Hi.
I have read couple times the documentation found here
OGA Hybrid Execution Mode — Ryzen AI Software 1.3 documentation
however cannot understand if we get a AMD AI 395, like the Framework Desktop or the HP Z machine, and put M.2 to Oculink adaptor and connect a 7900XTX or a 7900W would be able to utilise Hybrid Execution Mode and ONNX optimised LLMs across all the devices? dGPU (7900XTX / 7900W), iGPU (8060S) and NPU?
If we can use it, would this work also if we use USB Mesh adaptor (or via Ethernet) to connect 2-4 AMD AI 395 PCBs each with M.2 to Oculink and a 7900XTX/7900W dGPU connected also to each one?
