nvidia grid M60 or M10 + windows 2016 + VDI + solidwrks

I want set up an new server only with windows 2016 and a grid card Tesla M60 or M10. I want have 5 - 10 VDI desktops with WIN10 and solidworks.
There are no documents to see how to do this with the new server system windows 2016. Have anybody this done yet? Can somebody give me some information how to do the best way. Are there some best practice guidelines?

Is do that a good idea or is it nonsense? Should i do this better with vmware or citrix?

I hope somebody have a few guidelines or good advices for me to realice this with windows 2016.
I hope also somebody from nvidia is writing some sheets also for windows 2016 in the same way there are sheets for vmware or citrix.

thanks Paul


Microsoft with Server 2016 and DDA now supports GPU Passthrough but no vGPU. That means you can use for example the 2 GPUs from a M60 or the 4 GPUs from a M10 to passthrough to the given VM. So depending on the user density you desire it might not be sufficient.
With Citrix XenServer or VMWare vSphere you can use vGPU which gives you a much better density.

Best regards


Hello, i think your reply is not correct. Please see this http://social.technet.microsoft.com/wiki/contents/articles/31771.experience-guide-for-enabling-opengl-support-for-vgpu-in-server-2016.aspx

I hope i can bocome the right answer from nvidia specialist or from someone who have alright tested this. Perhaps also from microsoft because this is very powerfull news to all in the VDI CAD bussiness. There is know need anymore for vmware or citrix i think.
Or is this thinking from me not the right way. What arguments there are for vmware or citrix. Documents?

There is some misunderstanding. NVidia vGPU is not RemoteFX vGPU (https://en.wikipedia.org/wiki/RemoteFX). You can read https://virtuallyvisual.wordpress.com/2017/01/18/nvidia-grid-and-microsoft-windows-server-oss-and-hyper-v/.

Ok thanks. But what is the best way to have 10 VDI desktops with solidworks. Is it a good way to use only windows with remote fx or better to realize with vmware horizon.

Either ESX or XenServer will support NVIDIA vGPU. You can use Teradici Cloud Access Software( http://www.teradici.com/products-and-solutions/pcoip-products/cloud-access-software ) with either of them or you can go with Horizon or XenDesktop.

[i]My opinion: The answer depends on yours and/or users expectation and/or network setup (CAPEX, OPEX, distance/latency, UX - user experience …). You can probably prepare some $$$ for POC and ask the experts with mixed knowledge to help (like @rspruijt @_POPPELGAARD @msandbu @graton_cleric @WorkingHardInIT @RBRConecto … or theirs blogs):

  • NVidia NGCA
  • Microsoft MVP
  • Citrix CTP
  • VMware vExpert


Just checking to see if the information in this thread is still current.

Does Microsoft Server 2016 support vGPU now?

I’d like to build a Hyper-V VDI solution (without VMware or Citrix) using an M60 and need more than 4 GPUs.

This is still valid. No vGPU support with Hyper-V at this stage but this will hopefully change in 2018.