Dzmitry Malyshau

These are all very interesting ideas! For me, there is still an elephant in the room: how much would that give us versus the current approach (where the user would...

@bbernhar I'm curious if this is just a Dawn performance issue that is on the radar, versus a conceptual problem with bind groups in the API.

@benvanik there is a lot to digest in your post. If you are interested to communicate your feedback clearer, consider filing a discussion and providing a bit more information on...

> If dispatch A and dispatch B work on mutually exclusive subranges of a larger buffer then you'd want them to be able to overlap (run concurrently and have B...

Thank you for the investigation! So the conclusion I read is: - when recording a GPUBundle, we assume a clean state and validate accordingly - after invoking a bundle from...

@devshgraphicsprogramming right, we aren't even considering adding transform feedback at this point. I just brought it up for picture completeness.

Triage: we could easily have a flag in `ComputePassDescriptor` that turns the whole pass into a sync scope.

Solution B, as I mentioned on the call, needs to have a more complex API in order to support more than a single resource being aliased (e.g. "A | (B...

It's still a ding against an explicit API. In some cases, for example, a render target can't be used simultaneously on 2 queues, but it also doesn't care which queue...

I tried to capture the debate to date. See [plain](https://kvark.github.io/webgpu-debate/MultiQueue.html) and [graph](https://kvark.github.io/webgpu-debate/MultiQueue.component.html) views. PRs to https://github.com/kvark/webgpu-debate are welcome!