GPU para-virtualization in Hyper-V

I have just received an NVIDIA Tesla P40 having sourced it for a decent price with the intent of installing it into my Hyper-V server. It’s an older card, but it’s still a beast with 24GB of memory and 3840 CUDA cores. The card is gaining popularity in the AI community due to its high memory capacity and relatively low price, it is also passively cooled and an official HPE optional product (according to HPE DL380 G9 specs) and is therefore recognized by iLO and it’s temperature can be monitored as well.