Launching an NVIDIA GPU Cloud VM from the Azure Portal

Creating Your GPU Cloud VM

  1. Log into the Azure portal (
  2. Select Create a Resource from the left-side menu.

  3. On the New pane, search for "nvidia", and then select the NVIDIA GPU Cloud image that you wan to use from the list - either the basic NVIDIA GPU Cloud image, or the NVIDIA GPU Cloud image with TensorFlow or PyTorch.

  4. Click Create from the NVIDIA GPU Cloud image pane.
  5. Complete the Basics settings as follows:
    • Name: Name of your choosing
    • VM disk type: SSD
    • Username: Your username
    • Authentication type: SSH public key
    • SSH public key: Paste in your SSH public key
    • Resource group: Use existing, then select your Resource group

  6. Click OK.
  7. From the Choose a size pane, filter the list by GPU compute type and SSD disk type, then select a P40, P100, or V100 SKU that meets your requirements and budget and click Select.

  8. Under Settings, set up a Basic or Advanced Network Security Group.
    • To use the network security group assigned to this VM, select Basic and then select ports SSH (22) and HTTPS (443). You can add inbound rules to the network security group later for other ports as needed.
    • To use an existing network security group, select Advanced, then click the Network security group (firewall) arrow, and then select one of the already created NSGs.

      You can also create a new NSG at this time by clicking Create new and then adding inbound rules.

      See the section Setting Up a Network Security Group for the list of rules to add.

  9. Make other Settings selections as needed, then click OK.

    After the validation passes, the portal presents the details of your new image which you can download as a template to automate deployment later.

  10. Click OK to deploy the image. The deployment starts, as indicated by the traveling bar underneath the Alert icon. It may take a few minutes to complete A pop-up alert will let you know when the VM instance is successfully deployed.

Connecting to Your GPU Cloud VM Instance

  1. Open the page for your GPU Cloud VM instance.
    1. Select Virtual machines from the left side menu.
    2. Locate and select your GPU Cloud VM.
  2. Click Connect from the top menu, then follow the instructions to establish an SSH connection to your VMI.

    If you plan to access locked NGC containers, you will need to log in to the NGC container registry. See Logging in to the NGC Container Registry) for instructions.

    If the instructions for SSH login do not work, see the Troubleshooting SSH connections to an Azure Linux VM that fails, errors out, or is refused documentation for additional information.

Starting and Stopping Your GPU Cloud VM

  1. Open the page for your GPU Cloud VM.
    1. Select Virtual machines from the left side menu.
    2. Locate and select your GPU Cloud VM.
  2. Use the menu at the top of your GPU Cloud VM instance page to Start, Restart, or Stop your VM.

Deleting Your GPU Cloud VM and Associated Resources

When you created your VM, other resources for that instance were automatically created for you, such as a network interface, public IP address, and boot disk. If you deleted your VM, you will also need to delete these resources.
  1. Click All resources from the left side menu.
  2. Click the check box for the GPU VM that you want to delete, as well as the associated disk and network interface.

    The associated resources use the same base name as your VM and then appends an additional identifier.

    Do not select the Public IP address resource yet. The Network interface must be deleted first.

  3. Click Delete from the top menu.
  4. Enter ‘yes’ in the confirmation box on the Delete Resources pane and then click Delete.

  5. After you’ve deleted the VM disk and network interface, delete the public IP address resource.
    1. Click the Public IP address check box for your VM and then click Delete.
    2. Enter ‘yes’ in the confirmation box on the Delete Resources pane and then click Delete.