I already found the support article which basically states Parallels will detect an eGPU that's working with an external display, and use that for graphics acceleration. However, that's not what I'm looking for. I am looking to attach an Nvidia (not support in macOS) eGPU directly to the virtual machine via device passthrough. Once connected, Nvidia drivers would be installed within the virtual machine itself. This would allow/enable CUDA functions to work perfectly within the VM as well as enable DirectX 12 gaming... again directly through the VM as the Nvidia eGPU is directly connected to the VM. It would be great if any consumer-grade nVidia GPU worked, but its understandable if it is limited to the Titan and Quadro series due to Nvidia licensing. However; the functionality would be a big boon for this software and would offer quite a few benefits to those who cannot use nVidia products on their Catalina devices. Additionally, this would prevent folks from having to dual-boot or triple-boot multiple OSs and have that direct access to the GPU in the VM. Is there currently a way for this to work, or a way to add a feature request? Current setup: - 2019 MacBook Pro 16" Core i9 + RX 5500M - Razer Core X Thunderbolt 3 enclosure - nVidia Titan Black 6GB. My device does not support High Sierra, which was the last version to work with nVidia web drivers. I already use Windows with Bootcamp, but that functionality for a VM to have GPU access, and especially to boot High Sierra with GPU support, is also lacking. Summary: Requesting to allow direct access to a Thunderbolt 3 nVidia eGPU from Windows 10 and High Sierra guest virtual machines, bypassing Catalina (as the GPUs are not supported in Catalina).
