Without GPU Passthough, it’s difficult to use Qubes OS as a daily driver but I was wondering if this counts as a simple easy way of making GPU Passthrough secure on Qubes OS.
Since so many CPU’s come with integrated graphics nowadays, wouldn’t it be possible to have the host and most Qubes only use the integrated graphics and completely ignore the dedicated GPU and only have the dedicated GPU tied to a certain Qube that really needs it? This way the GPU doesn’t need to interact with the host and other Qubes preventing any malware from spreading. Would this be both possible and secure?
I found this guide incredibly helpful to passthrough a discrete graphics card to a qube Create a Gaming HVM
The limitation right now is there is no ideal way to display the discrete graphics qube on the same monitor as your other qubes, but it’s better than nothing.