Wgpu: Memory usage keeps growing on the backend of dx12 of AMD R5-4600U
On the version 0.7. I set WGPU_BACKEND="dx12" and run the cube example cargo run --example cube on my notebook( the cpu is R5-4600U). The memory usage keeps growing fast. But it remained stable on the backend of vulkan.
On the other GPU like 1050Ti 1080TI, dx12 backend can work well. There is an issue in bevy about this Memory usage keeps growing
I think it's something wrong on the amd gpu when the backend is dx12.
linkstar0513
👍4
All 2 comments
I have seen this leak on dx12 on nvidia and amd as well.
cwfitzgerald
on 22 Feb 2021
I'm sure it's caused by the DX12 command allocator leak in gfx: https://github.com/gfx-rs/gfx/issues/3505
Haven't had time recently to implement a fix, but I think the issue is well documented if anyone would like to take a stab at it.
yzsolt
on 27 Feb 2021
👍3
Was this page helpful?
0 / 5 - 0 ratings
Related issues
kvark
路
15Comments
kvark
路
11Comments
z2oh
路
13Comments
MarekLg
路
39Comments
Aeledfyr
路
23Comments
Most helpful comment
I'm sure it's caused by the DX12 command allocator leak in
gfx: https://github.com/gfx-rs/gfx/issues/3505Haven't had time recently to implement a fix, but I think the issue is well documented if anyone would like to take a stab at it.