Hello! I'm a long-standing user of MPSCNN framework, it usually works fine but during the implementation of one my recent networks I started to get this errors: 2023-01-06 00:17:46.017908+0600 -[44642:879994] [GPUDebug] Invalid device load executing kernel function "cnnConvWinograd_8x8_3x3_32x32_256" encoder: "", dispatch: 0, at offset 120``` Also, strangely networks produces different results over the same inputs across multiple runs. I assume there are some race conditions inside which causes that. Is it possible to somehow enforce MPSCNN to use some other implementation of convolution? Or I'm stuck with it forever?

Hi, all! I have a workload where I have a real-time renderer (which runs at up to 120FPS rate depending on the device's screen) and non-real-time kernel that should run in a loop but it can't handle the 60 (even 30) FPS requirement. So what I want to achieve is to create two MTLCommandQueues where the first will handle high-priority render tasks and maintain stable 120FPS rendering and the other one will be continuously fed with non-real-time kernel tasks. Can't find anything for this in the API. Is there anyway to achieve this?

Hi! I've install 3090 into my Mac Pro and Windows 10 as a second OS. I see that my card is pretty hot as the temperature keeps around 73C with 80% fan speed. I think increasing middle front fan speed will help a lot with it. However I can't find any info on how to achieve this. Can somebody help with it?

I see that Apple developer use Bundle.module in Resources and localization session, but it doesn't seem to be available in Xcode 12. Is that an unreleased addition, or a private extension of some sort?