I have been using xcp-ng for a while now and Xen prior to that, but just starting to look into GPU with the need for ability to run AI and Machine Learning for developers that may spin up their own machines and want to use GPU processing.
I saw AWS provides GPU instances with 4, 8, 16 vGPU cores and we believe that if we can provide that in our environment, then the internal development can be done with similar specs.
So we purchased a single Nvidia Graphics card from eBay, and realized after setting up a dev server that we can only pass through the entire card, and not vGPU... due to Nvidia being proprietary... so I am now trying to figure out which GPU will work for vGPU such that we can do ML and deep learning in our VM.
I read that AMD and Intel would work, but would a Radeon Vega 56 work with vGPU? How many cores will it provide etc.
Can someone provide some general guidance and I am not sure how to proceed - GPU as well as AI/ML are new to me..