Skip to main content
Removed in Orka 4.0. GPU passthrough configuration applies only to Intel-based VMs, so this page applies to Orka 3.6 and earlier. In Orka 4.0, which supports Apple silicon only, the --gpu flag is removed, GET /api/v1/cluster-info no longer returns gpuPassthroughEnabled, and the API rejects a VM deploy or VM config that turns on gpuPassthrough with a 400 error. If you still run Intel workloads, stay on Orka 3.6, which MacStadium supports through March 2028. See the compatibility matrix for details.
GPU passthrough is always enabled on Apple silicon VMs and cannot be disabled. For more information, see Feature Parity: Apple Hardware and Compatibility: Performance Improving Features.
When GPU passthrough is enabled for a cluster, VMs with GPU passthrough enabled can access and use the host GPU for their workloads. GPU passthrough must be enabled on the Intel-based nodes in your cluster before you can use it on VMs. Contact the MacStadium team to request that GPU passthrough is enabled on your nodes.

​
Defaults

GPU passthrough is disabled at the VM level for all Intel-based VMs, even when it is enabled in the cluster. You must explicitly enable it at deploy time or in a VM config template.

​
Limitations

  • You can deploy only one VM with GPU passthrough per Intel-based node at a time.
  • GPU passthrough and VNC cannot work together on Intel-based VMs. Enabling GPU passthrough automatically disables the VNC console. Always pass --disable-vnc alongside --gpu. Make sure you can still access the VM via SSH or Apple Screen Sharing before enabling GPU passthrough.
  • GPU passthrough is always enabled for Apple silicon-based nodes and VMs and cannot be disabled.

​
Enable GPU passthrough in a VM config template

Create a reusable template with GPU passthrough enabled:

​
Deploy a VM with GPU passthrough

To deploy a VM with GPU passthrough enabled, pass --gpu and --disable-vnc:
You can also override a template’s GPU setting at deploy time: