If I intended to use 2 X Tesla P4 8GB per HyperVisor, utilizing Server 2016 DDA to map the cards 2 sessions hosts (1:1), would I still require GRID licensing?
As they’re single GPU and I’m using Server 2016 to do the DDA, I’m not utilizing the GRID software am I?
First time with this configuration so I could be wrong.
It doesn’t matter if you use vGPU or Passthrough. As long as you use Tesla boards for virtualization you will need vApps or QvDWS license depending on your use case.
Regards
Simon
Thanks for clarifying. I guess I still fail to understand why as you wouldn’t with single GPU card like a Quadro. If you’re using Server 2016 DDA implemented by Microsoft, what exactly are you licensing/getting from NVIDIA?
Since AMD/Radeon card’s are not nerfed in the same way as NVIDIA GeForce cards are, in theory any AMD/Radeon card can be passed through provided the underlyng hardware supports it. AMD Takeaways Cards with Graphics Core Next (GCN) will work ok with Windows Server to pass through to the guest VMs AMD integrated Vega graphics are a lot more useful for a server as the cores are on die making it space efficient. Unless you are running games, Vega is good enough Because this is just for messing about, I’m planning on picking up an RX 560 since it can be had for about $100-ish.