT}/test/sad_test.cc")
+endif ()
+
 set(AOM_DECODE_PERF_TEST_SOURCES "${AOM_ROOT}/test/decode_perf_test.cc")
 set(AOM_ENCODE_PERF_TEST_SOURCES "${AOM_ROOT}/test/encode_perf_test.cc")
 set(AOM_UNIT_TEST_WEBM_SOURCES "${AOM_ROOT}/test/webm_video_source.h")
@@ -85,149 +123,186 @@ set(AOM_TEST_INTRA_PRED_SPEED_SOURCES
     "${AOM_CONFIG_DIR}/usage_exit.c"
     "${AOM_ROOT}/test/test_intra_pred_speed.cc")
 
-if (CONFIG_AV1)
-  set(AOM_UNIT_TEST_COMMON_SOURCES
-      ${AOM_UNIT_TEST_COMMON_SOURCES}
-      "${AOM_ROOT}/test/av1_convolve_optimz_test.cc"
-      "${AOM_ROOT}/test/av1_convolve_test.cc"
-      "${AOM_ROOT}/test/av1_txfm_test.cc"
-      "${AOM_ROOT}/test/av1_txfm_test.h"
-      "${AOM_ROOT}/test/intrapred_test.cc"
-      "${AOM_ROOT}/test/lpf_8_test.cc"
-      "${AOM_ROOT}/test/simd_cmp_impl.h")
-
-  if (CONFIG_CDEF)
+if (NOT BUILD_SHARED_LIBS)
+  if (CONFIG_AV1_DECODER OR CONFIG_AV1_ENCODER)
     set(AOM_UNIT_TEST_COMMON_SOURCES
         ${AOM_UNIT_TEST_COMMON_SOURCES}
-        "${AOM_ROOT}/test/clpf_test.cc")
-  endif ()
+        "${AOM_ROOT}/test/av1_convolve_optimz_test.cc"
+        "${AOM_ROOT}/test/av1_convolve_test.cc"
+        "${AOM_ROOT}/test/av1_txfm_test.cc"
+        "${AOM_ROOT}/test/av1_txfm_test.h"
+        "${AOM_ROOT}/test/intrapred_test.cc"
+        "${AOM_ROOT}/test/lpf_8_test.cc"
+        "${AOM_ROOT}/test/motion_vector_test.cc"
+        "${AOM_ROOT}/test/simd_cmp_impl.h")
+
+    if (CONFIG_CDEF)
+      set(AOM_UNIT_TEST_COMMON_SOURCES
+          ${AOM_UNIT_TEST_COMMON_SOURCES}
+          "${AOM_ROOT}/test/clpf_test.cc"
+          "${AOM_ROOT}/test/dering_test.cc")
+    endif ()
+
+    if (CONFIG_FILTER_INTRA)
+      if (HAVE_SSE4_1)
+        set(AOM_UNIT_TEST_COMMON_SOURCES
+            ${AOM_UNIT_TEST_COMMON_SOURCES}
+            "${AOM_ROOT}/test/filterintra_predictors_test.cc")
+      endif ()
+    endif ()
+
+    if (CONFIG_INTRABC)
+        set(AOM_UNIT_TEST_COMMON_SOURCES
+            ${AOM_UNIT_TEST_COMMON_SOURCES}
+            "${AOM_ROOT}/test/intrabc_test.cc")
+    endif ()
 
-  if (CONFIG_FILTER_INTRA)
-    if (HAVE_SSE4_1)
+    if (CONFIG_LOOP_RESTORATION)
       set(AOM_UNIT_TEST_COMMON_SOURCES
           ${AOM_UNIT_TEST_COMMON_SOURCES}
-          "${AOM_ROOT}/test/filterintra_predictors_test.cc")
+           "${AOM_ROOT}/test/hiprec_convolve_test.cc"
+            "${AOM_ROOT}/test/hiprec_convolve_test_util.cc"
+            "${AOM_ROOT}/test/hiprec_convolve_test_util.h"
+            "${AOM_ROOT}/test/selfguided_filter_test.cc")
     endif ()
-  endif ()
 
-  set(AOM_UNIT_TEST_COMMON_INTRIN_NEON
-    ${AOM_UNIT_TEST_COMMON_INTRIN_NEON}
-      "${AOM_ROOT}/test/simd_cmp_neon.cc"
-      "${AOM_ROOT}/test/simd_neon_test.cc")
-  set(AOM_UNIT_TEST_COMMON_INTRIN_SSE2
-      ${AOM_UNIT_TEST_COMMON_INTRIN_SSE2}
-      "${AOM_ROOT}/test/simd_cmp_sse2.cc")
-  set(AOM_UNIT_TEST_COMMON_INTRIN_SSSE3
-      ${AOM_UNIT_TEST_COMMON_INTRIN_SSSE3}
-      "${AOM_ROOT}/test/simd_cmp_ssse3.cc")
-  set(AOM_UNIT_TEST_COMMON_INTRIN_SSE4_1
-      ${AOM_UNIT_TEST_COMMON_INTRIN_SSE4_1}
-      "${AOM_ROOT}/test/simd_cmp_sse4.cc")
+    set(AOM_UNIT_TEST_COMMON_INTRIN_NEON
+        ${AOM_UNIT_TEST_COMMON_INTRIN_NEON}
+        "${AOM_ROOT}/test/simd_cmp_neon.cc")
+    set(AOM_UNIT_TEST_COMMON_INTRIN_SSE2
+        ${AOM_UNIT_TEST_COMMON_INTRIN_SSE2}
+        "${AOM_ROOT}/test/simd_cmp_sse2.cc")
+    set(AOM_UNIT_TEST_COMMON_INTRIN_SSSE3
+        ${AOM_UNIT_TEST_COMMON_INTRIN_SSSE3}
+        "${AOM_ROOT}/test/simd_cmp_ssse3.cc")
+    set(AOM_UNIT_TEST_COMMON_INTRIN_SSE4_1
+        ${AOM_UNIT_TEST_COMMON_INTRIN_SSE4_1}
+        "${AOM_ROOT}/test/simd_cmp_sse4.cc")
+    set(AOM_UNIT_TEST_COMMON_INTRIN_AVX2
+        ${AOM_UNIT_TEST_COMMON_INTRIN_AVX2}
+        "${AOM_ROOT}/test/simd_cmp_avx2.cc")
+  endif ()
 endif ()
 
 if (CONFIG_AV1_ENCODER)
   set(AOM_UNIT_TEST_ENCODER_SOURCES
       ${AOM_UNIT_TEST_ENCODER_SOURCES}
       "${AOM_ROOT}/test/active_map_test.cc"
-      "${AOM_ROOT}/test/arf_freq_test.cc"
-      "${AOM_ROOT}/test/av1_dct_test.cc"
-      "${AOM_ROOT}/test/av1_fht16x16_test.cc"
-      "${AOM_ROOT}/test/av1_fht32x32_test.cc"
-      "${AOM_ROOT}/test/av1_fht8x8_test.cc"
-      "${AOM_ROOT}/test/av1_inv_txfm_test.cc"
-      "${AOM_ROOT}/test/av1_fwd_txfm1d_test.cc"
-      "${AOM_ROOT}/test/av1_fwd_txfm2d_test.cc"
-      "${AOM_ROOT}/test/av1_inv_txfm1d_test.cc"
-      "${AOM_ROOT}/test/av1_inv_txfm2d_test.cc"
-      "${AOM_ROOT}/test/avg_test.cc"
-      "${AOM_ROOT}/test/blend_a64_mask_1d_test.cc"
-      "${AOM_ROOT}/test/blend_a64_mask_test.cc"
       "${AOM_ROOT}/test/borders_test.cc"
       "${AOM_ROOT}/test/cpu_speed_test.cc"
       "${AOM_ROOT}/test/end_to_end_test.cc"
-      "${AOM_ROOT}/test/error_block_test.cc"
-      "${AOM_ROOT}/test/fdct4x4_test.cc"
-      "${AOM_ROOT}/test/fdct8x8_test.cc"
       "${AOM_ROOT}/test/frame_size_tests.cc"
-      "${AOM_ROOT}/test/hadamard_test.cc"
-      "${AOM_ROOT}/test/lossless_test.cc"
-      "${AOM_ROOT}/test/minmax_test.cc"
-      "${AOM_ROOT}/test/subtract_test.cc"
-      "${AOM_ROOT}/test/sum_squares_test.cc"
-      "${AOM_ROOT}/test/variance_test.cc")
-
-  if (CONFIG_EXT_INTER)
-    set(AOM_UNIT_TEST_ENCODER_SOURCES
-        ${AOM_UNIT_TEST_ENCODER_SOURCES}
-        "${AOM_ROOT}/test/av1_wedge_utils_test.cc"
-        "${AOM_ROOT}/test/masked_sad_test.cc"
-        "${AOM_ROOT}/test/masked_variance_test.cc")
-  endif ()
+      "${AOM_ROOT}/test/lossless_test.cc")
 
-  if (CONFIG_EXT_TX)
+  if (NOT BUILD_SHARED_LIBS)
     set(AOM_UNIT_TEST_ENCODER_SOURCES
         ${AOM_UNIT_TEST_ENCODER_SOURCES}
-        "${AOM_ROOT}/test/av1_fht16x32_test.cc"
-        "${AOM_ROOT}/test/av1_fht16x8_test.cc"
-        "${AOM_ROOT}/test/av1_fht32x16_test.cc"
-        "${AOM_ROOT}/test/av1_fht4x4_test.cc"
-        "${AOM_ROOT}/test/av1_fht4x8_test.cc"
-        "${AOM_ROOT}/test/av1_fht8x16_test.cc"
-        "${AOM_ROOT}/test/av1_fht8x4_test.cc")
-        
-  endif ()
+        "${AOM_ROOT}/test/arf_freq_test.cc"
+        "${AOM_ROOT}/test/av1_dct_test.cc"
+        "${AOM_ROOT}/test/av1_fht16x16_test.cc"
+        "${AOM_ROOT}/test/av1_fht32x32_test.cc"
+        "${AOM_ROOT}/test/av1_fht8x8_test.cc"
+        "${AOM_ROOT}/test/av1_inv_txfm_test.cc"
+        "${AOM_ROOT}/test/av1_fwd_txfm1d_test.cc"
+        "${AOM_ROOT}/test/av1_fwd_txfm2d_test.cc"
+        "${AOM_ROOT}/test/av1_inv_txfm1d_test.cc"
+        "${AOM_ROOT}/test/av1_inv_txfm2d_test.cc"
+        "${AOM_ROOT}/test/avg_test.cc"
+        "${AOM_ROOT}/test/blend_a64_mask_1d_test.cc"
+        "${AOM_ROOT}/test/blend_a64_mask_test.cc"
+        "${AOM_ROOT}/test/error_block_test.cc"
+        "${AOM_ROOT}/test/fdct4x4_test.cc"
+        "${AOM_ROOT}/test/fdct8x8_test.cc"
+        "${AOM_ROOT}/test/hadamard_test.cc"
+        "${AOM_ROOT}/test/minmax_test.cc"
+        "${AOM_ROOT}/test/quantize_func_test.cc"
+        "${AOM_ROOT}/test/subtract_test.cc"
+        "${AOM_ROOT}/test/sum_squares_test.cc"
+        "${AOM_ROOT}/test/variance_test.cc")
+
+    if (CONFIG_CONVOLVE_ROUND)
+      set(AOM_UNIT_TEST_ENCODER_SOURCES
+          ${AOM_UNIT_TEST_ENCODER_SOURCES}
+          "${AOM_ROOT}/test/av1_convolve_2d_test.cc"
+          "${AOM_ROOT}/test/av1_convolve_2d_test_util.cc"
+          "${AOM_ROOT}/test/av1_convolve_2d_test_util.h"
+          "${AOM_ROOT}/test/convolve_round_test.cc")
+      endif ()
 
-  if (CONFIG_GLOBAL_MOTION)
-    set(AOM_UNIT_TEST_ENCODER_INTRIN_SSE4_1
-        ${AOM_UNIT_TEST_ENCODER_INTRIN_SSE4_1}
-        "${AOM_ROOT}/test/corner_match_test.cc")
-  endif ()
+    if (CONFIG_EXT_INTER)
+      set(AOM_UNIT_TEST_ENCODER_SOURCES
+          ${AOM_UNIT_TEST_ENCODER_SOURCES}
+          "${AOM_ROOT}/test/av1_wedge_utils_test.cc"
+          "${AOM_ROOT}/test/masked_sad_test.cc"
+          "${AOM_ROOT}/test/masked_variance_test.cc")
+    endif ()
 
-  if (CONFIG_MOTION_VAR)
-    set(AOM_UNIT_TEST_ENCODER_SOURCES
-        ${AOM_UNIT_TEST_ENCODER_SOURCES}
-        "${AOM_ROOT}/test/obmc_sad_test.cc"
-        "${AOM_ROOT}/test/obmc_variance_test.cc")
-  endif ()
+    if (CONFIG_EXT_TX)
+      set(AOM_UNIT_TEST_ENCODER_SOURCES
+          ${AOM_UNIT_TEST_ENCODER_SOURCES}
+          "${AOM_ROOT}/test/av1_fht16x32_test.cc"
+          "${AOM_ROOT}/test/av1_fht16x8_test.cc"
+          "${AOM_ROOT}/test/av1_fht32x16_test.cc"
+          "${AOM_ROOT}/test/av1_fht4x4_test.cc"
+          "${AOM_ROOT}/test/av1_fht4x8_test.cc"
+          "${AOM_ROOT}/test/av1_fht8x16_test.cc"
+          "${AOM_ROOT}/test/av1_fht8x4_test.cc")
+    endif ()
 
-  if (CONFIG_TX64X64)
-    set(AOM_UNIT_TEST_ENCODER_SOURCES
-        ${AOM_UNIT_TEST_ENCODER_SOURCES}
-        "${AOM_ROOT}/test/av1_fht64x64_test.cc")
+    if (CONFIG_GLOBAL_MOTION)
+      set(AOM_UNIT_TEST_ENCODER_INTRIN_SSE4_1
+          ${AOM_UNIT_TEST_ENCODER_INTRIN_SSE4_1}
+          "${AOM_ROOT}/test/corner_match_test.cc")
+    endif ()
+
+    if (CONFIG_MOTION_VAR)
+      set(AOM_UNIT_TEST_ENCODER_SOURCES
+          ${AOM_UNIT_TEST_ENCODER_SOURCES}
+          "${AOM_ROOT}/test/obmc_sad_test.cc"
+          "${AOM_ROOT}/test/obmc_variance_test.cc")
+    endif ()
+
+    if (CONFIG_TX64X64)
+      set(AOM_UNIT_TEST_ENCODER_SOURCES
+          ${AOM_UNIT_TEST_ENCODER_SOURCES}
+          "${AOM_ROOT}/test/av1_fht64x64_test.cc")
+    endif ()
   endif ()
 endif ()
 
