Please help me on this.
Is it possible to run cloud XR server on non RTX GPU, like Tesla-V100?
Is it mandatory to use Hypervisor approach (VMware vSphere or Citrix, ) when using GPU passthrough for cloud XR server setup?
Does cloud XR server setup supports NVIDIA data center GPU driver?
V100 images (P3 instances) require a software license applied to the image through the AWS marketplace. Without this applied, the image will not register having a GPU for CloudXR to use.
To apply the license, deploy a Windows-based P3 instance using the marketplace listings from Ingram Micro on the AWS Marketplace. Then you should be able to move and install CloudXR manually from the package available on the NVIDIA DevZone
You can verify your licensing with nvidia-smi with the command "nvidia-smi -q | Select-String -Pattern “(License Status) | (Product Name)” which should show the License Status.
Thank you for the quick reply.
We are working on a Local Server (Ubuntu HOST + Virtual-manager with Windows 2010) with NVIDIA Telsa V100 GPU.
We are using vfio_pci driver for GPU passthrough and enabled GPU in Windows VM. Is this septup correct?
We have to use VMware Vsphere hypervisor with Windows 2010 for installing cloudXR?
I haven’t heard of trying without using a hypervisor and being successful, but know most/all of the VM implementations that are passthrough still leverage a hypervisor. WMware vSphere is the most common one we hear being used. Other hypervisors may work, but haven’t heard directly.