[Webkit-unassigned] [Bug 170444] WebGPU: implement ComputeCommandEncoder and related components

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=170444

--- Comment #8 from Myles C. Maxfield <mmaxfield at apple.com> ---
I think this is good and we probably want to accept this.

However, we should probably be clear about some things:

- Our implementation of WebGPU is just an experiment. It is designed to help us understand what kind of performance is possible. It is also designed to help us understand any design issues which may appear.

- The WebKit project tries to be standards-compliant where possible. We are currently involved in the "GPU on the Web"[1] standardization group. This group is still very young, and does not have a draft specification document. Therefore, as this group develops, our implementation will probably change dramatically. The changes will probably include both fundamental design as well as details of implementation.

- The apparent direction of this patch is to mirror the Metal API in Javascript. This is the approach that our current implication has taken. However, we will switch approaches to, instead, pursue the work being done in the standardization group. Therefore, this patch is acceptable because it is an extension of what we have already done, but the existing approach, including this patch, will likely have to be dramatically changed soon.

[1] https://www.w3.org/community/gpu/

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170407/428af221/attachment.html>