-if (CONFIG_AV1_DECODER AND CONFIG_AV1_ENCODER)
-  set(AOM_UNIT_TEST_COMMON_SOURCES
-      ${AOM_UNIT_TEST_COMMON_SOURCES}
-      "${AOM_ROOT}/test/divu_small_test.cc"
-      "${AOM_ROOT}/test/ethread_test.cc"
-      "${AOM_ROOT}/test/idct8x8_test.cc"
-      "${AOM_ROOT}/test/partial_idct_test.cc"
-      "${AOM_ROOT}/test/superframe_test.cc"
-      "${AOM_ROOT}/test/binary_codes_test.cc"
-      "${AOM_ROOT}/test/tile_independence_test.cc")
-
-  if (CONFIG_ANS)
+if (NOT BUILD_SHARED_LIBS)
+  if (CONFIG_AV1_DECODER AND CONFIG_AV1_ENCODER)
     set(AOM_UNIT_TEST_COMMON_SOURCES
         ${AOM_UNIT_TEST_COMMON_SOURCES}
-        "${AOM_ROOT}/test/ans_codec_test.cc"
-        "${AOM_ROOT}/test/ans_test.cc")
-  else ()
-    set(AOM_UNIT_TEST_COMMON_SOURCES
-        ${AOM_UNIT_TEST_COMMON_SOURCES}
-        "${AOM_ROOT}/test/boolcoder_test.cc")
-  endif ()
+        "${AOM_ROOT}/test/binary_codes_test.cc"
+        "${AOM_ROOT}/test/divu_small_test.cc"
+        "${AOM_ROOT}/test/ethread_test.cc"
+        "${AOM_ROOT}/test/idct8x8_test.cc"
+        "${AOM_ROOT}/test/partial_idct_test.cc"
+        "${AOM_ROOT}/test/superframe_test.cc"
+        "${AOM_ROOT}/test/tile_independence_test.cc")
+
+    if (CONFIG_ANS)
+      set(AOM_UNIT_TEST_COMMON_SOURCES
+          ${AOM_UNIT_TEST_COMMON_SOURCES}
+          "${AOM_ROOT}/test/ans_codec_test.cc"
+          "${AOM_ROOT}/test/ans_test.cc")
+    else ()
+      set(AOM_UNIT_TEST_COMMON_SOURCES
+          ${AOM_UNIT_TEST_COMMON_SOURCES}
+          "${AOM_ROOT}/test/boolcoder_test.cc")
+    endif ()
 
-  if (CONFIG_EXT_TILE)
-    set(AOM_UNIT_TEST_COMMON_SOURCES
-        ${AOM_UNIT_TEST_COMMON_SOURCES}
-        "${AOM_ROOT}/test/av1_ext_tile_test.cc")
+    if (CONFIG_EXT_TILE)
+      set(AOM_UNIT_TEST_COMMON_SOURCES
+          ${AOM_UNIT_TEST_COMMON_SOURCES}
+          "${AOM_ROOT}/test/av1_ext_tile_test.cc")
+    endif ()
   endif ()
 endif ()
 
 if (CONFIG_HIGHBITDEPTH)
-  if (CONFIG_AV1_ENCODER)
+  if (CONFIG_AV1_ENCODER AND NOT BUILD_SHARED_LIBS)
     set(AOM_UNIT_TEST_COMMON_INTRIN_SSE4_1
         ${AOM_UNIT_TEST_COMMON_INTRIN_SSE4_1}
         "${AOM_ROOT}/test/av1_highbd_iht_test.cc"
@@ -242,6 +317,12 @@ if (CONFIG_HIGHBITDEPTH)
 endif ()
 
 if (CONFIG_UNIT_TESTS)
+  find_package(PythonInterp)
+  if (NOT PYTHONINTERP_FOUND)
+    message(WARNING "--- Unit tests disabled: Python not found.")
+    set(CONFIG_UNIT_TESTS 0)
+  endif ()
+
   if (MSVC)
     # Force static run time to avoid collisions with googletest.
     include("${AOM_ROOT}/build/cmake/msvc_runtime.cmake")
@@ -249,6 +330,11 @@ if (CONFIG_UNIT_TESTS)
   include_directories(
     "${AOM_ROOT}/third_party/googletest/src/googletest/src"
     "${AOM_ROOT}/third_party/googletest/src/googletest/include")
+
+  if (BUILD_SHARED_LIBS AND APPLE)
+    # Silence an RPATH warning.
+    set(CMAKE_MACOSX_RPATH 1)
+  endif ()
   add_subdirectory("${AOM_ROOT}/third_party/googletest/src/googletest"
                    EXCLUDE_FROM_ALL)
 
@@ -263,55 +349,57 @@ endif ()
 # exist before this function is called.
 function (setup_aom_test_targets)
   add_library(test_aom_common OBJECT ${AOM_UNIT_TEST_COMMON_SOURCES})
+  add_dependencies(test_aom_common aom)
 
   if (CONFIG_AV1_DECODER)
     add_library(test_aom_decoder OBJECT ${AOM_UNIT_TEST_DECODER_SOURCES})
+    add_dependencies(test_aom_decoder aom)
   endif ()
 
   if (CONFIG_AV1_ENCODER)
     add_library(test_aom_encoder OBJECT ${AOM_UNIT_TEST_ENCODER_SOURCES})
+    add_dependencies(test_aom_encoder aom)
   endif ()
 
-  set(AOM_LIB_TARGETS ${AOM_LIB_TARGETS} test_aom_common test_aom_decoder
-      test_aom_encoder PARENT_SCOPE)
-
   add_executable(test_libaom ${AOM_UNIT_TEST_WRAPPER_SOURCES}
                  $<TARGET_OBJECTS:aom_common_app_util>
                  $<TARGET_OBJECTS:test_aom_common>)
 
   if (CONFIG_AV1_DECODER)
-    target_sources(test_libaom PUBLIC
+    target_sources(test_libaom PRIVATE
                    $<TARGET_OBJECTS:aom_decoder_app_util>
                    $<TARGET_OBJECTS:test_aom_decoder>)
 
     if (CONFIG_DECODE_PERF_TESTS AND CONFIG_WEBM_IO)
-      target_sources(test_libaom PUBLIC ${AOM_DECODE_PERF_TEST_SOURCES})
+      target_sources(test_libaom PRIVATE ${AOM_DECODE_PERF_TEST_SOURCES})
     endif ()
   endif ()
 
   if (CONFIG_AV1_ENCODER)
-    target_sources(test_libaom PUBLIC
+    target_sources(test_libaom PRIVATE
                    $<TARGET_OBJECTS:test_aom_encoder>
                    $<TARGET_OBJECTS:aom_encoder_app_util>)
 
     if (CONFIG_ENCODE_PERF_TESTS)
-      target_sources(test_libaom PUBLIC ${AOM_ENCODE_PERF_TEST_SOURCES})
+      target_sources(test_libaom PRIVATE ${AOM_ENCODE_PERF_TEST_SOURCES})
     endif ()
 
-    add_executable(test_intra_pred_speed
-                   ${AOM_TEST_INTRA_PRED_SPEED_SOURCES}
-                   $<TARGET_OBJECTS:aom_common_app_util>)
-    target_link_libraries(test_intra_pred_speed ${AOM_LIB_LINK_TYPE} aom gtest)
+    if (NOT BUILD_SHARED_LIBS)
+      add_executable(test_intra_pred_speed
+                     ${AOM_TEST_INTRA_PRED_SPEED_SOURCES}
+                     $<TARGET_OBJECTS:aom_common_app_util>)
+      target_link_libraries(test_intra_pred_speed ${AOM_LIB_LINK_TYPE}
+                            aom gtest)
+    endif ()
   endif ()
 
   target_link_libraries(test_libaom ${AOM_LIB_LINK_TYPE} aom gtest)
 
   if (CONFIG_LIBYUV)
-    target_sources(test_libaom PUBLIC $<TARGET_OBJECTS:yuv>)
+    target_sources(test_libaom PRIVATE $<TARGET_OBJECTS:yuv>)
   endif ()
   if (CONFIG_WEBM_IO)
-    target_sources(test_libaom PUBLIC ${AOM_UNIT_TEST_WEBM_SOURCES}
-                   $<TARGET_OBJECTS:webm>)
+    target_sources(test_libaom PRIVATE $<TARGET_OBJECTS:webm>)
   endif ()
   if (HAVE_SSE2)
     add_intrinsics_source_to_target("-msse2" "test_libaom"
@@ -331,11 +419,26 @@ function (setup_aom_test_targets)
       endif ()
     endif ()
   endif ()
+  if (HAVE_AVX2)
+    add_intrinsics_source_to_target("-mavx2" "test_libaom"
+                                    "AOM_UNIT_TEST_COMMON_INTRIN_AVX2")
+  endif ()
   if (HAVE_NEON)
     add_intrinsics_source_to_target("${AOM_NEON_INTRIN_FLAG}" "test_libaom"
                                     "AOM_UNIT_TEST_COMMON_INTRIN_NEON")
   endif ()
 
+  if (NOT ENABLE_IDE_TEST_HOSTING)
+    if (MSVC OR XCODE)
+      # Skip creation of test data download and test run targets when generating
+      # for Visual Studio and Xcode unless the user explicitly requests IDE test
+      # hosting. This is done to make build cycles in the IDE tolerable when the
+      # IDE command for build project is used to build AOM. Default behavior in
+      # IDEs is to build all targets, and the test run takes hours.
+      return ()
+    endif ()
+  endif ()
+
   make_test_data_lists("${AOM_UNIT_TEST_DATA_LIST_FILE}"
                        test_files test_file_checksums)
   list(LENGTH test_files num_test_files)
@@ -366,8 +469,6 @@ function (setup_aom_test_targets)
     set(num_test_targets 10)
   endif ()
 
-  # TODO(tomfinegan): This needs some work for MSVC and Xcode. Executable suffix
-  # and config based executable output paths are the obvious issues.
   math(EXPR max_shard_index "${num_test_targets} - 1")
   foreach (shard_index RANGE ${max_shard_index})
     set(test_name "test_${shard_index}")
@@ -382,15 +483,6 @@ function (setup_aom_test_targets)
   endforeach ()
   add_custom_target(runtests)
   add_dependencies(runtests ${test_targets})
-
-  if (MSVC)
-    set_target_properties(${testdata_targets} PROPERTIES
-                          EXCLUDE_FROM_DEFAULT_BUILD TRUE)
-    set_target_properties(${test_targets} PROPERTIES
-                          EXCLUDE_FROM_DEFAULT_BUILD TRUE)
-    set_target_properties(testdata runtests PROPERTIES
-                          EXCLUDE_FROM_DEFAULT_BUILD TRUE)
-  endif ()
 endfunction ()
 
 endif ()  # AOM_TEST_TEST_CMAKE_
diff --git a/third_party/aom/test/test.mk b/third_party/aom/test/test.mk
index 45bb21286..4132e4f74 100644
--- a/third_party/aom/test/test.mk
+++ b/third_party/aom/test/test.mk
@@ -33,6 +33,7 @@ LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += altref_test.cc
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += aq_segment_test.cc
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += datarate_test.cc
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += encode_api_test.cc
+LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += coding_path_sync.cc
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += error_resilience_test.cc
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += i420_video_source.h
 #LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER)    += realtime_test.cc
@@ -167,6 +168,7 @@ ifneq ($(CONFIG_NEW_QUANT), yes)
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER) += quantize_func_test.cc
 endif
 endif
+LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER) += block_error_test.cc
 
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER) += av1_inv_txfm_test.cc
 LIBAOM_TEST_SRCS-$(CONFIG_AV1_ENCODER) += av1_dct_test.cc
@@ -243,6 +245,12 @@ LIBAOM_TEST_SRCS-$(HAVE_SSE2) += hiprec_convolve_test.cc
 LIBAOM_TEST_SRCS-$(HAVE_SSE2) += hiprec_convolve_test_util.cc
 LIBAOM_TEST_SRCS-$(HAVE_SSE4_1) += selfguided_filter_test.cc
 endif
+ifeq ($(CONFIG_CONVOLVE_ROUND),yes)
+LIBAOM_TEST_SRCS-$(HAVE_SSE2) += av1_convolve_2d_test_util.h
+LIBAOM_TEST_SRCS-$(HAVE_SSE2) += av1_convolve_2d_test.cc
+LIBAOM_TEST_SRCS-$(HAVE_SSE2) += av1_convolve_2d_test_util.cc
+LIBAOM_TEST_SRCS-yes          += convolve_round_test.cc
+endif
 
 ifeq ($(CONFIG_GLOBAL_MOTION)$(CONFIG_AV1_ENCODER),yesyes)
 LIBAOM_TEST_SRCS-$(HAVE_SSE4_1) += corner_match_test.cc
diff --git a/third_party/aom/test/test_data_download_worker.cmake b/third_party/aom/test/test_data_download_worker.cmake
index d7bf99edd..b252dd960 100644
--- a/third_party/aom/test/test_data_download_worker.cmake
+++ b/third_party/aom/test/test_data_download_worker.cmake
@@ -19,7 +19,10 @@ endif ()
 
 set(AOM_TEST_DATA_URL
     "https://storage.googleapis.com/downloads.webmproject.org/test_data/libvpx")
-set(AOM_TEST_DATA_PATH "$ENV{LIBAOM_TEST_DATA_PATH}")
+
+if (NOT AOM_TEST_DATA_PATH)
+  set(AOM_TEST_DATA_PATH "$ENV{LIBAOM_TEST_DATA_PATH}")
+endif ()
 
 if ("${AOM_TEST_DATA_PATH}" STREQUAL "")
   message(WARNING "Writing test data to ${AOM_CONFIG_DIR}, set "
diff --git a/third_party/aom/test/test_runner.cmake b/third_party/aom/test/test_runner.cmake
index 48ebaf570..a1f399642 100644
--- a/third_party/aom/test/test_runner.cmake
+++ b/third_party/aom/test/test_runner.cmake
@@ -17,4 +17,6 @@ endif ()
 
 set($ENV{GTEST_SHARD_INDEX} ${GTEST_SHARD_INDEX})
 set($ENV{GTEST_TOTAL_SHARDS} ${GTEST_TOTAL_SHARDS})
-execute_process(COMMAND ${TEST_LIBAOM})
+execute_process(COMMAND ${TEST_LIBAOM} RESULT_VARIABLE test_result)
+set(test_message "Test shard ${GTEST_SHARD_INDEX}/${GTEST_TOTAL_SHARDS} result")
+message("${test_message}: ${test_result}")
diff --git a/third_party/aom/test/tile_independence_test.cc b/third_party/aom/test/tile_independence_test.cc
index a29051f2f..832227fb8 100644
--- a/third_party/aom/test/tile_independence_test.cc
+++ b/third_party/aom/test/tile_independence_test.cc
@@ -22,8 +22,8 @@
 
 namespace {
 class TileIndependenceTest
-    : public ::libaom_test::EncoderTest,
-      public ::libaom_test::CodecTestWith2Params<int, int> {
+    : public ::libaom_test::CodecTestWith2Params<int, int>,
+      public ::libaom_test::EncoderTest {
  protected:
   TileIndependenceTest()
       : EncoderTest(GET_PARAM(0)), md5_fw_order_(), md5_inv_order_(),
@@ -33,11 +33,12 @@ class TileIndependenceTest
     cfg.w = 704;
     cfg.h = 144;
     cfg.threads = 1;
+    cfg.allow_lowbitdepth = 1;
     fw_dec_ = codec_->CreateDecoder(cfg, 0);
     inv_dec_ = codec_->CreateDecoder(cfg, 0);
     inv_dec_->Control(AV1_INVERT_TILE_DECODE_ORDER, 1);
 
-#if CONFIG_AV1 && CONFIG_EXT_TILE
+#if CONFIG_AV1
     if (fw_dec_->IsAV1() && inv_dec_->IsAV1()) {
       fw_dec_->Control(AV1_SET_DECODE_TILE_ROW, -1);
       fw_dec_->Control(AV1_SET_DECODE_TILE_COL, -1);
@@ -62,9 +63,6 @@ class TileIndependenceTest
     if (video->frame() == 1) {
       encoder->Control(AV1E_SET_TILE_COLUMNS, n_tile_cols_);
       encoder->Control(AV1E_SET_TILE_ROWS, n_tile_rows_);
-#if CONFIG_EXT_TILE
-      encoder->Control(AV1E_SET_TILE_ENCODING_MODE, 0);  // TILE_NORMAL
-#endif                                                   // CONFIG_EXT_TILE
 #if CONFIG_LOOPFILTERING_ACROSS_TILES
       encoder->Control(AV1E_SET_TILE_LOOPFILTER, 0);
 #endif  // CONFIG_LOOPFILTERING_ACROSS_TILES
@@ -121,7 +119,12 @@ class TileIndependenceTest
 // run an encode with 2 or 4 tiles, and do the decode both in normal and
 // inverted tile ordering. Ensure that the MD5 of the output in both cases
 // is identical. If so, tiles are considered independent and the test passes.
-TEST_P(TileIndependenceTest, MD5Match) { DoTest(); }
+TEST_P(TileIndependenceTest, MD5Match) {
+#if CONFIG_EXT_TILE
+  cfg_.large_scale_tile = 0;
+#endif  // CONFIG_EXT_TILE
+  DoTest();
+}
 
 class TileIndependenceTestLarge : public TileIndependenceTest {
   virtual void SetCpuUsed(libaom_test::Encoder *encoder) {
@@ -130,18 +133,37 @@ class TileIndependenceTestLarge : public TileIndependenceTest {
   }
 };
 
