I passed through two of my Titan V cards using Hyper-v as hypervisor from within Windows Server 2016. I want to set up an Ubuntu 16.04 VM. From past experience the Gforce consumer card drivers cannot be installed within a virtual machine using Hyper-v because the Nvidia consumer card driver recognize that the environment is a virtual machine. However, Tesla card drivers install just fine within a virtual machine which seems to be a deliberate decision by Nvidia.
My question is: Do the Titan V drivers install just fine within a virtual machine just as the Tesla drivers do or will the driver install be rejected because I attempt to install them within a VM? Obviously the Titan V is marketed and advertised as a compute card rather than a gaming/consumer card and hence I would expect the drivers to install just fine inside a virtual machine, but I like to have this confirmed by someone who has already done so or by someone from Nvidia.
Any hints or thoughts?
P.S.: I am aware that some hypervisors such as KVM allow the setting of a cpu-flag for VMs in order to simulate a non-vm environment, however, MS’s Hyper-v does not allow such. But I am forced to use Hyper-v, hence my above question.
Thanks