Workspace integrates the NVIDIA Virtual GPU technology to provide GPU virtualization services for VM users in Windows 10, 3D graphic design, and multimedia environments. GPU virtualization increases GPU resource usage by allowing multiple VMs to share a physical GPU resource.
Applicable to automated GPU hardware resource allocation and release for virtualization services of radio and television, earthquake and meteorology, and art and design industries. You can perform this task to meet high-performance requirements of services such as high-definition video transmission, satellite cloud mapping, and geological structure and 3D graphics imaging.
Synchronize GPU devices and add vGPU to virtualize physical GPU devices into multiple vGPUs. For more information, see "Synchronize GPU devices" and "Add vGPUs."
Add iRS tasks for vGPU resources. For more information, see "Add an iRS service."
Apply vGPUs to VMs. For more information, see "Edit a VM."
Install the NVIDIA driver on each VM and connect the VMs to the vGPU license server.
Make sure the host is installed and configured with a physical GPU.
ARM hosts do not support GPU device management.
Before virtualizing a physical GPU, disable the Error Correcting Code (ECC) feature. After editing the ECC state, restart the host to make the new ECC state effective.
Only NVIDIA Tesla series graphics cards support GPU virtualization. Before adding a vGPU, make sure the physical graphics card operates in graphics mode.
You can virtualize a physical GPU into one type of vGPUs, but virtualize different physical GPUs on a graphics card into multiple types of vGPUs.
A physical GPU no longer supports passthrough after being virtualized into vGPUs. A physical GPU in passthrough mode cannot be virtualized into vGPUs.
From the left navigation pane, select Data Center > Virtualization > Cluster name > Host name.
Click the GPUs tab.
Click Synchronize. The system will synchronize GPU devices on the host to Space Console.
From the left navigation pane, select Data Center > Virtualization > Cluster name > Host name.
Click the GPUs tab.
Perform either of the following tasks to add vGPUs:
To add vGPUs for multiple GPUs, select the target GPUs, click Add vGPUs, specify the vGPU name, and then click OK.
To add a vGPU for a GPU, click Add vGPUs for the GPU, specify the vGPU name, and then click OK.
From the left navigation pane, select Data Center > Virtualization > Cluster name > Host name.
Click the GPUs tab.
Perform either of the following tasks to reclaim vGPUs:
To reclaim vGPUs from multiple GPUs, select the target GPUs, click Reclaim vGPUs, and then click OK.
To reclaim vGPUs from one GPU, click Reclaim vGPUs for the GPU, and then click OK.
From the left navigation pane, select Data Center > Virtualization > Cluster name > Host name.
Click the GPUs tab.
Click Enable ECC or Disable ECC for the target GPU.
Click OK.
From the left navigation pane, select Data Center > Virtualization > Cluster name > Host name.
Click the GPUs tab.
Click the target GPU. The vGPU list will display information about vGPUs created for the GPU.
GPUs
Slot Number: Slot number of the GPU.
Manufacturer: Vendor of the GPU.
Type: Model of the GPU.
ECC: ECC state.
vGPUs
UUID: UUID of the vGPU.
Type: Type of the vGPU.
Monitor Interfaces: Maximum number of monitor interfaces supported by the vGPU.
Buffer: Frame buffer size of the vGPU.