-TEST_P(TileIndependenceTestLarge, MD5Match) { DoTest(); }
-
+TEST_P(TileIndependenceTestLarge, MD5Match) {
 #if CONFIG_EXT_TILE
-AV1_INSTANTIATE_TEST_CASE(TileIndependenceTest, ::testing::Values(1, 2, 32),
-                          ::testing::Values(1, 2, 32));
-AV1_INSTANTIATE_TEST_CASE(TileIndependenceTestLarge,
-                          ::testing::Values(1, 2, 32),
-                          ::testing::Values(1, 2, 32));
-#else
+  cfg_.large_scale_tile = 0;
+#endif  // CONFIG_EXT_TILE
+  DoTest();
+}
+
 AV1_INSTANTIATE_TEST_CASE(TileIndependenceTest, ::testing::Values(0, 1),
                           ::testing::Values(0, 1));
 AV1_INSTANTIATE_TEST_CASE(TileIndependenceTestLarge, ::testing::Values(0, 1),
                           ::testing::Values(0, 1));
+
+#if CONFIG_EXT_TILE
+class TileIndependenceLSTest : public TileIndependenceTest {};
+
+TEST_P(TileIndependenceLSTest, MD5Match) {
+  cfg_.large_scale_tile = 1;
+  DoTest();
+}
+
+class TileIndependenceLSTestLarge : public TileIndependenceTestLarge {};
+
+TEST_P(TileIndependenceLSTestLarge, MD5Match) {
+  cfg_.large_scale_tile = 1;
+  DoTest();
+}
+
+AV1_INSTANTIATE_TEST_CASE(TileIndependenceLSTest, ::testing::Values(1, 2, 32),
+                          ::testing::Values(1, 2, 32));
+AV1_INSTANTIATE_TEST_CASE(TileIndependenceLSTestLarge,
+                          ::testing::Values(1, 2, 32),
+                          ::testing::Values(1, 2, 32));
 #endif  // CONFIG_EXT_TILE
 }  // namespace
diff --git a/third_party/aom/test/tools_common.sh b/third_party/aom/test/tools_common.sh
index 254e6b296..c1262b6c8 100755
--- a/third_party/aom/test/tools_common.sh
+++ b/third_party/aom/test/tools_common.sh
@@ -48,6 +48,21 @@ test_end() {
 # Echoes the target configuration being tested.
 test_configuration_target() {
   aom_config_mk="${LIBAOM_CONFIG_PATH}/config.mk"
+  # TODO(tomfinegan): Remove the parts requiring config.mk when the configure
+  # script is removed from the repository.
+  if [ ! -f "${aom_config_mk}" ]; then
+    aom_config_c="${LIBAOM_CONFIG_PATH}/aom_config.c"
+    # Clean up the cfg pointer line from aom_config.c for easier re-use by
+    # someone examining a failure in the example tests.
+    # 1. Run grep on aom_config.c for cfg and limit the results to 1.
+    # 2. Split the line using ' = ' as separator.
+    # 3. Abuse sed to consume the leading " and trailing "; from the assignment
+    #    to the cfg pointer.
+    cmake_config=$(awk -F ' = ' '/cfg/ { print $NF; exit }' "${aom_config_c}" \
+      | sed -e s/\"// -e s/\"\;//)
+    echo cmake generated via command: cmake path/to/aom ${cmake_config}
+    return
+  fi
   # Find the TOOLCHAIN line, split it using ':=' as the field separator, and
   # print the last field to get the value. Then pipe the value to tr to consume
   # any leading/trailing spaces while allowing tr to echo the output to stdout.
@@ -71,6 +86,18 @@ cleanup() {
 # no git hash is contained in VERSION_STRING.
 config_hash() {
   aom_config_mk="${LIBAOM_CONFIG_PATH}/config.mk"
+  if [ ! -f "${aom_config_mk}" ]; then
+    aom_config_c="${LIBAOM_CONFIG_PATH}/aom_config.c"
+    # Clean up the aom_git_hash pointer line from aom_config.c.
+    # 1. Run grep on aom_config.c for aom_git_hash and limit results to 1.
+    # 2. Split the line using ' = "' as separator.
+    # 3. Abuse sed to consume the trailing "; from the assignment to the
+    #    aom_git_hash pointer.
+    awk -F ' = "' '/aom_git_hash/ { print $NF; exit }' "${aom_config_c}" \
+      | sed s/\"\;//
+    return
+  fi
+
   # Find VERSION_STRING line, split it with "-g" and print the last field to
   # output the git hash to stdout.
   aom_version=$(awk -F -g '/VERSION_STRING/ {print $NF}' "${aom_config_mk}")
@@ -87,7 +114,7 @@ config_hash() {
 current_hash() {
   if git --version > /dev/null 2>&1; then
     (cd "$(dirname "${0}")"
-    git rev-parse --short HEAD)
+    git rev-parse HEAD)
   else
     # Return the config hash if git is unavailable: Fail silently, git hashes
     # are used only for warnings.
@@ -103,6 +130,7 @@ check_git_hashes() {
 
   if [ "${hash_at_configure_time}" != "${hash_now}" ]; then
     echo "Warning: git hash has changed since last configure."
+    vlog "  config hash: ${hash_at_configure_time} hash now: ${hash_now}"
   fi
 }
 
diff --git a/third_party/aom/test/transform_test_base.h b/third_party/aom/test/transform_test_base.h
index 4c1a55496..21441beef 100644
--- a/third_party/aom/test/transform_test_base.h
+++ b/third_party/aom/test/transform_test_base.h
@@ -15,6 +15,7 @@
 #include "./aom_config.h"
 #include "aom_mem/aom_mem.h"
 #include "aom/aom_codec.h"
+#include "aom_dsp/txfm_common.h"
 
 namespace libaom_test {
 
@@ -28,10 +29,10 @@ namespace libaom_test {
 const int kDctMaxValue = 16384;
 
 typedef void (*FhtFunc)(const int16_t *in, tran_low_t *out, int stride,
-                        int tx_type);
+                        TxfmParam *txfm_param);
 
 typedef void (*IhtFunc)(const tran_low_t *in, uint8_t *out, int stride,
-                        int tx_type);
+                        const TxfmParam *txfm_param);
 
 class TransformTestBase {
  public:
@@ -157,7 +158,7 @@ class TransformTestBase {
         }
       }
 
-      fwd_txfm_ref(input_block, output_ref_block, stride, tx_type_);
+      fwd_txfm_ref(input_block, output_ref_block, stride, &txfm_param_);
       ASM_REGISTER_STATE_CHECK(RunFwdTxfm(input_block, output_block, stride));
 
       // The minimum quant value is 4.
@@ -205,9 +206,9 @@ class TransformTestBase {
         }
       }
 
-      fwd_txfm_ref(input_block, trans_block, pitch_, tx_type_);
+      fwd_txfm_ref(input_block, trans_block, pitch_, &txfm_param_);
 
-      inv_txfm_ref(trans_block, output_ref_block, stride, tx_type_);
+      inv_txfm_ref(trans_block, output_ref_block, stride, &txfm_param_);
       ASM_REGISTER_STATE_CHECK(RunInvTxfm(trans_block, output_block, stride));
 
       for (j = 0; j < height_; ++j) {
@@ -247,7 +248,7 @@ class TransformTestBase {
         for (int j = 0; j < num_coeffs_; ++j) input_extreme_block[j] = -mask_;
       }
 
-      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, tx_type_);
+      fwd_txfm_ref(input_extreme_block, output_ref_block, pitch_, &txfm_param_);
       ASM_REGISTER_STATE_CHECK(
           RunFwdTxfm(input_extreme_block, output_block, pitch_));
 
@@ -303,7 +304,7 @@ class TransformTestBase {
         }
       }
 
-      fwd_txfm_ref(in, coeff, pitch_, tx_type_);
+      fwd_txfm_ref(in, coeff, pitch_, &txfm_param_);
 
       if (bit_depth_ == AOM_BITS_8) {
         ASM_REGISTER_STATE_CHECK(RunInvTxfm(coeff, dst, pitch_));
@@ -338,12 +339,12 @@ class TransformTestBase {
 
   int pitch_;
   int height_;
-  int tx_type_;
   FhtFunc fwd_txfm_ref;
   IhtFunc inv_txfm_ref;
   aom_bit_depth_t bit_depth_;
   int mask_;
   int num_coeffs_;
+  TxfmParam txfm_param_;
 
  private:
   //  Assume transform size is 4x4, 8x8, 16x16,...
diff --git a/third_party/aom/test/user_priv_test.cc b/third_party/aom/test/user_priv_test.cc
index 3052b27b1..da289c990 100644
--- a/third_party/aom/test/user_priv_test.cc
+++ b/third_party/aom/test/user_priv_test.cc
@@ -49,6 +49,7 @@ string DecodeFile(const string &filename) {
   video.Init();
 
   aom_codec_dec_cfg_t cfg = aom_codec_dec_cfg_t();
+  cfg.allow_lowbitdepth = 1;
   libaom_test::AV1Decoder decoder(cfg, 0);
 
   libaom_test::MD5 md5;
diff --git a/third_party/aom/test/util.h b/third_party/aom/test/util.h
index a20fab65c..d1587b6ea 100644
--- a/third_party/aom/test/util.h
+++ b/third_party/aom/test/util.h
@@ -15,6 +15,7 @@
 #include <stdio.h>
 #include <math.h>
 #include "third_party/googletest/src/googletest/include/gtest/gtest.h"
+#include "aom/aom_integer.h"
 #include "aom/aom_image.h"
 
 // Macros
diff --git a/third_party/aom/test/variance_test.cc b/third_party/aom/test/variance_test.cc
index c712f96e2..2f5c22283 100644
--- a/third_party/aom/test/variance_test.cc
+++ b/third_party/aom/test/variance_test.cc
@@ -531,42 +531,26 @@ using ::std::tr1::get;
 using ::std::tr1::make_tuple;
 using ::std::tr1::tuple;
 
-template <typename SubpelVarianceFunctionType>
+template <typename FunctionType>
 class SubpelVarianceTest
-    : public ::testing::TestWithParam<
-          tuple<int, int, SubpelVarianceFunctionType, int> > {
+    : public ::testing::TestWithParam<TestParams<FunctionType> > {
  public:
   virtual void SetUp() {
-    const tuple<int, int, SubpelVarianceFunctionType, int> &params =
-        this->GetParam();
-    log2width_ = get<0>(params);
-    width_ = 1 << log2width_;
-    log2height_ = get<1>(params);
-    height_ = 1 << log2height_;
-    subpel_variance_ = get<2>(params);
-    if (get<3>(params)) {
-      bit_depth_ = (aom_bit_depth_t)get<3>(params);
-      use_high_bit_depth_ = true;
-    } else {
-      bit_depth_ = AOM_BITS_8;
-      use_high_bit_depth_ = false;
-    }
-    mask_ = (1 << bit_depth_) - 1;
+    params_ = this->GetParam();
 
     rnd_.Reset(ACMRandom::DeterministicSeed());
-    block_size_ = width_ * height_;
-    if (!use_high_bit_depth_) {
-      src_ = reinterpret_cast<uint8_t *>(aom_memalign(16, block_size_));
-      sec_ = reinterpret_cast<uint8_t *>(aom_memalign(16, block_size_));
-      ref_ = new uint8_t[block_size_ + width_ + height_ + 1];
+    if (!use_high_bit_depth()) {
+      src_ = reinterpret_cast<uint8_t *>(aom_memalign(16, block_size()));
+      sec_ = reinterpret_cast<uint8_t *>(aom_memalign(16, block_size()));
+      ref_ = new uint8_t[block_size() + width() + height() + 1];
 #if CONFIG_HIGHBITDEPTH
     } else {
       src_ = CONVERT_TO_BYTEPTR(reinterpret_cast<uint16_t *>(
-          aom_memalign(16, block_size_ * sizeof(uint16_t))));
+          aom_memalign(16, block_size() * sizeof(uint16_t))));
       sec_ = CONVERT_TO_BYTEPTR(reinterpret_cast<uint16_t *>(
-          aom_memalign(16, block_size_ * sizeof(uint16_t))));
+          aom_memalign(16, block_size() * sizeof(uint16_t))));
       ref_ = CONVERT_TO_BYTEPTR(aom_memalign(
-          16, (block_size_ + width_ + height_ + 1) * sizeof(uint16_t)));
+          16, (block_size() + width() + height() + 1) * sizeof(uint16_t)));
 #endif  // CONFIG_HIGHBITDEPTH
     }
     ASSERT_TRUE(src_ != NULL);
@@ -575,7 +559,7 @@ class SubpelVarianceTest
   }
 
   virtual void TearDown() {
-    if (!use_high_bit_depth_) {
+    if (!use_high_bit_depth()) {
       aom_free(src_);
       delete[] ref_;
       aom_free(sec_);
@@ -597,42 +581,45 @@ class SubpelVarianceTest
   uint8_t *src_;
   uint8_t *ref_;
   uint8_t *sec_;
-  bool use_high_bit_depth_;
-  aom_bit_depth_t bit_depth_;
-  int width_, log2width_;
-  int height_, log2height_;
-  int block_size_, mask_;
-  SubpelVarianceFunctionType subpel_variance_;
+  TestParams<FunctionType> params_;
+
+  // some relay helpers
+  bool use_high_bit_depth() const { return params_.use_high_bit_depth; }
+  int byte_shift() const { return params_.bit_depth - 8; }
+  int block_size() const { return params_.block_size; }
+  int width() const { return params_.width; }
+  int height() const { return params_.height; }
+  uint32_t mask() const { return params_.mask; }
 };
 
