0

I am using nested virtualization in GCP. I can achieve it by adding the support of Intel "VT-x" via following the Google provided document: https://cloud.google.com/compute/docs/instances/enable-nested-virtualization-vm-instances.

Now I am looking at how to add the nested GPU support via Pass Through, it requires the support of Intel "VT-d". How to check whether my GCP VM supports it and how to enable it?

I read previous question. Someone mentioned:

If VT-d is enabled, Linux will configure DMA Remapping at boot time. The easiest way to find this is to look in dmesg for DMAR entries. If you don't see errors, then VT-d is enabled.

I tried it and I didn't find any DMAR related message in the dmesg (only a few lines for DMA). Does it mean my GCP VM doesn't support VT-d. If so, is there a way to achieve it or it is impossible?

Digil
  • 742
  • 4
  • 12
chen lin
  • 125
  • 7
  • As you can see in the [documentation](https://cloud.google.com/compute/docs/instances/enable-nested-virtualization-vm-instances) you're able to enable VT-x "Nested virtualization adds support for Intel VT-x processor virtualization instructions to Compute Engine VMs." and there's no VT-d option. Please provide more details about your use case. – Serhii Rohoza Apr 20 '20 at 10:46
  • @SerhiiRohoza I am setup a k8s cluster myself on GCP/GCE and also KubeVirt. By do that I can host traditional VM workload into k8s and co-exist with container workload. Now I have some new VMs which also use GPU so I try to adding the support of GPU as well. According Kubevirt document, I can do that via GPU pass through or vGPU. But I have no idea not to enable that on GCP vms. https://kubevirt.io/2020/KubeVirt_deep_dive-virtualized_gpu_workloads.html – chen lin Apr 20 '20 at 16:15
  • I would recommend visiting this [thread](https://stackoverflow.com/questions/51261999/check-if-vt-d-iommu-has-been-enabled-in-the-bios-uefi) with a similar question. – Vivak P May 01 '20 at 16:15

0 Answers0