Integrate VDA with WebRTC.

content/common/gpu/media/exynos_video_decode_accelerator.h

Index: content/common/gpu/media/exynos_video_decode_accelerator.h
diff --git a/content/common/gpu/media/exynos_video_decode_accelerator.h b/content/common/gpu/media/exynos_video_decode_accelerator.h
index 0eb3299c018e4465313ed35c00a55994dc80e053..2c817a5803156f30ff4c83fa596598768652c6ce 100644
--- a/content/common/gpu/media/exynos_video_decode_accelerator.h
+++ b/content/common/gpu/media/exynos_video_decode_accelerator.h
@@ -85,12 +85,12 @@ class CONTENT_EXPORT ExynosVideoDecodeAccelerator :
  private:
   // These are rather subjectively tuned.
   enum {
-    kMfcInputBufferCount = 8,
+    kMfcInputBufferCount = 12,

[sheu, 2013/06/14 04:15:37]

Each one of these are 512 kB, so it's not so bad to increase this size; just
don't overdo it.

     kMfcInputBufferMaxSize = 512 * 1024,
-    kGscInputBufferCount = 4,
+    kGscInputBufferCount = 8,

[sheu, 2013/06/14 04:15:37]

These buffers are basically placeholders for MFC buffers, so increasing
kGscInputBufferCount should not much affect performance.  The bottleneck is in
the MFC.

     // Number of output buffers to use for each VDA stage above what's required
     // by the decoder (e.g. DPB size, in H264).
-    kDpbOutputBufferExtraCount = 3,
+    kDpbOutputBufferExtraCount = 8,

[sheu, 2013/06/14 04:15:37]

The revert of the tearing hack landed.  Try the peformance again without
increasing this buffer count.

[wuchengli, 2013/06/18 16:11:23]

After rebasing chrome and chrome OS source, the performance became better. I
don't know why. I reverted all the changes in this file.

     // We're continuing to have issues with synchronization between Mali 3D and
     // Exynos video decode, so we buffer this many extra GSC output buffers in
     // the GSC free output buffers queue, to add a safety margin.