 template <typename SubpelVarianceFunctionType>
 void SubpelVarianceTest<SubpelVarianceFunctionType>::RefTest() {
   for (int x = 0; x < 8; ++x) {
     for (int y = 0; y < 8; ++y) {
-      if (!use_high_bit_depth_) {
-        for (int j = 0; j < block_size_; j++) {
+      if (!use_high_bit_depth()) {
+        for (int j = 0; j < block_size(); j++) {
           src_[j] = rnd_.Rand8();
         }
-        for (int j = 0; j < block_size_ + width_ + height_ + 1; j++) {
+        for (int j = 0; j < block_size() + width() + height() + 1; j++) {
           ref_[j] = rnd_.Rand8();
         }
 #if CONFIG_HIGHBITDEPTH
       } else {
-        for (int j = 0; j < block_size_; j++) {
-          CONVERT_TO_SHORTPTR(src_)[j] = rnd_.Rand16() & mask_;
+        for (int j = 0; j < block_size(); j++) {
+          CONVERT_TO_SHORTPTR(src_)[j] = rnd_.Rand16() & mask();
         }
-        for (int j = 0; j < block_size_ + width_ + height_ + 1; j++) {
-          CONVERT_TO_SHORTPTR(ref_)[j] = rnd_.Rand16() & mask_;
+        for (int j = 0; j < block_size() + width() + height() + 1; j++) {
+          CONVERT_TO_SHORTPTR(ref_)[j] = rnd_.Rand16() & mask();
         }
 #endif  // CONFIG_HIGHBITDEPTH
       }
       unsigned int sse1, sse2;
       unsigned int var1;
       ASM_REGISTER_STATE_CHECK(
-          var1 = subpel_variance_(ref_, width_ + 1, x, y, src_, width_, &sse1));
-      const unsigned int var2 =
-          subpel_variance_ref(ref_, src_, log2width_, log2height_, x, y, &sse2,
-                              use_high_bit_depth_, bit_depth_);
+          var1 = params_.func(ref_, width() + 1, x, y, src_, width(), &sse1));
+      const unsigned int var2 = subpel_variance_ref(
+          ref_, src_, params_.log2width, params_.log2height, x, y, &sse2,
+          use_high_bit_depth(), params_.bit_depth);
       EXPECT_EQ(sse1, sse2) << "at position " << x << ", " << y;
       EXPECT_EQ(var1, var2) << "at position " << x << ", " << y;
     }
@@ -646,28 +633,28 @@ void SubpelVarianceTest<SubpelVarianceFunctionType>::ExtremeRefTest() {
   // Ref: Set the first half of values to the maximum, the second half to 0.
   for (int x = 0; x < 8; ++x) {
     for (int y = 0; y < 8; ++y) {
-      const int half = block_size_ / 2;
-      if (!use_high_bit_depth_) {
+      const int half = block_size() / 2;
+      if (!use_high_bit_depth()) {
         memset(src_, 0, half);
         memset(src_ + half, 255, half);
         memset(ref_, 255, half);
-        memset(ref_ + half, 0, half + width_ + height_ + 1);
+        memset(ref_ + half, 0, half + width() + height() + 1);
 #if CONFIG_HIGHBITDEPTH
       } else {
-        aom_memset16(CONVERT_TO_SHORTPTR(src_), mask_, half);
+        aom_memset16(CONVERT_TO_SHORTPTR(src_), mask(), half);
         aom_memset16(CONVERT_TO_SHORTPTR(src_) + half, 0, half);
         aom_memset16(CONVERT_TO_SHORTPTR(ref_), 0, half);
-        aom_memset16(CONVERT_TO_SHORTPTR(ref_) + half, mask_,
-                     half + width_ + height_ + 1);
+        aom_memset16(CONVERT_TO_SHORTPTR(ref_) + half, mask(),
+                     half + width() + height() + 1);
 #endif  // CONFIG_HIGHBITDEPTH
       }
       unsigned int sse1, sse2;
       unsigned int var1;
       ASM_REGISTER_STATE_CHECK(
-          var1 = subpel_variance_(ref_, width_ + 1, x, y, src_, width_, &sse1));
-      const unsigned int var2 =
-          subpel_variance_ref(ref_, src_, log2width_, log2height_, x, y, &sse2,
-                              use_high_bit_depth_, bit_depth_);
+          var1 = params_.func(ref_, width() + 1, x, y, src_, width(), &sse1));
+      const unsigned int var2 = subpel_variance_ref(
+          ref_, src_, params_.log2width, params_.log2height, x, y, &sse2,
+          use_high_bit_depth(), params_.bit_depth);
       EXPECT_EQ(sse1, sse2) << "for xoffset " << x << " and yoffset " << y;
       EXPECT_EQ(var1, var2) << "for xoffset " << x << " and yoffset " << y;
     }
@@ -678,33 +665,32 @@ template <>
 void SubpelVarianceTest<SubpixAvgVarMxNFunc>::RefTest() {
   for (int x = 0; x < 8; ++x) {
     for (int y = 0; y < 8; ++y) {
-      if (!use_high_bit_depth_) {
-        for (int j = 0; j < block_size_; j++) {
+      if (!use_high_bit_depth()) {
+        for (int j = 0; j < block_size(); j++) {
           src_[j] = rnd_.Rand8();
           sec_[j] = rnd_.Rand8();
         }
-        for (int j = 0; j < block_size_ + width_ + height_ + 1; j++) {
+        for (int j = 0; j < block_size() + width() + height() + 1; j++) {
           ref_[j] = rnd_.Rand8();
         }
 #if CONFIG_HIGHBITDEPTH
       } else {
-        for (int j = 0; j < block_size_; j++) {
-          CONVERT_TO_SHORTPTR(src_)[j] = rnd_.Rand16() & mask_;
-          CONVERT_TO_SHORTPTR(sec_)[j] = rnd_.Rand16() & mask_;
+        for (int j = 0; j < block_size(); j++) {
+          CONVERT_TO_SHORTPTR(src_)[j] = rnd_.Rand16() & mask();
+          CONVERT_TO_SHORTPTR(sec_)[j] = rnd_.Rand16() & mask();
         }
-        for (int j = 0; j < block_size_ + width_ + height_ + 1; j++) {
-          CONVERT_TO_SHORTPTR(ref_)[j] = rnd_.Rand16() & mask_;
+        for (int j = 0; j < block_size() + width() + height() + 1; j++) {
+          CONVERT_TO_SHORTPTR(ref_)[j] = rnd_.Rand16() & mask();
         }
 #endif  // CONFIG_HIGHBITDEPTH
       }
       uint32_t sse1, sse2;
       uint32_t var1, var2;
-      ASM_REGISTER_STATE_CHECK(var1 =
-                                   subpel_variance_(ref_, width_ + 1, x, y,
-                                                    src_, width_, &sse1, sec_));
-      var2 = subpel_avg_variance_ref(ref_, src_, sec_, log2width_, log2height_,
-                                     x, y, &sse2, use_high_bit_depth_,
-                                     static_cast<aom_bit_depth_t>(bit_depth_));
+      ASM_REGISTER_STATE_CHECK(var1 = params_.func(ref_, width() + 1, x, y,
+                                                   src_, width(), &sse1, sec_));
+      var2 = subpel_avg_variance_ref(ref_, src_, sec_, params_.log2width,
+                                     params_.log2height, x, y, &sse2,
+                                     use_high_bit_depth(), params_.bit_depth);
       EXPECT_EQ(sse1, sse2) << "at position " << x << ", " << y;
       EXPECT_EQ(var1, var2) << "at position " << x << ", " << y;
     }
@@ -763,37 +749,41 @@ INSTANTIATE_TEST_CASE_P(
                       VarianceParams(2, 3, &aom_variance4x8_c),
                       VarianceParams(2, 2, &aom_variance4x4_c)));
 
+typedef TestParams<SubpixVarMxNFunc> SubpelVarianceParams;
 INSTANTIATE_TEST_CASE_P(
     C, AvxSubpelVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_variance64x64_c, 0),
-                      make_tuple(6, 5, &aom_sub_pixel_variance64x32_c, 0),
-                      make_tuple(5, 6, &aom_sub_pixel_variance32x64_c, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_variance32x32_c, 0),
-                      make_tuple(5, 4, &aom_sub_pixel_variance32x16_c, 0),
-                      make_tuple(4, 5, &aom_sub_pixel_variance16x32_c, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_variance16x16_c, 0),
-                      make_tuple(4, 3, &aom_sub_pixel_variance16x8_c, 0),
-                      make_tuple(3, 4, &aom_sub_pixel_variance8x16_c, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_variance8x8_c, 0),
-                      make_tuple(3, 2, &aom_sub_pixel_variance8x4_c, 0),
-                      make_tuple(2, 3, &aom_sub_pixel_variance4x8_c, 0),
-                      make_tuple(2, 2, &aom_sub_pixel_variance4x4_c, 0)));
-
+    ::testing::Values(
+        SubpelVarianceParams(6, 6, &aom_sub_pixel_variance64x64_c, 0),
+        SubpelVarianceParams(6, 5, &aom_sub_pixel_variance64x32_c, 0),
+        SubpelVarianceParams(5, 6, &aom_sub_pixel_variance32x64_c, 0),
+        SubpelVarianceParams(5, 5, &aom_sub_pixel_variance32x32_c, 0),
+        SubpelVarianceParams(5, 4, &aom_sub_pixel_variance32x16_c, 0),
+        SubpelVarianceParams(4, 5, &aom_sub_pixel_variance16x32_c, 0),
+        SubpelVarianceParams(4, 4, &aom_sub_pixel_variance16x16_c, 0),
+        SubpelVarianceParams(4, 3, &aom_sub_pixel_variance16x8_c, 0),
+        SubpelVarianceParams(3, 4, &aom_sub_pixel_variance8x16_c, 0),
+        SubpelVarianceParams(3, 3, &aom_sub_pixel_variance8x8_c, 0),
+        SubpelVarianceParams(3, 2, &aom_sub_pixel_variance8x4_c, 0),
+        SubpelVarianceParams(2, 3, &aom_sub_pixel_variance4x8_c, 0),
+        SubpelVarianceParams(2, 2, &aom_sub_pixel_variance4x4_c, 0)));
+
+typedef TestParams<SubpixAvgVarMxNFunc> SubpelAvgVarianceParams;
 INSTANTIATE_TEST_CASE_P(
     C, AvxSubpelAvgVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_avg_variance64x64_c, 0),
-                      make_tuple(6, 5, &aom_sub_pixel_avg_variance64x32_c, 0),
-                      make_tuple(5, 6, &aom_sub_pixel_avg_variance32x64_c, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_avg_variance32x32_c, 0),
-                      make_tuple(5, 4, &aom_sub_pixel_avg_variance32x16_c, 0),
-                      make_tuple(4, 5, &aom_sub_pixel_avg_variance16x32_c, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_avg_variance16x16_c, 0),
-                      make_tuple(4, 3, &aom_sub_pixel_avg_variance16x8_c, 0),
-                      make_tuple(3, 4, &aom_sub_pixel_avg_variance8x16_c, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_avg_variance8x8_c, 0),
-                      make_tuple(3, 2, &aom_sub_pixel_avg_variance8x4_c, 0),
-                      make_tuple(2, 3, &aom_sub_pixel_avg_variance4x8_c, 0),
-                      make_tuple(2, 2, &aom_sub_pixel_avg_variance4x4_c, 0)));
+    ::testing::Values(
+        SubpelAvgVarianceParams(6, 6, &aom_sub_pixel_avg_variance64x64_c, 0),
+        SubpelAvgVarianceParams(6, 5, &aom_sub_pixel_avg_variance64x32_c, 0),
+        SubpelAvgVarianceParams(5, 6, &aom_sub_pixel_avg_variance32x64_c, 0),
+        SubpelAvgVarianceParams(5, 5, &aom_sub_pixel_avg_variance32x32_c, 0),
+        SubpelAvgVarianceParams(5, 4, &aom_sub_pixel_avg_variance32x16_c, 0),
+        SubpelAvgVarianceParams(4, 5, &aom_sub_pixel_avg_variance16x32_c, 0),
+        SubpelAvgVarianceParams(4, 4, &aom_sub_pixel_avg_variance16x16_c, 0),
+        SubpelAvgVarianceParams(4, 3, &aom_sub_pixel_avg_variance16x8_c, 0),
+        SubpelAvgVarianceParams(3, 4, &aom_sub_pixel_avg_variance8x16_c, 0),
+        SubpelAvgVarianceParams(3, 3, &aom_sub_pixel_avg_variance8x8_c, 0),
+        SubpelAvgVarianceParams(3, 2, &aom_sub_pixel_avg_variance8x4_c, 0),
+        SubpelAvgVarianceParams(2, 3, &aom_sub_pixel_avg_variance4x8_c, 0),
+        SubpelAvgVarianceParams(2, 2, &aom_sub_pixel_avg_variance4x4_c, 0)));
 
 #if CONFIG_HIGHBITDEPTH
 typedef MainTestClass<VarianceMxNFunc> AvxHBDMseTest;
@@ -896,120 +886,147 @@ INSTANTIATE_TEST_CASE_P(
         VarianceParams(2, 2, &aom_highbd_12_variance4x4_sse4_1, 12)));
 #endif  // HAVE_SSE4_1 && CONFIG_HIGHBITDEPTH
 
