The 3D rendering options and other pool settings offer various advantages and drawbacks. Select the option that best supports your vSphere hardware infrastructure and your users' requirements for graphics rendering.
When to Choose the Automatic Option
The Automatic option is the best choice for many View deployments that require 3D rendering. vSGA (Virtual Shared Graphics Acceleration)-enabled virtual machines can dynamically switch between software and hardware 3D rendering, without your having to reconfigure. This option ensures that some type of 3D rendering takes place even when GPU resources are completely reserved. In a mixed cluster of ESXi 5.1 and ESXi 5.0 hosts, this option ensures that a virtual machine is powered on successfully and uses 3D rendering even if, for example, vMotion moved the virtual machine to an ESXi 5.0 host.
The only drawback with the Automatic option is that you cannot easily tell whether a virtual machine is using hardware or software 3D rendering.
When to Choose the Hardware Option
The Hardware option guarantees that every virtual machine in the pool uses hardware 3D rendering, provided that GPU resources are available on the ESXi hosts. This option might be the best choice when all your users run graphically intensive applications. You can use this option when configuring vSGA (Virtual Shared Graphics Acceleration).
With the Hardware option, you must strictly control your vSphere environment. All ESXi hosts must be version 5.1 or later and must have GPU graphics cards installed.
When all GPU resources on an ESXi host are reserved, View cannot power on a virtual machine for the next user who tries to log in to a desktop. You must manage the allocation of GPU resources and the use of vMotion to ensure that resources are available for your desktops.
When to Choose the Option to Manage Using vSphere Client
When you select the Manage using vSphere Client option, you can use vSphere Web Client to configure individual virtual machines with different options and VRAM values.
- For vSGA (Virtual Shared Graphics Acceleration), you can support a mixed configuration of 3D rendering and VRAM sizes for virtual machines in a pool.
- For vDGA (Virtual Dedicated Graphics Acceleration), each virtual machine must be individually configured to share a specific PCI device with the ESXi host and all memory must be reserved. For more information, see Preparing for vDGA Capabilities.
All ESXi hosts must be version 5.5 or later and must have GPU graphics cards installed.Note: Some Intel vDGA cards require a certain vSphere 6 version. See the VMware Hardware Compatibility List at http://www.vmware.com/resources/compatibility/search.php. Also, for Intel vDGA, the Intel integrated GPU is used rather than discrete GPUs, as is the case with other vendors.
- For AMD Multiuser GPU using vDGA, each virtual machine must be individually configured to share a specific PCI device with the ESXi host and all memory must be reserved. This feature allows a PCI device to appear to be multiple separate physical PCI devices so that the GPU can be shared between 2 to 15 users. For more information, see Preparing to Use the Capabilities of AMD Multiuser GPU Using vDGA.
All ESXi hosts must be version 6.0 or later and must have GPU graphics cards installed.
You might also choose this option if you want to explicitly manage graphics settings of clones and linked clones by having the clones inherit settings from the parent virtual machine.
When to Choose the NVIDIA GRID vGPU Option
With the NVIDIA GRID vGPU option, you can achieve a higher consolidation ratio of virtual machines on an NVIDIA GRID vGPU-enabled ESXi host than is possible by using vDGA, while maintaining the same performance level. As with vDGA (Dedicated Virtual Graphics), the ESXi and virtual machine also use GPU pass-through for NVIDIA GRID vGPU.
vGPU.consolidation = "true"
By default, the ESXi host assigns virtual machines to the physical GPU with the fewest virtual machines already assigned. This is called performance mode. If you would rather have the ESXi host assign virtual machines to the same physical GPU until the maximum number of virtual machines is reached before placing virtual machines on the next physical GPU, you can use consolidation mode.
Because a GPU does not need to be dedicated to one specific virtual machine, with the NVIDIA GRID vGPU option, you can create and configure a parent virtual machine or virtual machine template to be NVIDIA GRID vGPU-enabled and then create a desktop pool of virtual machines that can share the same physical GPU.
If all GPU resources on an ESXi host are being used by other virtual machines, when the next user tries to log in to a desktop, View can move the virtual machine to another NVIDIA GRID vGPU-enabled ESXi server in the cluster and then power on the virtual machine. All ESXi hosts must be version 6.0 or later and must have GPU graphics cards installed.
For more information, see Preparing for NVIDIA GRID vGPU Capabilities.
When to Choose the Software Option
Select the Software option if you have ESXi 5.0 hosts only, or if ESXi 5.1 or later hosts do not have GPU graphics cards, or if your users only run applications such as AERO and Microsoft Office, which do not require hardware graphics acceleration.
Configuring Desktop Settings to Manage GPU Resources
You can configure other desktop settings to ensure that GPU resources are not wasted when users are not actively using them.
For floating pools, set a session timeout so that GPU resources are freed up for other users when a user is not using the desktop.
For dedicated pools, you can configure the Automatically logoff after disconnect setting to Immediately and a Suspend power policy if these settings are appropriate for your users. For example, do not use these settings for a pool of researchers who execute long-running simulations. Note that the Suspend power policy is not available if you use the NVIDIA GRID vGPU option.