Use lea to remove 2 instructions from the inner loop of YUV Scale.

Use MMX2 to remove emms.
unittest has code to time yuv scale and convert 100 times so it takes about 1 second... useful for catching performance issues.  But the code is disabled.
BUG=23263
TEST=media_unittest should pass and run much faster: 28% faster than 32 bit version.

Review URL: http://codereview.chromium.org/273047

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@29019 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 30 insertions, 25 deletions

diff --git a/media/base/yuv_convert_unittest.cc b/media/base/yuv_convert_unittest.cc
index d75488a..00152b6 100644
--- a/media/base/yuv_convert_unittest.cc
+++ b/media/base/yuv_convert_unittest.cc
@@ -33,7 +33,6 @@ static const size_t kRGBSize = kWidth * kHeight * kBpp;
 static const size_t kRGBSizeConverted = kWidth * kHeight * kBpp;
 
 // Set to 100 to time ConvertYUVToRGB32.
-// This will take approximately 40 to 200 ms.
 static const int kTestTimes = 1;
 
 TEST(YUVConvertTest, YV12) {
@@ -99,16 +98,18 @@ TEST(YUVConvertTest, YV16) {
                                 reinterpret_cast<char*>(yuv_bytes.get()),
                                 static_cast<int>(kYUV16Size)));
 
-  // Convert a frame of YUV to 32 bit ARGB.
-  media::ConvertYUVToRGB32(yuv_bytes.get(),                             // Y
-                           yuv_bytes.get() + kWidth * kHeight,          // U
-                           yuv_bytes.get() + kWidth * kHeight * 3 / 2,  // V
-                           rgb_converted_bytes.get(),             // RGB output
-                           kWidth, kHeight,                       // Dimensions
-                           kWidth,                                // YStride
-                           kWidth / 2,                            // UVStride
-                           kWidth * kBpp,                         // RGBStride
-                           media::YV16);
+  for (int i = 0; i < kTestTimes; ++i) {
+    // Convert a frame of YUV to 32 bit ARGB.
+    media::ConvertYUVToRGB32(yuv_bytes.get(),                             // Y
+                             yuv_bytes.get() + kWidth * kHeight,          // U
+                             yuv_bytes.get() + kWidth * kHeight * 3 / 2,  // V
+                             rgb_converted_bytes.get(),  // RGB output
+                             kWidth, kHeight,            // Dimensions
+                             kWidth,                     // YStride
+                             kWidth / 2,                 // UVStride
+                             kWidth * kBpp,              // RGBStride
+                             media::YV16);
+  }
 
   unsigned int rgb_hash = DJB2Hash(rgb_converted_bytes.get(), kRGBSizeConverted,
                                    kDJB2HashSeed);
@@ -143,17 +144,19 @@ TEST(YuvScaleTest, YV12) {
   const size_t size_of_rgb_scaled = kScaledWidth * kScaledHeight * kBpp;
   scoped_array<uint8> rgb_scaled_bytes(new uint8[size_of_rgb_scaled]);
 
-  media::ScaleYUVToRGB32(yuv_bytes.get(),                             // Y
+  for (int i = 0; i < kTestTimes; ++i) {
+    media::ScaleYUVToRGB32(yuv_bytes.get(),                           // Y
                          yuv_bytes.get() + kWidth * kHeight,          // U
                          yuv_bytes.get() + kWidth * kHeight * 5 / 4,  // V
-                         rgb_scaled_bytes.get(),                // Rgb output
-                         kWidth, kHeight,                       // Dimensions
-                         kScaledWidth, kScaledHeight,           // Dimensions
-                         kWidth,                                // YStride
-                         kWidth / 2,                            // UvStride
-                         kScaledWidth * kBpp,                   // RgbStride
+                         rgb_scaled_bytes.get(),       // Rgb output
+                         kWidth, kHeight,              // Dimensions
+                         kScaledWidth, kScaledHeight,  // Dimensions
+                         kWidth,                       // YStride
+                         kWidth / 2,                   // UvStride
+                         kScaledWidth * kBpp,          // RgbStride
                          media::YV12,
                          media::ROTATE_0);
+  }
 
   unsigned int rgb_hash = DJB2Hash(rgb_scaled_bytes.get(), size_of_rgb_scaled,
                                    kDJB2HashSeed);
@@ -188,17 +191,19 @@ TEST(YuvScaleTest, YV16) {
   const size_t size_of_rgb_scaled = kScaledWidth * kScaledHeight * kBpp;
   scoped_array<uint8> rgb_scaled_bytes(new uint8[size_of_rgb_scaled]);
 
-  media::ScaleYUVToRGB32(yuv_bytes.get(),                             // Y
+  for (int i = 0; i < kTestTimes; ++i) {
+    media::ScaleYUVToRGB32(yuv_bytes.get(),                           // Y
                          yuv_bytes.get() + kWidth * kHeight,          // U
                          yuv_bytes.get() + kWidth * kHeight * 3 / 2,  // V
-                         rgb_scaled_bytes.get(),        // Rgb output
-                         kWidth, kHeight,               // Dimensions
-                         kScaledWidth, kScaledHeight,   // Dimensions
-                         kWidth,                        // YStride
-                         kWidth / 2,                    // UvStride
-                         kScaledWidth * kBpp,           // RgbStride
+                         rgb_scaled_bytes.get(),       // Rgb output
+                         kWidth, kHeight,              // Dimensions
+                         kScaledWidth, kScaledHeight,  // Dimensions
+                         kWidth,                       // YStride
+                         kWidth / 2,                   // UvStride
+                         kScaledWidth * kBpp,          // RgbStride
                          media::YV16,
                          media::ROTATE_0);
+  }
 
   unsigned int rgb_hash = DJB2Hash(rgb_scaled_bytes.get(), size_of_rgb_scaled,
                                    kDJB2HashSeed);