-const AvxHBDSubpelVarianceTest::ParamType kArrayHBDSubpelVariance_c[] = {
+const SubpelVarianceParams kArrayHBDSubpelVariance_c[] = {
 #if CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(7, 7, &aom_highbd_8_sub_pixel_variance128x128_c, 8),
-  make_tuple(7, 6, &aom_highbd_8_sub_pixel_variance128x64_c, 8),
-  make_tuple(6, 7, &aom_highbd_8_sub_pixel_variance64x128_c, 8),
+  SubpelVarianceParams(7, 7, &aom_highbd_8_sub_pixel_variance128x128_c, 8),
+  SubpelVarianceParams(7, 6, &aom_highbd_8_sub_pixel_variance128x64_c, 8),
+  SubpelVarianceParams(6, 7, &aom_highbd_8_sub_pixel_variance64x128_c, 8),
 #endif  // CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(6, 6, &aom_highbd_8_sub_pixel_variance64x64_c, 8),
-  make_tuple(6, 5, &aom_highbd_8_sub_pixel_variance64x32_c, 8),
-  make_tuple(5, 6, &aom_highbd_8_sub_pixel_variance32x64_c, 8),
-  make_tuple(5, 5, &aom_highbd_8_sub_pixel_variance32x32_c, 8),
-  make_tuple(5, 4, &aom_highbd_8_sub_pixel_variance32x16_c, 8),
-  make_tuple(4, 5, &aom_highbd_8_sub_pixel_variance16x32_c, 8),
-  make_tuple(4, 4, &aom_highbd_8_sub_pixel_variance16x16_c, 8),
-  make_tuple(4, 3, &aom_highbd_8_sub_pixel_variance16x8_c, 8),
-  make_tuple(3, 4, &aom_highbd_8_sub_pixel_variance8x16_c, 8),
-  make_tuple(3, 3, &aom_highbd_8_sub_pixel_variance8x8_c, 8),
-  make_tuple(3, 2, &aom_highbd_8_sub_pixel_variance8x4_c, 8),
-  make_tuple(2, 3, &aom_highbd_8_sub_pixel_variance4x8_c, 8),
-  make_tuple(2, 2, &aom_highbd_8_sub_pixel_variance4x4_c, 8),
+  SubpelVarianceParams(6, 6, &aom_highbd_8_sub_pixel_variance64x64_c, 8),
+  SubpelVarianceParams(6, 5, &aom_highbd_8_sub_pixel_variance64x32_c, 8),
+  SubpelVarianceParams(5, 6, &aom_highbd_8_sub_pixel_variance32x64_c, 8),
+  SubpelVarianceParams(5, 5, &aom_highbd_8_sub_pixel_variance32x32_c, 8),
+  SubpelVarianceParams(5, 4, &aom_highbd_8_sub_pixel_variance32x16_c, 8),
+  SubpelVarianceParams(4, 5, &aom_highbd_8_sub_pixel_variance16x32_c, 8),
+  SubpelVarianceParams(4, 4, &aom_highbd_8_sub_pixel_variance16x16_c, 8),
+  SubpelVarianceParams(4, 3, &aom_highbd_8_sub_pixel_variance16x8_c, 8),
+  SubpelVarianceParams(3, 4, &aom_highbd_8_sub_pixel_variance8x16_c, 8),
+  SubpelVarianceParams(3, 3, &aom_highbd_8_sub_pixel_variance8x8_c, 8),
+  SubpelVarianceParams(3, 2, &aom_highbd_8_sub_pixel_variance8x4_c, 8),
+  SubpelVarianceParams(2, 3, &aom_highbd_8_sub_pixel_variance4x8_c, 8),
+  SubpelVarianceParams(2, 2, &aom_highbd_8_sub_pixel_variance4x4_c, 8),
 #if CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(7, 7, &aom_highbd_10_sub_pixel_variance128x128_c, 10),
-  make_tuple(7, 6, &aom_highbd_10_sub_pixel_variance128x64_c, 10),
-  make_tuple(6, 7, &aom_highbd_10_sub_pixel_variance64x128_c, 10),
+  SubpelVarianceParams(7, 7, &aom_highbd_10_sub_pixel_variance128x128_c, 10),
+  SubpelVarianceParams(7, 6, &aom_highbd_10_sub_pixel_variance128x64_c, 10),
+  SubpelVarianceParams(6, 7, &aom_highbd_10_sub_pixel_variance64x128_c, 10),
 #endif  // CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(6, 6, &aom_highbd_10_sub_pixel_variance64x64_c, 10),
-  make_tuple(6, 5, &aom_highbd_10_sub_pixel_variance64x32_c, 10),
-  make_tuple(5, 6, &aom_highbd_10_sub_pixel_variance32x64_c, 10),
-  make_tuple(5, 5, &aom_highbd_10_sub_pixel_variance32x32_c, 10),
-  make_tuple(5, 4, &aom_highbd_10_sub_pixel_variance32x16_c, 10),
-  make_tuple(4, 5, &aom_highbd_10_sub_pixel_variance16x32_c, 10),
-  make_tuple(4, 4, &aom_highbd_10_sub_pixel_variance16x16_c, 10),
-  make_tuple(4, 3, &aom_highbd_10_sub_pixel_variance16x8_c, 10),
-  make_tuple(3, 4, &aom_highbd_10_sub_pixel_variance8x16_c, 10),
-  make_tuple(3, 3, &aom_highbd_10_sub_pixel_variance8x8_c, 10),
-  make_tuple(3, 2, &aom_highbd_10_sub_pixel_variance8x4_c, 10),
-  make_tuple(2, 3, &aom_highbd_10_sub_pixel_variance4x8_c, 10),
-  make_tuple(2, 2, &aom_highbd_10_sub_pixel_variance4x4_c, 10),
+  SubpelVarianceParams(6, 6, &aom_highbd_10_sub_pixel_variance64x64_c, 10),
+  SubpelVarianceParams(6, 5, &aom_highbd_10_sub_pixel_variance64x32_c, 10),
+  SubpelVarianceParams(5, 6, &aom_highbd_10_sub_pixel_variance32x64_c, 10),
+  SubpelVarianceParams(5, 5, &aom_highbd_10_sub_pixel_variance32x32_c, 10),
+  SubpelVarianceParams(5, 4, &aom_highbd_10_sub_pixel_variance32x16_c, 10),
+  SubpelVarianceParams(4, 5, &aom_highbd_10_sub_pixel_variance16x32_c, 10),
+  SubpelVarianceParams(4, 4, &aom_highbd_10_sub_pixel_variance16x16_c, 10),
+  SubpelVarianceParams(4, 3, &aom_highbd_10_sub_pixel_variance16x8_c, 10),
+  SubpelVarianceParams(3, 4, &aom_highbd_10_sub_pixel_variance8x16_c, 10),
+  SubpelVarianceParams(3, 3, &aom_highbd_10_sub_pixel_variance8x8_c, 10),
+  SubpelVarianceParams(3, 2, &aom_highbd_10_sub_pixel_variance8x4_c, 10),
+  SubpelVarianceParams(2, 3, &aom_highbd_10_sub_pixel_variance4x8_c, 10),
+  SubpelVarianceParams(2, 2, &aom_highbd_10_sub_pixel_variance4x4_c, 10),
 #if CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(7, 7, &aom_highbd_12_sub_pixel_variance128x128_c, 12),
-  make_tuple(7, 6, &aom_highbd_12_sub_pixel_variance128x64_c, 12),
-  make_tuple(6, 7, &aom_highbd_12_sub_pixel_variance64x128_c, 12),
+  SubpelVarianceParams(7, 7, &aom_highbd_12_sub_pixel_variance128x128_c, 12),
+  SubpelVarianceParams(7, 6, &aom_highbd_12_sub_pixel_variance128x64_c, 12),
+  SubpelVarianceParams(6, 7, &aom_highbd_12_sub_pixel_variance64x128_c, 12),
 #endif  // CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(6, 6, &aom_highbd_12_sub_pixel_variance64x64_c, 12),
-  make_tuple(6, 5, &aom_highbd_12_sub_pixel_variance64x32_c, 12),
-  make_tuple(5, 6, &aom_highbd_12_sub_pixel_variance32x64_c, 12),
-  make_tuple(5, 5, &aom_highbd_12_sub_pixel_variance32x32_c, 12),
-  make_tuple(5, 4, &aom_highbd_12_sub_pixel_variance32x16_c, 12),
-  make_tuple(4, 5, &aom_highbd_12_sub_pixel_variance16x32_c, 12),
-  make_tuple(4, 4, &aom_highbd_12_sub_pixel_variance16x16_c, 12),
-  make_tuple(4, 3, &aom_highbd_12_sub_pixel_variance16x8_c, 12),
-  make_tuple(3, 4, &aom_highbd_12_sub_pixel_variance8x16_c, 12),
-  make_tuple(3, 3, &aom_highbd_12_sub_pixel_variance8x8_c, 12),
-  make_tuple(3, 2, &aom_highbd_12_sub_pixel_variance8x4_c, 12),
-  make_tuple(2, 3, &aom_highbd_12_sub_pixel_variance4x8_c, 12),
-  make_tuple(2, 2, &aom_highbd_12_sub_pixel_variance4x4_c, 12),
+  SubpelVarianceParams(6, 6, &aom_highbd_12_sub_pixel_variance64x64_c, 12),
+  SubpelVarianceParams(6, 5, &aom_highbd_12_sub_pixel_variance64x32_c, 12),
+  SubpelVarianceParams(5, 6, &aom_highbd_12_sub_pixel_variance32x64_c, 12),
+  SubpelVarianceParams(5, 5, &aom_highbd_12_sub_pixel_variance32x32_c, 12),
+  SubpelVarianceParams(5, 4, &aom_highbd_12_sub_pixel_variance32x16_c, 12),
+  SubpelVarianceParams(4, 5, &aom_highbd_12_sub_pixel_variance16x32_c, 12),
+  SubpelVarianceParams(4, 4, &aom_highbd_12_sub_pixel_variance16x16_c, 12),
+  SubpelVarianceParams(4, 3, &aom_highbd_12_sub_pixel_variance16x8_c, 12),
+  SubpelVarianceParams(3, 4, &aom_highbd_12_sub_pixel_variance8x16_c, 12),
+  SubpelVarianceParams(3, 3, &aom_highbd_12_sub_pixel_variance8x8_c, 12),
+  SubpelVarianceParams(3, 2, &aom_highbd_12_sub_pixel_variance8x4_c, 12),
+  SubpelVarianceParams(2, 3, &aom_highbd_12_sub_pixel_variance4x8_c, 12),
+  SubpelVarianceParams(2, 2, &aom_highbd_12_sub_pixel_variance4x4_c, 12),
 };
 INSTANTIATE_TEST_CASE_P(C, AvxHBDSubpelVarianceTest,
                         ::testing::ValuesIn(kArrayHBDSubpelVariance_c));
 
-const AvxHBDSubpelAvgVarianceTest::ParamType kArrayHBDSubpelAvgVariance_c[] = {
+const SubpelAvgVarianceParams kArrayHBDSubpelAvgVariance_c[] = {
 #if CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(7, 7, &aom_highbd_8_sub_pixel_avg_variance128x128_c, 8),
-  make_tuple(7, 6, &aom_highbd_8_sub_pixel_avg_variance128x64_c, 8),
-  make_tuple(6, 7, &aom_highbd_8_sub_pixel_avg_variance64x128_c, 8),
+  SubpelAvgVarianceParams(7, 7, &aom_highbd_8_sub_pixel_avg_variance128x128_c,
+                          8),
+  SubpelAvgVarianceParams(7, 6, &aom_highbd_8_sub_pixel_avg_variance128x64_c,
+                          8),
+  SubpelAvgVarianceParams(6, 7, &aom_highbd_8_sub_pixel_avg_variance64x128_c,
+                          8),
 #endif  // CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(6, 6, &aom_highbd_8_sub_pixel_avg_variance64x64_c, 8),
-  make_tuple(6, 5, &aom_highbd_8_sub_pixel_avg_variance64x32_c, 8),
-  make_tuple(5, 6, &aom_highbd_8_sub_pixel_avg_variance32x64_c, 8),
-  make_tuple(5, 5, &aom_highbd_8_sub_pixel_avg_variance32x32_c, 8),
-  make_tuple(5, 4, &aom_highbd_8_sub_pixel_avg_variance32x16_c, 8),
-  make_tuple(4, 5, &aom_highbd_8_sub_pixel_avg_variance16x32_c, 8),
-  make_tuple(4, 4, &aom_highbd_8_sub_pixel_avg_variance16x16_c, 8),
-  make_tuple(4, 3, &aom_highbd_8_sub_pixel_avg_variance16x8_c, 8),
-  make_tuple(3, 4, &aom_highbd_8_sub_pixel_avg_variance8x16_c, 8),
-  make_tuple(3, 3, &aom_highbd_8_sub_pixel_avg_variance8x8_c, 8),
-  make_tuple(3, 2, &aom_highbd_8_sub_pixel_avg_variance8x4_c, 8),
-  make_tuple(2, 3, &aom_highbd_8_sub_pixel_avg_variance4x8_c, 8),
-  make_tuple(2, 2, &aom_highbd_8_sub_pixel_avg_variance4x4_c, 8),
+  SubpelAvgVarianceParams(6, 6, &aom_highbd_8_sub_pixel_avg_variance64x64_c, 8),
+  SubpelAvgVarianceParams(6, 5, &aom_highbd_8_sub_pixel_avg_variance64x32_c, 8),
+  SubpelAvgVarianceParams(5, 6, &aom_highbd_8_sub_pixel_avg_variance32x64_c, 8),
+  SubpelAvgVarianceParams(5, 5, &aom_highbd_8_sub_pixel_avg_variance32x32_c, 8),
+  SubpelAvgVarianceParams(5, 4, &aom_highbd_8_sub_pixel_avg_variance32x16_c, 8),
+  SubpelAvgVarianceParams(4, 5, &aom_highbd_8_sub_pixel_avg_variance16x32_c, 8),
+  SubpelAvgVarianceParams(4, 4, &aom_highbd_8_sub_pixel_avg_variance16x16_c, 8),
+  SubpelAvgVarianceParams(4, 3, &aom_highbd_8_sub_pixel_avg_variance16x8_c, 8),
+  SubpelAvgVarianceParams(3, 4, &aom_highbd_8_sub_pixel_avg_variance8x16_c, 8),
+  SubpelAvgVarianceParams(3, 3, &aom_highbd_8_sub_pixel_avg_variance8x8_c, 8),
+  SubpelAvgVarianceParams(3, 2, &aom_highbd_8_sub_pixel_avg_variance8x4_c, 8),
+  SubpelAvgVarianceParams(2, 3, &aom_highbd_8_sub_pixel_avg_variance4x8_c, 8),
+  SubpelAvgVarianceParams(2, 2, &aom_highbd_8_sub_pixel_avg_variance4x4_c, 8),
 #if CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(7, 7, &aom_highbd_10_sub_pixel_avg_variance128x128_c, 10),
-  make_tuple(7, 6, &aom_highbd_10_sub_pixel_avg_variance128x64_c, 10),
-  make_tuple(6, 7, &aom_highbd_10_sub_pixel_avg_variance64x128_c, 10),
+  SubpelAvgVarianceParams(7, 7, &aom_highbd_10_sub_pixel_avg_variance128x128_c,
+                          10),
+  SubpelAvgVarianceParams(7, 6, &aom_highbd_10_sub_pixel_avg_variance128x64_c,
+                          10),
+  SubpelAvgVarianceParams(6, 7, &aom_highbd_10_sub_pixel_avg_variance64x128_c,
+                          10),
 #endif  // CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(6, 6, &aom_highbd_10_sub_pixel_avg_variance64x64_c, 10),
-  make_tuple(6, 5, &aom_highbd_10_sub_pixel_avg_variance64x32_c, 10),
-  make_tuple(5, 6, &aom_highbd_10_sub_pixel_avg_variance32x64_c, 10),
-  make_tuple(5, 5, &aom_highbd_10_sub_pixel_avg_variance32x32_c, 10),
-  make_tuple(5, 4, &aom_highbd_10_sub_pixel_avg_variance32x16_c, 10),
-  make_tuple(4, 5, &aom_highbd_10_sub_pixel_avg_variance16x32_c, 10),
-  make_tuple(4, 4, &aom_highbd_10_sub_pixel_avg_variance16x16_c, 10),
-  make_tuple(4, 3, &aom_highbd_10_sub_pixel_avg_variance16x8_c, 10),
-  make_tuple(3, 4, &aom_highbd_10_sub_pixel_avg_variance8x16_c, 10),
-  make_tuple(3, 3, &aom_highbd_10_sub_pixel_avg_variance8x8_c, 10),
-  make_tuple(3, 2, &aom_highbd_10_sub_pixel_avg_variance8x4_c, 10),
-  make_tuple(2, 3, &aom_highbd_10_sub_pixel_avg_variance4x8_c, 10),
-  make_tuple(2, 2, &aom_highbd_10_sub_pixel_avg_variance4x4_c, 10),
+  SubpelAvgVarianceParams(6, 6, &aom_highbd_10_sub_pixel_avg_variance64x64_c,
+                          10),
+  SubpelAvgVarianceParams(6, 5, &aom_highbd_10_sub_pixel_avg_variance64x32_c,
+                          10),
+  SubpelAvgVarianceParams(5, 6, &aom_highbd_10_sub_pixel_avg_variance32x64_c,
+                          10),
+  SubpelAvgVarianceParams(5, 5, &aom_highbd_10_sub_pixel_avg_variance32x32_c,
+                          10),
+  SubpelAvgVarianceParams(5, 4, &aom_highbd_10_sub_pixel_avg_variance32x16_c,
+                          10),
+  SubpelAvgVarianceParams(4, 5, &aom_highbd_10_sub_pixel_avg_variance16x32_c,
+                          10),
+  SubpelAvgVarianceParams(4, 4, &aom_highbd_10_sub_pixel_avg_variance16x16_c,
+                          10),
+  SubpelAvgVarianceParams(4, 3, &aom_highbd_10_sub_pixel_avg_variance16x8_c,
+                          10),
+  SubpelAvgVarianceParams(3, 4, &aom_highbd_10_sub_pixel_avg_variance8x16_c,
+                          10),
+  SubpelAvgVarianceParams(3, 3, &aom_highbd_10_sub_pixel_avg_variance8x8_c, 10),
+  SubpelAvgVarianceParams(3, 2, &aom_highbd_10_sub_pixel_avg_variance8x4_c, 10),
+  SubpelAvgVarianceParams(2, 3, &aom_highbd_10_sub_pixel_avg_variance4x8_c, 10),
+  SubpelAvgVarianceParams(2, 2, &aom_highbd_10_sub_pixel_avg_variance4x4_c, 10),
 #if CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(7, 7, &aom_highbd_12_sub_pixel_avg_variance128x128_c, 12),
-  make_tuple(7, 6, &aom_highbd_12_sub_pixel_avg_variance128x64_c, 12),
-  make_tuple(6, 7, &aom_highbd_12_sub_pixel_avg_variance64x128_c, 12),
+  SubpelAvgVarianceParams(7, 7, &aom_highbd_12_sub_pixel_avg_variance128x128_c,
+                          12),
+  SubpelAvgVarianceParams(7, 6, &aom_highbd_12_sub_pixel_avg_variance128x64_c,
+                          12),
+  SubpelAvgVarianceParams(6, 7, &aom_highbd_12_sub_pixel_avg_variance64x128_c,
+                          12),
 #endif  // CONFIG_AV1 && CONFIG_EXT_PARTITION
-  make_tuple(6, 6, &aom_highbd_12_sub_pixel_avg_variance64x64_c, 12),
-  make_tuple(6, 5, &aom_highbd_12_sub_pixel_avg_variance64x32_c, 12),
-  make_tuple(5, 6, &aom_highbd_12_sub_pixel_avg_variance32x64_c, 12),
-  make_tuple(5, 5, &aom_highbd_12_sub_pixel_avg_variance32x32_c, 12),
-  make_tuple(5, 4, &aom_highbd_12_sub_pixel_avg_variance32x16_c, 12),
-  make_tuple(4, 5, &aom_highbd_12_sub_pixel_avg_variance16x32_c, 12),
-  make_tuple(4, 4, &aom_highbd_12_sub_pixel_avg_variance16x16_c, 12),
-  make_tuple(4, 3, &aom_highbd_12_sub_pixel_avg_variance16x8_c, 12),
-  make_tuple(3, 4, &aom_highbd_12_sub_pixel_avg_variance8x16_c, 12),
-  make_tuple(3, 3, &aom_highbd_12_sub_pixel_avg_variance8x8_c, 12),
-  make_tuple(3, 2, &aom_highbd_12_sub_pixel_avg_variance8x4_c, 12),
-  make_tuple(2, 3, &aom_highbd_12_sub_pixel_avg_variance4x8_c, 12),
-  make_tuple(2, 2, &aom_highbd_12_sub_pixel_avg_variance4x4_c, 12)
+  SubpelAvgVarianceParams(6, 6, &aom_highbd_12_sub_pixel_avg_variance64x64_c,
+                          12),
+  SubpelAvgVarianceParams(6, 5, &aom_highbd_12_sub_pixel_avg_variance64x32_c,
+                          12),
+  SubpelAvgVarianceParams(5, 6, &aom_highbd_12_sub_pixel_avg_variance32x64_c,
+                          12),
+  SubpelAvgVarianceParams(5, 5, &aom_highbd_12_sub_pixel_avg_variance32x32_c,
+                          12),
+  SubpelAvgVarianceParams(5, 4, &aom_highbd_12_sub_pixel_avg_variance32x16_c,
+                          12),
+  SubpelAvgVarianceParams(4, 5, &aom_highbd_12_sub_pixel_avg_variance16x32_c,
+                          12),
+  SubpelAvgVarianceParams(4, 4, &aom_highbd_12_sub_pixel_avg_variance16x16_c,
+                          12),
+  SubpelAvgVarianceParams(4, 3, &aom_highbd_12_sub_pixel_avg_variance16x8_c,
+                          12),
+  SubpelAvgVarianceParams(3, 4, &aom_highbd_12_sub_pixel_avg_variance8x16_c,
+                          12),
+  SubpelAvgVarianceParams(3, 3, &aom_highbd_12_sub_pixel_avg_variance8x8_c, 12),
+  SubpelAvgVarianceParams(3, 2, &aom_highbd_12_sub_pixel_avg_variance8x4_c, 12),
+  SubpelAvgVarianceParams(2, 3, &aom_highbd_12_sub_pixel_avg_variance4x8_c, 12),
+  SubpelAvgVarianceParams(2, 2, &aom_highbd_12_sub_pixel_avg_variance4x4_c, 12)
 };
 INSTANTIATE_TEST_CASE_P(C, AvxHBDSubpelAvgVarianceTest,
                         ::testing::ValuesIn(kArrayHBDSubpelAvgVariance_c));
@@ -1043,51 +1060,61 @@ INSTANTIATE_TEST_CASE_P(
 
