Is my NVIDIA M60, K1, K2, K5000 or my AMD Firepro W7000 supported by vSphere/XenServer/Other Hypervisor?

This is a question I continually see from customers and even those involved in deploying the technologies. The reasons behind the answer are quite complicated, but the answer itself is a simple yes or no, and the steps to find it out are very easy, once you know them.

The Basic Fact

GPU cards are not supported for “hypervisors”! GPUs are only supported for the combination of Server model, hypervisor version, GPU model and GPU technology, for example:

support

 

How to check if your GPU is supported for VMware vSphere

  • Is your combination of server, vSphere version, GPU technology and GPU on the vSphere/ESXi HCL (Hardware compatibility list), here: vSphere HCL/ESXi HCL
  • If yes it is fully supported.

 

How to check if your GPU is supported for Citrix XenServer

  • Is your combination of server, XenServer version, GPU technology and GPU on the XenServer HCL (Hardware compatibility list), here: Citrix XenServer HCL,
  • If yes it is fully supported.

 

What if it works but isn’t listed on the HCL?

It is likely that both server manufacturer and hypervisor vendor will refuse a support call involving your GPU. Anyone who has contact with hypervisor vendors such as Citrix or VMware will know that they will not provide support for uncertified hardware combinations.

Can NVIDIA provide a support statement if VMware/Citrix/Other doesn’t have a combination on its HCL?

No we cannot as the type of problems that arise are usually server or hypervisor related and we do not have access to partners closed code nor the power to resolve their hardware faults.

What should I do if my combination is not on an HCL?

  • Contact your server OEM (e.g. Dell, Cisco, IBM, HP, Supermicro etc, as they submit their servers with a limit range of GPUs they choose to support and test to the hypervisor vendors for certification
  • Contact the hypervisor certification teams at Citrix and VMware who liaise with the OEM vendors so that they are aware you want to use certain hardware with their hypervisor

The complicated background and reasons for certifications being done like this

I wrote a long detailed blog on why the certification involves this combination of hardware and hypervisor a long while ago if you really want to know more details: https://www.citrix.com/blogs/2014/01/14/xenserver-and-xendesktop-nvidia-support-for-gpu-passthrough-on-cards-including-the-k5000/

Warnings

  • Just because a GPU and server combination is certified for use as a physical server does not mean it will even work when a hypervisor is added. Even if it does work it may well not be certified, tested or supported. Adding a hypervisor is like changing the server and so neither server OEMs nor hypervisor manufacturers will support systems with hypervisors unless certified with that hypervisor.
  • The server is an essential component as is the hypervisor_version number_, a question such as – “Is the NVIDIA Quadro K5000 supported for GPU-passthrough with vSphere?” doesn’t contain all the components and can’t be answered.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s