Me and my colleagues are sharing server with multiple GPU nodes on which we run OpenCL code. We do not run any resource manager on top of it. We just make agreements that: &