 INSTANTIATE_TEST_CASE_P(
     SSE2, AvxSubpelVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_variance64x64_sse2, 0),
-                      make_tuple(6, 5, &aom_sub_pixel_variance64x32_sse2, 0),
-                      make_tuple(5, 6, &aom_sub_pixel_variance32x64_sse2, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_variance32x32_sse2, 0),
-                      make_tuple(5, 4, &aom_sub_pixel_variance32x16_sse2, 0),
-                      make_tuple(4, 5, &aom_sub_pixel_variance16x32_sse2, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_variance16x16_sse2, 0),
-                      make_tuple(4, 3, &aom_sub_pixel_variance16x8_sse2, 0),
-                      make_tuple(3, 4, &aom_sub_pixel_variance8x16_sse2, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_variance8x8_sse2, 0),
-                      make_tuple(3, 2, &aom_sub_pixel_variance8x4_sse2, 0),
-                      make_tuple(2, 3, &aom_sub_pixel_variance4x8_sse2, 0),
-                      make_tuple(2, 2, &aom_sub_pixel_variance4x4_sse2, 0)));
+    ::testing::Values(
+        SubpelVarianceParams(6, 6, &aom_sub_pixel_variance64x64_sse2, 0),
+        SubpelVarianceParams(6, 5, &aom_sub_pixel_variance64x32_sse2, 0),
+        SubpelVarianceParams(5, 6, &aom_sub_pixel_variance32x64_sse2, 0),
+        SubpelVarianceParams(5, 5, &aom_sub_pixel_variance32x32_sse2, 0),
+        SubpelVarianceParams(5, 4, &aom_sub_pixel_variance32x16_sse2, 0),
+        SubpelVarianceParams(4, 5, &aom_sub_pixel_variance16x32_sse2, 0),
+        SubpelVarianceParams(4, 4, &aom_sub_pixel_variance16x16_sse2, 0),
+        SubpelVarianceParams(4, 3, &aom_sub_pixel_variance16x8_sse2, 0),
+        SubpelVarianceParams(3, 4, &aom_sub_pixel_variance8x16_sse2, 0),
+        SubpelVarianceParams(3, 3, &aom_sub_pixel_variance8x8_sse2, 0),
+        SubpelVarianceParams(3, 2, &aom_sub_pixel_variance8x4_sse2, 0),
+        SubpelVarianceParams(2, 3, &aom_sub_pixel_variance4x8_sse2, 0),
+        SubpelVarianceParams(2, 2, &aom_sub_pixel_variance4x4_sse2, 0)));
 
 INSTANTIATE_TEST_CASE_P(
     SSE2, AvxSubpelAvgVarianceTest,
     ::testing::Values(
-        make_tuple(6, 6, &aom_sub_pixel_avg_variance64x64_sse2, 0),
-        make_tuple(6, 5, &aom_sub_pixel_avg_variance64x32_sse2, 0),
-        make_tuple(5, 6, &aom_sub_pixel_avg_variance32x64_sse2, 0),
-        make_tuple(5, 5, &aom_sub_pixel_avg_variance32x32_sse2, 0),
-        make_tuple(5, 4, &aom_sub_pixel_avg_variance32x16_sse2, 0),
-        make_tuple(4, 5, &aom_sub_pixel_avg_variance16x32_sse2, 0),
-        make_tuple(4, 4, &aom_sub_pixel_avg_variance16x16_sse2, 0),
-        make_tuple(4, 3, &aom_sub_pixel_avg_variance16x8_sse2, 0),
-        make_tuple(3, 4, &aom_sub_pixel_avg_variance8x16_sse2, 0),
-        make_tuple(3, 3, &aom_sub_pixel_avg_variance8x8_sse2, 0),
-        make_tuple(3, 2, &aom_sub_pixel_avg_variance8x4_sse2, 0),
-        make_tuple(2, 3, &aom_sub_pixel_avg_variance4x8_sse2, 0),
-        make_tuple(2, 2, &aom_sub_pixel_avg_variance4x4_sse2, 0)));
+        SubpelAvgVarianceParams(6, 6, &aom_sub_pixel_avg_variance64x64_sse2, 0),
+        SubpelAvgVarianceParams(6, 5, &aom_sub_pixel_avg_variance64x32_sse2, 0),
+        SubpelAvgVarianceParams(5, 6, &aom_sub_pixel_avg_variance32x64_sse2, 0),
+        SubpelAvgVarianceParams(5, 5, &aom_sub_pixel_avg_variance32x32_sse2, 0),
+        SubpelAvgVarianceParams(5, 4, &aom_sub_pixel_avg_variance32x16_sse2, 0),
+        SubpelAvgVarianceParams(4, 5, &aom_sub_pixel_avg_variance16x32_sse2, 0),
+        SubpelAvgVarianceParams(4, 4, &aom_sub_pixel_avg_variance16x16_sse2, 0),
+        SubpelAvgVarianceParams(4, 3, &aom_sub_pixel_avg_variance16x8_sse2, 0),
+        SubpelAvgVarianceParams(3, 4, &aom_sub_pixel_avg_variance8x16_sse2, 0),
+        SubpelAvgVarianceParams(3, 3, &aom_sub_pixel_avg_variance8x8_sse2, 0),
+        SubpelAvgVarianceParams(3, 2, &aom_sub_pixel_avg_variance8x4_sse2, 0),
+        SubpelAvgVarianceParams(2, 3, &aom_sub_pixel_avg_variance4x8_sse2, 0),
+        SubpelAvgVarianceParams(2, 2, &aom_sub_pixel_avg_variance4x4_sse2, 0)));
 
 #if HAVE_SSE4_1 && CONFIG_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(
     SSE4_1, AvxSubpelVarianceTest,
     ::testing::Values(
-        make_tuple(2, 2, &aom_highbd_8_sub_pixel_variance4x4_sse4_1, 8),
-        make_tuple(2, 2, &aom_highbd_10_sub_pixel_variance4x4_sse4_1, 10),
-        make_tuple(2, 2, &aom_highbd_12_sub_pixel_variance4x4_sse4_1, 12)));
+        SubpelVarianceParams(2, 2, &aom_highbd_8_sub_pixel_variance4x4_sse4_1,
+                             8),
+        SubpelVarianceParams(2, 2, &aom_highbd_10_sub_pixel_variance4x4_sse4_1,
+                             10),
+        SubpelVarianceParams(2, 2, &aom_highbd_12_sub_pixel_variance4x4_sse4_1,
+                             12)));
 
 INSTANTIATE_TEST_CASE_P(
     SSE4_1, AvxSubpelAvgVarianceTest,
     ::testing::Values(
-        make_tuple(2, 2, &aom_highbd_8_sub_pixel_avg_variance4x4_sse4_1, 8),
-        make_tuple(2, 2, &aom_highbd_10_sub_pixel_avg_variance4x4_sse4_1, 10),
-        make_tuple(2, 2, &aom_highbd_12_sub_pixel_avg_variance4x4_sse4_1, 12)));
+        SubpelAvgVarianceParams(2, 2,
+                                &aom_highbd_8_sub_pixel_avg_variance4x4_sse4_1,
+                                8),
+        SubpelAvgVarianceParams(2, 2,
+                                &aom_highbd_10_sub_pixel_avg_variance4x4_sse4_1,
+                                10),
+        SubpelAvgVarianceParams(2, 2,
+                                &aom_highbd_12_sub_pixel_avg_variance4x4_sse4_1,
+                                12)));
 #endif  // HAVE_SSE4_1 && CONFIG_HIGHBITDEPTH
 
 #if CONFIG_HIGHBITDEPTH
@@ -1142,115 +1169,160 @@ INSTANTIATE_TEST_CASE_P(
         VarianceParams(3, 4, &aom_highbd_8_variance8x16_sse2, 8),
         VarianceParams(3, 3, &aom_highbd_8_variance8x8_sse2, 8)));
 
-INSTANTIATE_TEST_CASE_P(
-    SSE2, AvxHBDSubpelVarianceTest,
-    ::testing::Values(
-        make_tuple(6, 6, &aom_highbd_12_sub_pixel_variance64x64_sse2, 12),
-        make_tuple(6, 5, &aom_highbd_12_sub_pixel_variance64x32_sse2, 12),
-        make_tuple(5, 6, &aom_highbd_12_sub_pixel_variance32x64_sse2, 12),
-        make_tuple(5, 5, &aom_highbd_12_sub_pixel_variance32x32_sse2, 12),
-        make_tuple(5, 4, &aom_highbd_12_sub_pixel_variance32x16_sse2, 12),
-        make_tuple(4, 5, &aom_highbd_12_sub_pixel_variance16x32_sse2, 12),
-        make_tuple(4, 4, &aom_highbd_12_sub_pixel_variance16x16_sse2, 12),
-        make_tuple(4, 3, &aom_highbd_12_sub_pixel_variance16x8_sse2, 12),
-        make_tuple(3, 4, &aom_highbd_12_sub_pixel_variance8x16_sse2, 12),
-        make_tuple(3, 3, &aom_highbd_12_sub_pixel_variance8x8_sse2, 12),
-        make_tuple(3, 2, &aom_highbd_12_sub_pixel_variance8x4_sse2, 12),
-        make_tuple(6, 6, &aom_highbd_10_sub_pixel_variance64x64_sse2, 10),
-        make_tuple(6, 5, &aom_highbd_10_sub_pixel_variance64x32_sse2, 10),
-        make_tuple(5, 6, &aom_highbd_10_sub_pixel_variance32x64_sse2, 10),
-        make_tuple(5, 5, &aom_highbd_10_sub_pixel_variance32x32_sse2, 10),
-        make_tuple(5, 4, &aom_highbd_10_sub_pixel_variance32x16_sse2, 10),
-        make_tuple(4, 5, &aom_highbd_10_sub_pixel_variance16x32_sse2, 10),
-        make_tuple(4, 4, &aom_highbd_10_sub_pixel_variance16x16_sse2, 10),
-        make_tuple(4, 3, &aom_highbd_10_sub_pixel_variance16x8_sse2, 10),
-        make_tuple(3, 4, &aom_highbd_10_sub_pixel_variance8x16_sse2, 10),
-        make_tuple(3, 3, &aom_highbd_10_sub_pixel_variance8x8_sse2, 10),
-        make_tuple(3, 2, &aom_highbd_10_sub_pixel_variance8x4_sse2, 10),
-        make_tuple(6, 6, &aom_highbd_8_sub_pixel_variance64x64_sse2, 8),
-        make_tuple(6, 5, &aom_highbd_8_sub_pixel_variance64x32_sse2, 8),
-        make_tuple(5, 6, &aom_highbd_8_sub_pixel_variance32x64_sse2, 8),
-        make_tuple(5, 5, &aom_highbd_8_sub_pixel_variance32x32_sse2, 8),
-        make_tuple(5, 4, &aom_highbd_8_sub_pixel_variance32x16_sse2, 8),
-        make_tuple(4, 5, &aom_highbd_8_sub_pixel_variance16x32_sse2, 8),
-        make_tuple(4, 4, &aom_highbd_8_sub_pixel_variance16x16_sse2, 8),
-        make_tuple(4, 3, &aom_highbd_8_sub_pixel_variance16x8_sse2, 8),
-        make_tuple(3, 4, &aom_highbd_8_sub_pixel_variance8x16_sse2, 8),
-        make_tuple(3, 3, &aom_highbd_8_sub_pixel_variance8x8_sse2, 8),
-        make_tuple(3, 2, &aom_highbd_8_sub_pixel_variance8x4_sse2, 8)));
+const SubpelVarianceParams kArrayHBDSubpelVariance_sse2[] = {
+  SubpelVarianceParams(6, 6, &aom_highbd_12_sub_pixel_variance64x64_sse2, 12),
+  SubpelVarianceParams(6, 5, &aom_highbd_12_sub_pixel_variance64x32_sse2, 12),
+  SubpelVarianceParams(5, 6, &aom_highbd_12_sub_pixel_variance32x64_sse2, 12),
+  SubpelVarianceParams(5, 5, &aom_highbd_12_sub_pixel_variance32x32_sse2, 12),
+  SubpelVarianceParams(5, 4, &aom_highbd_12_sub_pixel_variance32x16_sse2, 12),
+  SubpelVarianceParams(4, 5, &aom_highbd_12_sub_pixel_variance16x32_sse2, 12),
+  SubpelVarianceParams(4, 4, &aom_highbd_12_sub_pixel_variance16x16_sse2, 12),
+  SubpelVarianceParams(4, 3, &aom_highbd_12_sub_pixel_variance16x8_sse2, 12),
+  SubpelVarianceParams(3, 4, &aom_highbd_12_sub_pixel_variance8x16_sse2, 12),
+  SubpelVarianceParams(3, 3, &aom_highbd_12_sub_pixel_variance8x8_sse2, 12),
+  SubpelVarianceParams(3, 2, &aom_highbd_12_sub_pixel_variance8x4_sse2, 12),
+  SubpelVarianceParams(6, 6, &aom_highbd_10_sub_pixel_variance64x64_sse2, 10),
+  SubpelVarianceParams(6, 5, &aom_highbd_10_sub_pixel_variance64x32_sse2, 10),
+  SubpelVarianceParams(5, 6, &aom_highbd_10_sub_pixel_variance32x64_sse2, 10),
+  SubpelVarianceParams(5, 5, &aom_highbd_10_sub_pixel_variance32x32_sse2, 10),
+  SubpelVarianceParams(5, 4, &aom_highbd_10_sub_pixel_variance32x16_sse2, 10),
+  SubpelVarianceParams(4, 5, &aom_highbd_10_sub_pixel_variance16x32_sse2, 10),
+  SubpelVarianceParams(4, 4, &aom_highbd_10_sub_pixel_variance16x16_sse2, 10),
+  SubpelVarianceParams(4, 3, &aom_highbd_10_sub_pixel_variance16x8_sse2, 10),
+  SubpelVarianceParams(3, 4, &aom_highbd_10_sub_pixel_variance8x16_sse2, 10),
+  SubpelVarianceParams(3, 3, &aom_highbd_10_sub_pixel_variance8x8_sse2, 10),
+  SubpelVarianceParams(3, 2, &aom_highbd_10_sub_pixel_variance8x4_sse2, 10),
+  SubpelVarianceParams(6, 6, &aom_highbd_8_sub_pixel_variance64x64_sse2, 8),
+  SubpelVarianceParams(6, 5, &aom_highbd_8_sub_pixel_variance64x32_sse2, 8),
+  SubpelVarianceParams(5, 6, &aom_highbd_8_sub_pixel_variance32x64_sse2, 8),
+  SubpelVarianceParams(5, 5, &aom_highbd_8_sub_pixel_variance32x32_sse2, 8),
+  SubpelVarianceParams(5, 4, &aom_highbd_8_sub_pixel_variance32x16_sse2, 8),
+  SubpelVarianceParams(4, 5, &aom_highbd_8_sub_pixel_variance16x32_sse2, 8),
+  SubpelVarianceParams(4, 4, &aom_highbd_8_sub_pixel_variance16x16_sse2, 8),
+  SubpelVarianceParams(4, 3, &aom_highbd_8_sub_pixel_variance16x8_sse2, 8),
+  SubpelVarianceParams(3, 4, &aom_highbd_8_sub_pixel_variance8x16_sse2, 8),
+  SubpelVarianceParams(3, 3, &aom_highbd_8_sub_pixel_variance8x8_sse2, 8),
+  SubpelVarianceParams(3, 2, &aom_highbd_8_sub_pixel_variance8x4_sse2, 8)
+};
 
