Then the request should be edited in the first post. Currently the request is to increase the simultaneous image processing limit higher than 10.
I don’t think this has been demonstrated. Are we sure the appropriate cores designed to do this kind of work in the GPU aren’t fully utilized?
I haven’t seen any evidence that they aren’t. At least not on windows.
Anyway suggest you clarify exactly what it is you’re asking for with a headline edit and an edit to the first post.
Hi Mike,
I don’t see a way to edit the post - the only option is to delete it and start another one.
I’m not doing this work on Windows. I bought the Mac Studio specifically for the NNE (neural network engine) capability, and have been taken back by how lightly loaded the rest of the system is while processing a bunch of images with DeepPRIME XD:
This is the screenshot I was referring to. The GPU cluster is clearly not loaded and the CPU cluster is only periodically doing something that does not fully load them. The fan on this system does not spin high enough for me to hear it. The system is just not stressed enough. I’ve also said elsewhere that I don’t have a tool to measure the neural network engine capability, as that’s what should be loaded at 100%… when I get my hands on such a tool, I’ll post the result here.
You can use NVidia inspector if you want to check. (Assuming you’ve got a NVidia card. Don’t know about AMD).
