Chromium code: Implementation of GPU->GPU Direct picture transfer, do not need to transfer through the CPU

Commit0c4e9d8781aea6e52fdb4a7aee978817910c67ea
Authordongseong.hwang Thu Jan 08 20:11:13 2015
Committercommit bot Thu Jan 08 20:12:02 2015

Media:optimize HW Video to 2D Canvas copy. Currently, when we draws GPUs decoded Video on accelerated 2D Canvas, chromiumreads back pixel from GPUs and then uploads th e Pixel to the GPU to make a skbitmap.it's so inefficient for both speed and battery. On the other hand, only androidcopies Gpu-gpu in this case, but Android doesn ' t has cache mechanism Whichskcanvasvideoren Derer provides. This CL makes all platforms the copy Gpu-gpu with the cache mechanism. Cache Mechanismis useful when 2d canvas draws a video frame many times.e.g. http://craftymind.com/factory/html5video/CanvasVideo.htmlIn addition, fix white video background on Android Loaded. Other Platformsdraw black background thanks to Skcanvasvideorenderer::P aint (). In detail of the changes;1. Implement gpu-gpu copy in Skcanvasvideorenderer::P aint () like previouswebmediaplayerandroid::p aint (). 2. Move duplicated GPU code on WEBMEDIAPLAYERIMPL and Webmediaplayerandroid toskcanvasvideorenderer.perf data on i5 IvyBridge Blink_perf.all:Canvas_draw-video-to-hw-accelerated-canvas-2d15.8x speed up:116.27 runs/s-1847.23 runs/snote:measure after disabling C Ache in skcanvasvideorendererbug=401058, 263667Review url:https://codereview.chromium.org/ 445013002cr-commit-position:refs/heads/[email protected]{#310577}

• Content/renderer/media/android/webmediaplayer_android.cc[diff]
• Content/renderer/media/android/webmediaplayer_android.h[diff]
• Content/renderer/media/webmediaplayer_ms.cc[diff]
• Content/renderer/render_frame_impl.cc[diff]
• Media/build.gn[diff]
• Media/deps[diff]
• Media/blink/build.gn[diff]
• Media/blink/deps[diff]
• Media/blink/media_blink.gyp[diff]
• Media/blink/webmediaplayer_impl.cc[diff]
• Media/blink/webmediaplayer_impl.h[diff]
• Media/blink/webmediaplayer_params.cc[diff]
• Media/blink/webmediaplayer_params.h[diff]
• Media/filters/context_3d.h[added-diff]
• Media/filters/skcanvas_video_renderer.cc[diff]
• Media/filters/skcanvas_video_renderer.h[diff]
• Media/filters/skcanvas_video_renderer_unittest.cc[diff]
• Media/media.gyp[diff]
• Mojo/services/html_viewer/webmediaplayer_factory.cc[diff]

Copy from video to picture, GPU to GPU direct, no CPU memory required?

//StaticvoidSkcanvasvideorenderer::copyvideoframetexturetogltexture (gpu::gles2::gles2interface* GL, VideoFrame* video_frame,unsigned intTextureunsigned intLevelunsigned intInternal_format,unsigned intTypeBOOLPremultiply_alpha,BOOLflip_y) {Dcheck (Video_frame && video_frame->format () = = Videoframe::native_texture);Constgpu::mailboxholder* Mailbox_holder = Video_frame->mailbox_holder ();         Dcheck (Mailbox_holder->texture_target = = Gl_texture_2d | |  Mailbox_holder->texture_target = = gl_texture_external_oes);  Gl->waitsyncpointchromium (Mailbox_holder->sync_point); UInt32 source_texture = Gl->createandconsumetexturechromium (Mailbox_holder->texture_target, mailbox_holder-& Gt;mailbox. Name);//The video is stored in a unmultiplied format, so premultiply  //if necessary.Gl->pixelstorei (Gl_unpack_premultiply_alpha_chromium, Premultiply_alpha);//application itself needs to take care of setting the right |flip_y|  //value down to get the expected result.  //"flip_y = = true" means to reverse the video orientation while  //"flip_y = = false" means to keep the intrinsic orientation.Gl->pixelstorei (Gl_unpack_flip_y_chromium, flip_y);   Gl->copytexturechromium (gl_texture_2d, Source_texture, TEXTURE, Level, Internal_format, type); Gl->pixelstorei (Gl_unpack_flip_y_chromium,false); Gl->pixelstorei (Gl_unpack_premultiply_alpha_chromium,false); Gl->deletetextures (1, &source_texture);  Gl->flush ();  Syncpointclientimpl Client (GL); Video_frame->updatereleasesyncpoint (&client);}

Main implementation code here, the main drawback is: from the code can be seen, still need GPU internal to GPU memory copy, do not zero-copy (if you can do the GPU memory pointer swap, or c++11 in the move structure, it is better). But obviously it's better than GPU-->CPU-->GPU's transfer efficiency.

In theory, GPU hardware-accelerated webview-to-view GPU snapshot snapshots can be delivered directly.

Chromium code: Implementation of GPU->GPU Direct picture transfer, do not need to transfer through the CPU