-INSTANTIATE_TEST_CASE_P(
-    SSE2, AvxHBDSubpelAvgVarianceTest,
-    ::testing::Values(
-        make_tuple(6, 6, &aom_highbd_12_sub_pixel_avg_variance64x64_sse2, 12),
-        make_tuple(6, 5, &aom_highbd_12_sub_pixel_avg_variance64x32_sse2, 12),
-        make_tuple(5, 6, &aom_highbd_12_sub_pixel_avg_variance32x64_sse2, 12),
-        make_tuple(5, 5, &aom_highbd_12_sub_pixel_avg_variance32x32_sse2, 12),
-        make_tuple(5, 4, &aom_highbd_12_sub_pixel_avg_variance32x16_sse2, 12),
-        make_tuple(4, 5, &aom_highbd_12_sub_pixel_avg_variance16x32_sse2, 12),
-        make_tuple(4, 4, &aom_highbd_12_sub_pixel_avg_variance16x16_sse2, 12),
-        make_tuple(4, 3, &aom_highbd_12_sub_pixel_avg_variance16x8_sse2, 12),
-        make_tuple(3, 4, &aom_highbd_12_sub_pixel_avg_variance8x16_sse2, 12),
-        make_tuple(3, 3, &aom_highbd_12_sub_pixel_avg_variance8x8_sse2, 12),
-        make_tuple(3, 2, &aom_highbd_12_sub_pixel_avg_variance8x4_sse2, 12),
-        make_tuple(6, 6, &aom_highbd_10_sub_pixel_avg_variance64x64_sse2, 10),
-        make_tuple(6, 5, &aom_highbd_10_sub_pixel_avg_variance64x32_sse2, 10),
-        make_tuple(5, 6, &aom_highbd_10_sub_pixel_avg_variance32x64_sse2, 10),
-        make_tuple(5, 5, &aom_highbd_10_sub_pixel_avg_variance32x32_sse2, 10),
-        make_tuple(5, 4, &aom_highbd_10_sub_pixel_avg_variance32x16_sse2, 10),
-        make_tuple(4, 5, &aom_highbd_10_sub_pixel_avg_variance16x32_sse2, 10),
-        make_tuple(4, 4, &aom_highbd_10_sub_pixel_avg_variance16x16_sse2, 10),
-        make_tuple(4, 3, &aom_highbd_10_sub_pixel_avg_variance16x8_sse2, 10),
-        make_tuple(3, 4, &aom_highbd_10_sub_pixel_avg_variance8x16_sse2, 10),
-        make_tuple(3, 3, &aom_highbd_10_sub_pixel_avg_variance8x8_sse2, 10),
-        make_tuple(3, 2, &aom_highbd_10_sub_pixel_avg_variance8x4_sse2, 10),
-        make_tuple(6, 6, &aom_highbd_8_sub_pixel_avg_variance64x64_sse2, 8),
-        make_tuple(6, 5, &aom_highbd_8_sub_pixel_avg_variance64x32_sse2, 8),
-        make_tuple(5, 6, &aom_highbd_8_sub_pixel_avg_variance32x64_sse2, 8),
-        make_tuple(5, 5, &aom_highbd_8_sub_pixel_avg_variance32x32_sse2, 8),
-        make_tuple(5, 4, &aom_highbd_8_sub_pixel_avg_variance32x16_sse2, 8),
-        make_tuple(4, 5, &aom_highbd_8_sub_pixel_avg_variance16x32_sse2, 8),
-        make_tuple(4, 4, &aom_highbd_8_sub_pixel_avg_variance16x16_sse2, 8),
-        make_tuple(4, 3, &aom_highbd_8_sub_pixel_avg_variance16x8_sse2, 8),
-        make_tuple(3, 4, &aom_highbd_8_sub_pixel_avg_variance8x16_sse2, 8),
-        make_tuple(3, 3, &aom_highbd_8_sub_pixel_avg_variance8x8_sse2, 8),
-        make_tuple(3, 2, &aom_highbd_8_sub_pixel_avg_variance8x4_sse2, 8)));
+INSTANTIATE_TEST_CASE_P(SSE2, AvxHBDSubpelVarianceTest,
+                        ::testing::ValuesIn(kArrayHBDSubpelVariance_sse2));
+
+const SubpelAvgVarianceParams kArrayHBDSubpelAvgVariance_sse2[] = {
+  SubpelAvgVarianceParams(6, 6, &aom_highbd_12_sub_pixel_avg_variance64x64_sse2,
+                          12),
+  SubpelAvgVarianceParams(6, 5, &aom_highbd_12_sub_pixel_avg_variance64x32_sse2,
+                          12),
+  SubpelAvgVarianceParams(5, 6, &aom_highbd_12_sub_pixel_avg_variance32x64_sse2,
+                          12),
+  SubpelAvgVarianceParams(5, 5, &aom_highbd_12_sub_pixel_avg_variance32x32_sse2,
+                          12),
+  SubpelAvgVarianceParams(5, 4, &aom_highbd_12_sub_pixel_avg_variance32x16_sse2,
+                          12),
+  SubpelAvgVarianceParams(4, 5, &aom_highbd_12_sub_pixel_avg_variance16x32_sse2,
+                          12),
+  SubpelAvgVarianceParams(4, 4, &aom_highbd_12_sub_pixel_avg_variance16x16_sse2,
+                          12),
+  SubpelAvgVarianceParams(4, 3, &aom_highbd_12_sub_pixel_avg_variance16x8_sse2,
+                          12),
+  SubpelAvgVarianceParams(3, 4, &aom_highbd_12_sub_pixel_avg_variance8x16_sse2,
+                          12),
+  SubpelAvgVarianceParams(3, 3, &aom_highbd_12_sub_pixel_avg_variance8x8_sse2,
+                          12),
+  SubpelAvgVarianceParams(3, 2, &aom_highbd_12_sub_pixel_avg_variance8x4_sse2,
+                          12),
+  SubpelAvgVarianceParams(6, 6, &aom_highbd_10_sub_pixel_avg_variance64x64_sse2,
+                          10),
+  SubpelAvgVarianceParams(6, 5, &aom_highbd_10_sub_pixel_avg_variance64x32_sse2,
+                          10),
+  SubpelAvgVarianceParams(5, 6, &aom_highbd_10_sub_pixel_avg_variance32x64_sse2,
+                          10),
+  SubpelAvgVarianceParams(5, 5, &aom_highbd_10_sub_pixel_avg_variance32x32_sse2,
+                          10),
+  SubpelAvgVarianceParams(5, 4, &aom_highbd_10_sub_pixel_avg_variance32x16_sse2,
+                          10),
+  SubpelAvgVarianceParams(4, 5, &aom_highbd_10_sub_pixel_avg_variance16x32_sse2,
+                          10),
+  SubpelAvgVarianceParams(4, 4, &aom_highbd_10_sub_pixel_avg_variance16x16_sse2,
+                          10),
+  SubpelAvgVarianceParams(4, 3, &aom_highbd_10_sub_pixel_avg_variance16x8_sse2,
+                          10),
+  SubpelAvgVarianceParams(3, 4, &aom_highbd_10_sub_pixel_avg_variance8x16_sse2,
+                          10),
+  SubpelAvgVarianceParams(3, 3, &aom_highbd_10_sub_pixel_avg_variance8x8_sse2,
+                          10),
+  SubpelAvgVarianceParams(3, 2, &aom_highbd_10_sub_pixel_avg_variance8x4_sse2,
+                          10),
+  SubpelAvgVarianceParams(6, 6, &aom_highbd_8_sub_pixel_avg_variance64x64_sse2,
+                          8),
+  SubpelAvgVarianceParams(6, 5, &aom_highbd_8_sub_pixel_avg_variance64x32_sse2,
+                          8),
+  SubpelAvgVarianceParams(5, 6, &aom_highbd_8_sub_pixel_avg_variance32x64_sse2,
+                          8),
+  SubpelAvgVarianceParams(5, 5, &aom_highbd_8_sub_pixel_avg_variance32x32_sse2,
+                          8),
+  SubpelAvgVarianceParams(5, 4, &aom_highbd_8_sub_pixel_avg_variance32x16_sse2,
+                          8),
+  SubpelAvgVarianceParams(4, 5, &aom_highbd_8_sub_pixel_avg_variance16x32_sse2,
+                          8),
+  SubpelAvgVarianceParams(4, 4, &aom_highbd_8_sub_pixel_avg_variance16x16_sse2,
+                          8),
+  SubpelAvgVarianceParams(4, 3, &aom_highbd_8_sub_pixel_avg_variance16x8_sse2,
+                          8),
+  SubpelAvgVarianceParams(3, 4, &aom_highbd_8_sub_pixel_avg_variance8x16_sse2,
+                          8),
+  SubpelAvgVarianceParams(3, 3, &aom_highbd_8_sub_pixel_avg_variance8x8_sse2,
+                          8),
+  SubpelAvgVarianceParams(3, 2, &aom_highbd_8_sub_pixel_avg_variance8x4_sse2, 8)
+};
+
+INSTANTIATE_TEST_CASE_P(SSE2, AvxHBDSubpelAvgVarianceTest,
+                        ::testing::ValuesIn(kArrayHBDSubpelAvgVariance_sse2));
 #endif  // CONFIG_HIGHBITDEPTH
 #endif  // HAVE_SSE2
 
 #if HAVE_SSSE3
 INSTANTIATE_TEST_CASE_P(
     SSSE3, AvxSubpelVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_variance64x64_ssse3, 0),
-                      make_tuple(6, 5, &aom_sub_pixel_variance64x32_ssse3, 0),
-                      make_tuple(5, 6, &aom_sub_pixel_variance32x64_ssse3, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_variance32x32_ssse3, 0),
-                      make_tuple(5, 4, &aom_sub_pixel_variance32x16_ssse3, 0),
-                      make_tuple(4, 5, &aom_sub_pixel_variance16x32_ssse3, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_variance16x16_ssse3, 0),
-                      make_tuple(4, 3, &aom_sub_pixel_variance16x8_ssse3, 0),
-                      make_tuple(3, 4, &aom_sub_pixel_variance8x16_ssse3, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_variance8x8_ssse3, 0),
-                      make_tuple(3, 2, &aom_sub_pixel_variance8x4_ssse3, 0),
-                      make_tuple(2, 3, &aom_sub_pixel_variance4x8_ssse3, 0),
-                      make_tuple(2, 2, &aom_sub_pixel_variance4x4_ssse3, 0)));
+    ::testing::Values(
+        SubpelVarianceParams(6, 6, &aom_sub_pixel_variance64x64_ssse3, 0),
+        SubpelVarianceParams(6, 5, &aom_sub_pixel_variance64x32_ssse3, 0),
+        SubpelVarianceParams(5, 6, &aom_sub_pixel_variance32x64_ssse3, 0),
+        SubpelVarianceParams(5, 5, &aom_sub_pixel_variance32x32_ssse3, 0),
+        SubpelVarianceParams(5, 4, &aom_sub_pixel_variance32x16_ssse3, 0),
+        SubpelVarianceParams(4, 5, &aom_sub_pixel_variance16x32_ssse3, 0),
+        SubpelVarianceParams(4, 4, &aom_sub_pixel_variance16x16_ssse3, 0),
+        SubpelVarianceParams(4, 3, &aom_sub_pixel_variance16x8_ssse3, 0),
+        SubpelVarianceParams(3, 4, &aom_sub_pixel_variance8x16_ssse3, 0),
+        SubpelVarianceParams(3, 3, &aom_sub_pixel_variance8x8_ssse3, 0),
+        SubpelVarianceParams(3, 2, &aom_sub_pixel_variance8x4_ssse3, 0),
+        SubpelVarianceParams(2, 3, &aom_sub_pixel_variance4x8_ssse3, 0),
+        SubpelVarianceParams(2, 2, &aom_sub_pixel_variance4x4_ssse3, 0)));
 
