I have a Nvidia Tesla P40 GPU set up on my Windows 11 machine, and I'm using it through WSL2 for applications that need its CUDA capabilities. These applications are running in various CUDA toolkit versions within Conda environments. To achieve this, I have to operate the P40 in WDDM mode since WSL2 can't recognize the GPU when it's in TCC mode.
While in WDDM mode, the P40 appears to have a virtual VGA port attached to a virtual display. In certain scenarios, Windows defaults to using this virtual display as the main one, ignoring the actual display connected to another graphics card in my system.
Is there a way to convince Windows that nothing is connected to the virtual VGA on the P40? Could I possibly modify a registry entry to make this change? I'd appreciate any help on this matter.