I want to pass a Quadro GPU to a container on a Windows Server host.
What I want to change in our build infrastructure are idleing machines. We got Linux machines and Windows machines. Different Linux distributions. Some of them are idleing most of the time. So I thought about moving everything, Linux and Windows, in a container and run the GPU tests (CUDA and OpenGL) there. Since Linux containers don’t have access to the GPU I was wondering if I can make use of and assign the GPU to any of my containers at runtime?
If this is possible, can you please explain how this is done?