 INSTANTIATE_TEST_CASE_P(
     SSSE3, AvxSubpelAvgVarianceTest,
     ::testing::Values(
-        make_tuple(6, 6, &aom_sub_pixel_avg_variance64x64_ssse3, 0),
-        make_tuple(6, 5, &aom_sub_pixel_avg_variance64x32_ssse3, 0),
-        make_tuple(5, 6, &aom_sub_pixel_avg_variance32x64_ssse3, 0),
-        make_tuple(5, 5, &aom_sub_pixel_avg_variance32x32_ssse3, 0),
-        make_tuple(5, 4, &aom_sub_pixel_avg_variance32x16_ssse3, 0),
-        make_tuple(4, 5, &aom_sub_pixel_avg_variance16x32_ssse3, 0),
-        make_tuple(4, 4, &aom_sub_pixel_avg_variance16x16_ssse3, 0),
-        make_tuple(4, 3, &aom_sub_pixel_avg_variance16x8_ssse3, 0),
-        make_tuple(3, 4, &aom_sub_pixel_avg_variance8x16_ssse3, 0),
-        make_tuple(3, 3, &aom_sub_pixel_avg_variance8x8_ssse3, 0),
-        make_tuple(3, 2, &aom_sub_pixel_avg_variance8x4_ssse3, 0),
-        make_tuple(2, 3, &aom_sub_pixel_avg_variance4x8_ssse3, 0),
-        make_tuple(2, 2, &aom_sub_pixel_avg_variance4x4_ssse3, 0)));
+        SubpelAvgVarianceParams(6, 6, &aom_sub_pixel_avg_variance64x64_ssse3,
+                                0),
+        SubpelAvgVarianceParams(6, 5, &aom_sub_pixel_avg_variance64x32_ssse3,
+                                0),
+        SubpelAvgVarianceParams(5, 6, &aom_sub_pixel_avg_variance32x64_ssse3,
+                                0),
+        SubpelAvgVarianceParams(5, 5, &aom_sub_pixel_avg_variance32x32_ssse3,
+                                0),
+        SubpelAvgVarianceParams(5, 4, &aom_sub_pixel_avg_variance32x16_ssse3,
+                                0),
+        SubpelAvgVarianceParams(4, 5, &aom_sub_pixel_avg_variance16x32_ssse3,
+                                0),
+        SubpelAvgVarianceParams(4, 4, &aom_sub_pixel_avg_variance16x16_ssse3,
+                                0),
+        SubpelAvgVarianceParams(4, 3, &aom_sub_pixel_avg_variance16x8_ssse3, 0),
+        SubpelAvgVarianceParams(3, 4, &aom_sub_pixel_avg_variance8x16_ssse3, 0),
+        SubpelAvgVarianceParams(3, 3, &aom_sub_pixel_avg_variance8x8_ssse3, 0),
+        SubpelAvgVarianceParams(3, 2, &aom_sub_pixel_avg_variance8x4_ssse3, 0),
+        SubpelAvgVarianceParams(2, 3, &aom_sub_pixel_avg_variance4x8_ssse3, 0),
+        SubpelAvgVarianceParams(2, 2, &aom_sub_pixel_avg_variance4x4_ssse3,
+                                0)));
 #endif  // HAVE_SSSE3
 
 #if HAVE_AVX2
@@ -1267,14 +1339,16 @@ INSTANTIATE_TEST_CASE_P(
 
 INSTANTIATE_TEST_CASE_P(
     AVX2, AvxSubpelVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_variance64x64_avx2, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_variance32x32_avx2, 0)));
+    ::testing::Values(
+        SubpelVarianceParams(6, 6, &aom_sub_pixel_variance64x64_avx2, 0),
+        SubpelVarianceParams(5, 5, &aom_sub_pixel_variance32x32_avx2, 0)));
 
 INSTANTIATE_TEST_CASE_P(
     AVX2, AvxSubpelAvgVarianceTest,
     ::testing::Values(
-        make_tuple(6, 6, &aom_sub_pixel_avg_variance64x64_avx2, 0),
-        make_tuple(5, 5, &aom_sub_pixel_avg_variance32x32_avx2, 0)));
+        SubpelAvgVarianceParams(6, 6, &aom_sub_pixel_avg_variance64x64_avx2, 0),
+        SubpelAvgVarianceParams(5, 5, &aom_sub_pixel_avg_variance32x32_avx2,
+                                0)));
 #endif  // HAVE_AVX2
 
 #if HAVE_NEON
@@ -1298,10 +1372,11 @@ INSTANTIATE_TEST_CASE_P(
 
 INSTANTIATE_TEST_CASE_P(
     NEON, AvxSubpelVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_variance64x64_neon, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_variance32x32_neon, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_variance16x16_neon, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_variance8x8_neon, 0)));
+    ::testing::Values(
+        SubpelVarianceParams(6, 6, &aom_sub_pixel_variance64x64_neon, 0),
+        SubpelVarianceParams(5, 5, &aom_sub_pixel_variance32x32_neon, 0),
+        SubpelVarianceParams(4, 4, &aom_sub_pixel_variance16x16_neon, 0),
+        SubpelVarianceParams(3, 3, &aom_sub_pixel_variance8x8_neon, 0)));
 #endif  // HAVE_NEON
 
 #if HAVE_MSA
@@ -1336,34 +1411,36 @@ INSTANTIATE_TEST_CASE_P(
 
 INSTANTIATE_TEST_CASE_P(
     MSA, AvxSubpelVarianceTest,
-    ::testing::Values(make_tuple(2, 2, &aom_sub_pixel_variance4x4_msa, 0),
-                      make_tuple(2, 3, &aom_sub_pixel_variance4x8_msa, 0),
-                      make_tuple(3, 2, &aom_sub_pixel_variance8x4_msa, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_variance8x8_msa, 0),
-                      make_tuple(3, 4, &aom_sub_pixel_variance8x16_msa, 0),
-                      make_tuple(4, 3, &aom_sub_pixel_variance16x8_msa, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_variance16x16_msa, 0),
-                      make_tuple(4, 5, &aom_sub_pixel_variance16x32_msa, 0),
-                      make_tuple(5, 4, &aom_sub_pixel_variance32x16_msa, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_variance32x32_msa, 0),
-                      make_tuple(5, 6, &aom_sub_pixel_variance32x64_msa, 0),
-                      make_tuple(6, 5, &aom_sub_pixel_variance64x32_msa, 0),
-                      make_tuple(6, 6, &aom_sub_pixel_variance64x64_msa, 0)));
+    ::testing::Values(
+        SubpelVarianceParams(2, 2, &aom_sub_pixel_variance4x4_msa, 0),
+        SubpelVarianceParams(2, 3, &aom_sub_pixel_variance4x8_msa, 0),
+        SubpelVarianceParams(3, 2, &aom_sub_pixel_variance8x4_msa, 0),
+        SubpelVarianceParams(3, 3, &aom_sub_pixel_variance8x8_msa, 0),
+        SubpelVarianceParams(3, 4, &aom_sub_pixel_variance8x16_msa, 0),
+        SubpelVarianceParams(4, 3, &aom_sub_pixel_variance16x8_msa, 0),
+        SubpelVarianceParams(4, 4, &aom_sub_pixel_variance16x16_msa, 0),
+        SubpelVarianceParams(4, 5, &aom_sub_pixel_variance16x32_msa, 0),
+        SubpelVarianceParams(5, 4, &aom_sub_pixel_variance32x16_msa, 0),
+        SubpelVarianceParams(5, 5, &aom_sub_pixel_variance32x32_msa, 0),
+        SubpelVarianceParams(5, 6, &aom_sub_pixel_variance32x64_msa, 0),
+        SubpelVarianceParams(6, 5, &aom_sub_pixel_variance64x32_msa, 0),
+        SubpelVarianceParams(6, 6, &aom_sub_pixel_variance64x64_msa, 0)));
 
 INSTANTIATE_TEST_CASE_P(
     MSA, AvxSubpelAvgVarianceTest,
-    ::testing::Values(make_tuple(6, 6, &aom_sub_pixel_avg_variance64x64_msa, 0),
-                      make_tuple(6, 5, &aom_sub_pixel_avg_variance64x32_msa, 0),
-                      make_tuple(5, 6, &aom_sub_pixel_avg_variance32x64_msa, 0),
-                      make_tuple(5, 5, &aom_sub_pixel_avg_variance32x32_msa, 0),
-                      make_tuple(5, 4, &aom_sub_pixel_avg_variance32x16_msa, 0),
-                      make_tuple(4, 5, &aom_sub_pixel_avg_variance16x32_msa, 0),
-                      make_tuple(4, 4, &aom_sub_pixel_avg_variance16x16_msa, 0),
-                      make_tuple(4, 3, &aom_sub_pixel_avg_variance16x8_msa, 0),
-                      make_tuple(3, 4, &aom_sub_pixel_avg_variance8x16_msa, 0),
-                      make_tuple(3, 3, &aom_sub_pixel_avg_variance8x8_msa, 0),
-                      make_tuple(3, 2, &aom_sub_pixel_avg_variance8x4_msa, 0),
-                      make_tuple(2, 3, &aom_sub_pixel_avg_variance4x8_msa, 0),
-                      make_tuple(2, 2, &aom_sub_pixel_avg_variance4x4_msa, 0)));
+    ::testing::Values(
+        SubpelAvgVarianceParams(6, 6, &aom_sub_pixel_avg_variance64x64_msa, 0),
+        SubpelAvgVarianceParams(6, 5, &aom_sub_pixel_avg_variance64x32_msa, 0),
+        SubpelAvgVarianceParams(5, 6, &aom_sub_pixel_avg_variance32x64_msa, 0),
+        SubpelAvgVarianceParams(5, 5, &aom_sub_pixel_avg_variance32x32_msa, 0),
+        SubpelAvgVarianceParams(5, 4, &aom_sub_pixel_avg_variance32x16_msa, 0),
+        SubpelAvgVarianceParams(4, 5, &aom_sub_pixel_avg_variance16x32_msa, 0),
+        SubpelAvgVarianceParams(4, 4, &aom_sub_pixel_avg_variance16x16_msa, 0),
+        SubpelAvgVarianceParams(4, 3, &aom_sub_pixel_avg_variance16x8_msa, 0),
+        SubpelAvgVarianceParams(3, 4, &aom_sub_pixel_avg_variance8x16_msa, 0),
+        SubpelAvgVarianceParams(3, 3, &aom_sub_pixel_avg_variance8x8_msa, 0),
+        SubpelAvgVarianceParams(3, 2, &aom_sub_pixel_avg_variance8x4_msa, 0),
+        SubpelAvgVarianceParams(2, 3, &aom_sub_pixel_avg_variance4x8_msa, 0),
+        SubpelAvgVarianceParams(2, 2, &aom_sub_pixel_avg_variance4x4_msa, 0)));
 #endif  // HAVE_MSA
 }  // namespace
diff --git a/third_party/aom/test/warp_filter_test.cc b/third_party/aom/test/warp_filter_test.cc
index 2e4e6c32b..0353c074e 100644
--- a/third_party/aom/test/warp_filter_test.cc
+++ b/third_party/aom/test/warp_filter_test.cc
@@ -28,9 +28,11 @@ INSTANTIATE_TEST_CASE_P(
     SSE2, AV1WarpFilterTest,
     libaom_test::AV1WarpFilter::BuildParams(av1_warp_affine_sse2));
 
+#if HAVE_SSSE3
 INSTANTIATE_TEST_CASE_P(
     SSSE3, AV1WarpFilterTest,
     libaom_test::AV1WarpFilter::BuildParams(av1_warp_affine_ssse3));
+#endif
 
 #if CONFIG_HIGHBITDEPTH && HAVE_SSSE3
 TEST_P(AV1HighbdWarpFilterTest, CheckOutput) {
diff --git a/third_party/aom/test/warp_filter_test_util.cc b/third_party/aom/test/warp_filter_test_util.cc
index ac7518f47..ab3c153e7 100644
--- a/third_party/aom/test/warp_filter_test_util.cc
+++ b/third_party/aom/test/warp_filter_test_util.cc
@@ -112,6 +112,7 @@ void AV1WarpFilterTest::RunCheckOutput(warp_affine_func test_impl) {
   uint8_t *output2 = new uint8_t[output_n];
   int32_t mat[8];
   int16_t alpha, beta, gamma, delta;
+  ConvolveParams conv_params = get_conv_params(0, 0, 0);
 
   // Generate an input block and extend its borders horizontally
   for (i = 0; i < h; ++i)
@@ -126,10 +127,10 @@ void AV1WarpFilterTest::RunCheckOutput(warp_affine_func test_impl) {
       for (sub_y = 0; sub_y < 2; ++sub_y) {
         generate_model(mat, &alpha, &beta, &gamma, &delta);
         av1_warp_affine_c(mat, input, w, h, stride, output, 32, 32, out_w,
-                          out_h, out_w, sub_x, sub_y, 0, alpha, beta, gamma,
-                          delta);
+                          out_h, out_w, sub_x, sub_y, &conv_params, alpha, beta,
+                          gamma, delta);
         test_impl(mat, input, w, h, stride, output2, 32, 32, out_w, out_h,
-                  out_w, sub_x, sub_y, 0, alpha, beta, gamma, delta);
+                  out_w, sub_x, sub_y, &conv_params, alpha, beta, gamma, delta);
 
         for (j = 0; j < out_w * out_h; ++j)
           ASSERT_EQ(output[j], output2[j])
@@ -248,6 +249,7 @@ void AV1HighbdWarpFilterTest::RunCheckOutput(
   uint16_t *output2 = new uint16_t[output_n];
   int32_t mat[8];
   int16_t alpha, beta, gamma, delta;
+  ConvolveParams conv_params = get_conv_params(0, 0, 0);
 
   // Generate an input block and extend its borders horizontally
   for (i = 0; i < h; ++i)
@@ -265,10 +267,11 @@ void AV1HighbdWarpFilterTest::RunCheckOutput(
         generate_model(mat, &alpha, &beta, &gamma, &delta);
 
         av1_highbd_warp_affine_c(mat, input, w, h, stride, output, 32, 32,
-                                 out_w, out_h, out_w, sub_x, sub_y, bd, 0,
-                                 alpha, beta, gamma, delta);
+                                 out_w, out_h, out_w, sub_x, sub_y, bd,
+                                 &conv_params, alpha, beta, gamma, delta);
         test_impl(mat, input, w, h, stride, output2, 32, 32, out_w, out_h,
-                  out_w, sub_x, sub_y, bd, 0, alpha, beta, gamma, delta);
+                  out_w, sub_x, sub_y, bd, &conv_params, alpha, beta, gamma,
+                  delta);
 
         for (j = 0; j < out_w * out_h; ++j)
           ASSERT_EQ(output[j], output2[j])
diff --git a/third_party/aom/test/warp_filter_test_util.h b/third_party/aom/test/warp_filter_test_util.h
index 651a9f830..4828f3785 100644
--- a/third_party/aom/test/warp_filter_test_util.h
+++ b/third_party/aom/test/warp_filter_test_util.h
@@ -31,8 +31,8 @@ typedef void (*warp_affine_func)(const int32_t *mat, const uint8_t *ref,
                                  uint8_t *pred, int p_col, int p_row,
                                  int p_width, int p_height, int p_stride,
                                  int subsampling_x, int subsampling_y,
-                                 int ref_frm, int16_t alpha, int16_t beta,
-                                 int16_t gamma, int16_t delta);
+                                 ConvolveParams *conv_params, int16_t alpha,
+                                 int16_t beta, int16_t gamma, int16_t delta);
 
 typedef std::tr1::tuple<int, int, int, warp_affine_func> WarpTestParam;
 
@@ -60,11 +60,14 @@ class AV1WarpFilterTest : public ::testing::TestWithParam<WarpTestParam> {
 
 #if CONFIG_HIGHBITDEPTH
 namespace AV1HighbdWarpFilter {
-typedef void (*highbd_warp_affine_func)(
-    const int32_t *mat, const uint16_t *ref, int width, int height, int stride,
-    uint16_t *pred, int p_col, int p_row, int p_width, int p_height,
-    int p_stride, int subsampling_x, int subsampling_y, int bd, int ref_frm,
-    int16_t alpha, int16_t beta, int16_t gamma, int16_t delta);
+typedef void (*highbd_warp_affine_func)(const int32_t *mat, const uint16_t *ref,
+                                        int width, int height, int stride,
+                                        uint16_t *pred, int p_col, int p_row,
+                                        int p_width, int p_height, int p_stride,
+                                        int subsampling_x, int subsampling_y,
+                                        int bd, ConvolveParams *conv_params,
+                                        int16_t alpha, int16_t beta,
+                                        int16_t gamma, int16_t delta);
 
 typedef std::tr1::tuple<int, int, int, int> HighbdWarpTestParam;