Steps I have done to make NVIDIA vGPU works:

1. Install XCP-ng 8.2.1
2. Install all update yum update
3. reboot
4. Download NVIDIA vGPU drivers for XenServer 8.2 from NVIDIA site. Version NVIDIA-GRID-CitrixHypervisor-8.2-510.108.03-513.91
5. Unzip and install rpm from Host-Drivers
6. reboot again
7. Download free CitrixHypervisor-8.2.0-install-cd.iso from Citrix site
8. Open CitrixHypervisor-8.2.0-install-cd.iso with 7-zip, then unzip vgpu binary file from Packages->vgpu....rpm->vgpu....cpio->.->usr->lib64->xen->bin
9. Upload vgpu to XCP-ng host to /usr/lib64/xen/bin and made it executable chmod +x /usr/lib64/xen/bin/vgpu
10. Deployed VM with vGPU and it started without any problems

So I did not make any modifications with emu-manager.

My test server is far away from me and it will take some time to download the windows ISO to this test location. Then I will check how it works in the guest OS and report back here.

Hello!

I'm system and network engineer.
I have started work with xenserver since 2013 (I think it was opensource 6.2 version).
In 2015 I started deploing my own control panel for hosting based on xenserver.

After Citrix closed free version I started to move to XCP-ng.

Now I'm playing with xcp-ng + GPU, testing performance in different scenarios and sharing on this forum I believe that knowledge should be open so that everyone can move forward faster.

Some time later I faced with problem that VMs (Linux and Windows) can't correctly start GPU (ADM MxGPU).
In windows device manager there was error #43.
I have solved this error without host reboot by reloading gim module.

Hope this will help somebody else.

rmmod gim gim_api
modprobe gim gim_api

+1 to @Andrew setup

@wyatt-made I need few days to test it.
Will report later here

Finally after a week I found the solution!
There is no problem with emu-manager.
XCP does not contain necessary package vgpu.
I copied vgpu from Citrix ISO and now it is alive! : )

Live migration is not possible in this case, because of lack of free space.

You have to shutdown VM, then click this button and chose another host

Some time later I faced with problem that VMs (Linux and Windows) can't correctly start GPU (ADM MxGPU).
In windows device manager there was error #43.
I have solved this error without host reboot by reloading gim module.

Hope this will help somebody else.

rmmod gim gim_api
modprobe gim gim_api

In my case problem was in bad connection.
I have reassembled server, cleaned PCI-e (GPU) contacts and now it is stable.

But it will be nice to have some tool to control and monitor AMD GPUs.

@hani it began asking for license after one day but without throttling.

I have switched to AMD GPUs

@Dani Strange. Is it executable? Did you tried to follow step by step my instruction to make vGPU work?

Do you have vgpu binary file at /usr/lib64/xen/bin/vgpu?

Why it is not working?

Did you read my post about how to run Nvidia vGPU on XCP-ng?
https://xcp-ng.org/forum/post/55774

But in any case you will need Nvidia license sever

That's good sign.

On host you can check to which pci-e controller drives are connected by command
ls -al /sys/block/sd*

You will see something like this

Before ata... you will find pci-e address

Check if you passthrough correct SATA controller.
I thing you have 2 SATA controllers because your MB have M.2 connector.

Can you see drives on the host if disable pass through?

Recently I have some troubles with OCulink to NVMe cables.
That was Supermicro server and they have different cables for SATA/SAS and NVMe.
The difference is in wire resistance - for SATA/SAS you should use cables with 100Ohm resistance and for NVM with 85Ohm.
In such way motherboard detects what type of drives are connected to hybrid port.

It is just my ideas what happens.

First you should check if host can see drives without pass through.

Live migration is not possible in this case, because of lack of free space.

You have to shutdown VM, then click this button and chose another host

Hello!

Are all VMs on this host belongs to you and you certainly know what processes running on them?

I had same issue with Dell R630.
The solution was to update to latest BIOS.
I think that some clients ran some software that triggered some bug and host rebooted.

XCP-ng security updates does not helped.
In my case only BIOS update fixed suddenly crushes.

So the work around will be to move VMs one by one to another host and check if it will solve the problem.