How to share memory between the WebGL and GPU process

[jerry.m...@gmail.com]

Hi, all

 

    Now I'm focusing on the share mechanism in chrome and webkit, so I want to know the share mechanism between the render process and GPU process when I create a 3D element in web in chrome. 

   Anyone would tell me what's the detail procedure in the the application scenarios of creating 3D games, especially How to interacte between the WebGL, GPU process and the last display stage.

 

Cheers,

Jerry

[Gregg Tavares]

There's a bunch of gpu design docs here

http://dev.chromium.org/developers/design-documents

In particular

http://dev.chromium.org/developers/design-documents/gpu-command-buffer

http://dev.chromium.org/developers/design-documents/rendering-architecture-diagrams

http://dev.chromium.org/developers/design-documents/gpu-accelerated-compositing-in-chrome

Do those answer your question?

--
--
Chromium Developers mailing list: chromi...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-dev
 
 
 

[jerry.m...@gmail.com]

Thanks, Gregg

 

From your link, I know how does the communication between render process and GPU processing.

 

By the way, I have aother question:

In the hardware render patch, the GPU process does the hardware render and then call the D3D APIs to display it.  In this case, I split this procedure into two stages, the first stage is that GPU process uses the GPU capability to do some, such as calling the GL to draw some, the second stage is that GPU process uses the D3D(on Windows platform) to display the web.  in the first stage, the output of processing is in the GPU, but it is not in the GPU surface which is used to show in screen by D3D on Windows. but the second stage needs a D3D surface as the input on Windows. 

 

So I want to know how to do this communication between the GL output and the D3D input in GPU process. Is there a copy from the output of GL into system memory, then do another copy from system memroy into D3D surface ?

 

Cheers,

Jerry

在 2013年4月27日星期六UTC+8下午11时44分15秒，Gregg Tavares写道：

[Gregg Tavares]

On Sat, Apr 27, 2013 at 8:05 PM, <jerry.m...@gmail.com> wrote:

Thanks, Gregg

 

From your link, I know how does the communication between render process and GPU processing.

 

By the way, I have aother question:

In the hardware render patch, the GPU process does the hardware render and then call the D3D APIs to display it.  In this case, I split this procedure into two stages, the first stage is that GPU process uses the GPU capability to do some, such as calling the GL to draw some, the second stage is that GPU process uses the D3D(on Windows platform) to display the web.  in the first stage, the output of processing is in the GPU, but it is not in the GPU surface which is used to show in screen by D3D on Windows. but the second stage needs a D3D surface as the input on Windows. 

 

So I want to know how to do this communication between the GL output and the D3D input in GPU process. Is there a copy from the output of GL into system memory, then do another copy from system memroy into D3D surface ?

GL is emulated using ANGLE (http://code.google.com/p/angleprogroject) so there is no GL being used on windows, it's all D3D so there is no need to copy anything, it's already in a D3D surface.