planar blend use signed images

unit_test/planar_test.cc

 /*
  *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS. All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
@@ skipping 1145 matching lines @@
   EXPECT_LE(max_diff, 1);
 }
 
 TEST_F(LibYUVPlanarTest, ARGBBlend_Opt) {
   int max_diff = TestBlend(benchmark_width_, benchmark_height_,
                            benchmark_iterations_,
                            disable_cpu_flags_, benchmark_cpu_info_, +1, 0);
   EXPECT_LE(max_diff, 1);
 }
 
+#ifdef HAS_BLENDPLANEROW_SSSE3
+
+static int TestBlendPlane(int width, int height, int benchmark_iterations,
+                          int invert, int off) {
+  int has_ssse3 = TestCpuFlag(kCpuHasSSSE3);
+  if (width < 1) {
+    width = 1;
+  }
+  const int kBpp = 1;
+  const int kStride = width * kBpp;
+  align_buffer_64(src_argb_a, kStride * height + off);
+  align_buffer_64(src_argb_b, kStride * height + off);
+  align_buffer_64(src_argb_alpha, kStride * height + off);
+  align_buffer_64(dst_argb_c, kStride * height);
+  align_buffer_64(dst_argb_opt, kStride * height);
+  for (int i = 0; i < kStride * height; ++i) {
+    src_argb_a[i + off] = (fastrand() & 0xff);
+    src_argb_b[i + off] = (fastrand() & 0xff);
+    src_argb_alpha[i + off] = (fastrand() & 0xff);
+  }
+  memset(dst_argb_c, 255, kStride * height);
+  memset(dst_argb_opt, 255, kStride * height);
+
+  BlendPlaneRow_C(src_argb_a + off,
+                  src_argb_b + off,
+                  src_argb_alpha + off,
+                  dst_argb_c,
+                  width * height);
+  for (int i = 0; i < benchmark_iterations; ++i) {
+    if (has_ssse3) {
+      BlendPlaneRow_SSSE3(src_argb_a + off,
+                          src_argb_b + off,
+                          src_argb_alpha + off,
+                          dst_argb_opt,
+                          width * height);
+    } else {
+      BlendPlaneRow_C(src_argb_a + off,
+                      src_argb_b + off,
+                      src_argb_alpha + off,
+                      dst_argb_opt,
+                      width * height);
+    }
+  }
+  int max_diff = 0;
+  for (int i = 0; i < kStride * height; ++i) {
+    EXPECT_EQ(dst_argb_c[i], dst_argb_opt[i]);
+  }
+  free_aligned_buffer_64(src_argb_a);
+  free_aligned_buffer_64(src_argb_b);
+  free_aligned_buffer_64(dst_argb_c);
+  free_aligned_buffer_64(dst_argb_opt);
+  return max_diff;
+}
+
+TEST_F(LibYUVPlanarTest, BlendPlane_Opt) {
+  int max_diff = TestBlendPlane(benchmark_width_, benchmark_height_,
+                                benchmark_iterations_,
+                                +1, 0);
+}
+#endif
+
 TEST_F(LibYUVPlanarTest, TestAffine) {
   SIMD_ALIGNED(uint8 orig_pixels_0[1280][4]);
   SIMD_ALIGNED(uint8 interpolate_pixels_C[1280][4]);
 
   for (int i = 0; i < 1280; ++i) {
     for (int j = 0; j < 4; ++j) {
       orig_pixels_0[i][j] = i;
     }
   }
 
@@ skipping 1169 matching lines @@
 
 TEST_F(LibYUVPlanarTest, SetPlane_Opt) {
   int max_diff = TestARGBRect(benchmark_width_, benchmark_height_,
                               benchmark_iterations_,
                               disable_cpu_flags_, benchmark_cpu_info_,
                               +1, 0, 1);
   EXPECT_EQ(0, max_diff);
 }
 
 }  // namespace libyuv