Configure a Shared PCI Device for vGPU on the Linux Virtual Machine

To use NVIDIA vGPU, you must configure a shared PCI device for the Linux virtual machine.

Prerequisites

• Verify that the Linux virtual machine is prepared for use as a desktop. See Create a Virtual Machine and Install Linux and Prepare a Linux Machine for Remote Desktop Deployment.
• Verify that Horizon Agent is not installed on the Linux virtual machine.
• Verify that the NVIDIA VIB is installed on the ESXi host. See Install the VIB for the NVIDIA GRID vGPU Graphics Card on the ESXi Host.
• Familiarize yourself with the virtual GPU types that are available with NVIDIA vGPU, which you select with the GPU Profile setting. The virtual GPU types provide varying capabilities on the physical GPUs installed on the ESXi host.

Note: For information about the NVIDIA graphics cards and Linux distributions that support vGPU capabilities, see https://docs.nvidia.com/grid/latest/product-support-matrix/index.html.

Procedure

1. Power off the virtual machine.
2. In vSphere Client, select the virtual machine and, under the VM Hardware tab, click Edit Settings.
3. In the New device menu, select Shared PCI Device.
4. Click Add and select NVIDIA GRID vGPU from the drop-down menu.
5. For the GPU Profile setting, select a virtual GPU type from the drop-down menu.
   Note: To support vGPU capabilities on Linux application pools, you must select the profile that assigns the full memory of the physical GPU to the VM. This means selecting the highest available vGPU profile from the list of options.
6. Click Reserve all memory and click OK.
   You must reserve all virtual machine memory to enable the GPU to support NVIDIA GRID vGPU.
7. Power on the virtual machine.