#Update transform position on GPU
8 messages · Page 1 of 1 (latest)
I actually don't need the data on ram at all, i just need to get it so that I can pass it to the extract render step next iteration. Is it possible that I create some data that lives solely on GPU?
why would you need that? Are you having performance issues?
Yes, it would definitely help as I am trying to simulate a couple of millions entities that move every frame based on some logic.
It's mostly used for actually simulating
I think the most logical thing then is to create a custom compute node in the render graph. That will give you access to the wgpu api and is where you can create buffers
you look in the examples to see how it is done with a custom image, but a buffer shouldn't be that different
Yeah, i have done it. But now I need to get the data from the GPU back to CPU. Is there any example of this?