[Webkit-unassigned] [Bug 100341] [META][GTK] Implement threaded model of Coordinated Graphics

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Fri Jan 18 01:24:01 PST 2013


https://bugs.webkit.org/show_bug.cgi?id=100341





--- Comment #16 from Jae Hyun Park <jae.park at company100.net>  2013-01-18 01:25:47 PST ---
(In reply to comment #10)
> Created an attachment (id=182799)
 --> (https://bugs.webkit.org/attachment.cgi?id=182799&action=review) [details]
> Proof-of-concept prototype. Not for commit.
> 
> Hi, I have been working on a proof-of-concept prototype that implements a thread for compositing layers in the TextureMapperLayer. I am not fully aware of the final implementation proposed in this meta bug, but I believe the prototype follows roughly the same path.
> 
> https://github.com/dnomi/webkit/tree/threaded-coordinated-gfx
> 
> The idea is to just run TextureMapperLayer::paint() on a new thread, and to check the performance improvement that the threaded coordinated graphics will potentially give us on the WebKit2 GTK+ port.
> 
> So the initial results are encouraging, suggesting an improvement of about 20%~30% in the frame rate when displaying some pages with WebGL on a Linux PC (e.g. http://code.google.com/p/webglsamples).
> 
> I have also uploaded the patch here for reference and review. Comments are welcome.

Hi, Denis.

Thanks for sharing your prototype with us!

Our design of Threaded Coordinated Graphics is quite different from your prototype. As you can see in the design document, our plan is to refactor and reuse current Coordinated Graphics System by implementing communication mechanism between threads, instead of IPC. That's why we call this "Threaded Coordinated Graphics". 

It's still great to see a performance improvement just by running TextureMapperLayer::paint() on a new thread. 

Thank you for your interest in this project.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.


More information about the webkit-unassigned mailing list