Yeah I'd check BIOS settings like @olivierlambert mentioned here. But also, you mention pass through, this isn't pass through this is vGPU which is different (and only supported by some GPUs in specific), which allows you to have a single GPU that can run multiple VMs. If you want passthrough you'll need to follow the docs here https://docs.xcp-ng.org/compute/ This will let you pass through the entire card, make sure to passthrough all devices that are part of the card (so HD audio and stuff too, everything that is under the same PCI address). I've done this with a few GPUs and it works great.

In XOA, Import, drag and drop the XVA and that's it

@nodesocket Depends on your setup. For mine, I use sudo systemctl restart xo-server.service. You could also just restart the VM.

Yes, they are

No problem, happy to help

You can reattach the drive (in XO web UI or with xe by recreating the VBD: vbd-create vm-uuid=4aef7b34-2e7e-53c5-62e3-331be8151287 vdi-uuid=193f9105-e23c-4fe7-a8ae-40bed37ede4e device=0). Then try to boot the VM.

You can't delete a base copy. You need to understand how it works before doing dangerous operations. A base copy exists if there's a snapshot. If you have 0 snap and still base copy, it could be either: because the coalesce isn't finished because the coalesce doesn't work In both case, you have to take a look at your SMlog. If you have no idea about it, you should either get pro support or start to read the doc in more details You can start here: https://xcp-ng.org/docs/storage.html#coalesce