Skip to content
Portal

vGPUs

Virtual GPUs (vGPUs) allow a physical GPU to be shared among multiple virtual machines, giving VMs access to GPU compute resources for graphics-intensive or GPU-accelerated workloads.

Prerequisites

Before you can use vGPUs:

  • Your location must have vGPU-capable hardware (physical GPU cards configured for vGPU allocation).
  • vGPU support must be enabled for your location. This depends on the GPU hardware available at your location.
  • The VM you want to attach a vGPU to must be stopped (Halted state) before attaching or detaching the vGPU.

Tip

Contact your organisation administrator if you need vGPU access and it is not available in your location.

Billing

vGPU usage is billed in VGU (Virtual Graphics Processing Unit) units:

  • 1 VGU = 1 vGPU allocated for one month.
  • VGU is an allocation-based unit — you are charged for the vGPU as long as it is reserved, even if the VM is not actively using it.
  • Detached vGPUs (reserved but not attached to a VM) are still billed at the full VGU rate.
  • vGPUs in the Recycle Bin are still reserved for you and will appear on your invoice. Only after deleting a vGPU from the recycle bin is the resource released.

Tip

To minimise vGPU costs, delete vGPUs from the recycle bin promptly when no longer needed. See Recycle Bin for details on retention periods.

For the full billing unit reference, see Billing & Resource Consumption.

vGPU Profiles

Each location defines a set of available vGPU profiles. Profiles determine the amount of GPU resources allocated to a single vGPU. Common profiles specify the amount of GPU memory (e.g. 1 GB, 2 GB, 4 GB) or the number of virtual display heads.

The exact profiles available depend on the underlying physical GPU hardware at your location. Not all locations will offer the same vGPU profiles.

Reserving a vGPU

To reserve a vGPU before attaching it to a VM:

  1. Navigate to the vGPUs page in the left navigation (it has one tab per location), or the vGPUs tab on a VM details page.
  2. Click RESERVE VGPU (on the vGPUs page) — note that on the VM details page the button is Attach vGPU, not RESERVE VGPU.
  3. Select a GPU profile from the Available GPU profiles and enter a vGPU name.
  4. Click RESERVE.

The vGPU is created and appears in your list of reserved vGPUs. You can attach it to a VM when needed.

Attaching a vGPU to a VM

To attach a vGPU to a virtual machine, from the VM details page's vGPUs tab, choose one of two options: Reserve and Attach vGPU Profile (reserve a new one from a GPU profile in one step) or Attach reserved vGPU Profile (attach one you've already reserved).

Tip

Attaching a vGPU to a running VM is not supported — the portal blocks it with an error asking you to stop the VM first.

Note

A VM with an attached vGPU cannot be moved to another node or cloudspace — detach the vGPU first.

Detaching a vGPU from a VM

To detach a vGPU from a VM:

  1. Navigate to the vGPUs tab on the VM details page.
  2. Make sure the VM is in Halted state.
  3. Use the detach inline row action icon on the vGPU's row (this is an inline row action, not a select-then-action flow), then confirm on the Detach vGPU dialog.

The vGPU returns to your pool of reserved vGPUs. It continues to be billed at the VGU rate while reserved.

Deleting a vGPU

To permanently delete a vGPU (and stop billing):

  1. Navigate to the vGPUs tab on the VM details page (or the global vGPU list).
  2. Make sure the VM is halted. The vGPU does not need to be detached first — deleting an attached vGPU auto-detaches it.
  3. Use the delete inline row action icon on the vGPU's row (this is an inline row action, not a select-then-action flow), then confirm on the Delete vGPU dialog.

The vGPU moves to the Recycle Bin. See Recycle Bin & Restore for how long it is kept before permanent deletion (7 days by default).

Checking vGPU Status

On the VM details page, the vGPUs tab lists vGPU (name) and GPU (the underlying physical GPU), plus an Actions column. The customer-wide vGPUs page (left navigation) adds Cloudspace, Virtual Machine and Status columns, and is organised into tabs by location.

vGPUs with VDI

When creating a VDI profile, you can optionally assign a vGPU profile to the VMs provisioned by that profile, including the number of vGPU instances per VM. See the VDI Profiles section for details.

vGPU Management Workflow

  1. List available vGPU profiles: Navigate to the vGPUs page in the left navigation (under the customer). The list shows all available vGPU profiles with their allocation status.
  2. Reserve a vGPU: Click Reserve vGPU on the vGPUs page. Select a profile and enter a name for the reservation.
  3. Attach a vGPU to a VM: On the VM details page, open the vGPUs tab and click Attach vGPU. Choose either Reserve and Attach (reserves a new vGPU and attaches it in one step) or Attach reserved (attaches an existing reservation).
  4. Detach a vGPU: On the VM's vGPUs tab, click the detach icon on the target row.
  5. Delete a vGPU reservation: On the vGPUs page, click the delete icon on the target row.