Deploying a GPU instance – Support Guides

Learn how to deploy a GPU instance on Linux or Windows.

GPU instances are technically similar to CPU instances, but they also have a graphics card (Graphics Processing Unit, or GPU). The technology used (pci_passthrough) allows the instance’s operating system to control the GPU as a physical machine would.

The GPUs offered are powered by NVIDIA® V100S, L4, and L40S.

Requirements

A Public Cloud project with access to the regions where GPUs are available
An SSH key created to deploy a Linux GPU instance

OVHcloud Control Panel Access

Direct link: Public Cloud Projects
Navigation path: Public Cloud > Select your project

Instructions

You will find the information needed to deploy a GPU instance on Linux or Windows below.

Once you have selected your project, click Create an instance. Choose an appropriate GPU instance model to benefit from resources suited to graphics or intensive computing workloads.

Next, follow the remaining steps in this guide. The process may take a few minutes.

On Linux

All the images we offer can be used on a GPU instance.

During the image selection step, open the Unix distributions tab, then choose a UNIX image that suits your needs.

If you don’t feel comfortable with manually compiling a kernel module, we recommend using a distribution that is officially supported by NVIDIA and for which they provide turnkey drivers: https://developer.nvidia.com/cuda-downloads.

Once the instance is delivered, you can then log in and check for the graphics card:

lspci | grep -i nvidia
00:05.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
00:06.0 Audio device: NVIDIA Corporation Device 10f1 (rev a1)

The graphics card is there, but cannot be used yet. To do so, you must first install the NVIDIA driver. You can find the list of packages at this address: List of available Linux packages.

You will then need to enter the following commands:

NOTE: The Linux command can vary based on your distribution. If in doubt, please check the official guide for your version of Linux.

wget URL_of_packet_to_download
sudo dpkg -i cuda-repo-XXXX-XXXXXX
sudo apt-get update
sudo apt-get upgrade
sudo apt-get install cuda
sudo apt-get install -y cuda-drivers
sudo apt-get install linux-headers-$(uname -r)
sudo reboot

Once the instance has been rebooted, the graphics card will appear in the NVIDIA utility program:

nvidia-smi
Wed Apr 26 13:05:25 2017
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 375.51                 Driver Version: 375.51                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 106...  Off  | 0000:00:05.0     Off |                  N/A |
|  0%   22C    P0    26W / 120W |      0MiB /  6072MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID  Type  Process name                               Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

The GPU instance is now fully functional and usable.

On Windows

Running an NVIDIA GPU instance with Windows requires UEFI images for the KVM/pci_passthrough virtualization to function correctly. For the NVIDIA drivers to work, you must use an image based on a virtual UEFI BIOS; non-UEFI images do not work.

NOTE: We only offer the possibility to install the special images on a few selected models (T2-45, T2-90, T2-180). Additionally, depending on the region selected, these special images may not be available.

During the image selection step, open the Windows distributions tab, then select a Windows image that is compatible with the chosen instance model.

NOTE: We cannot guarantee that the solution will work with all future versions of the NVIDIA driver.

Before performing any update of the NVIDIA driver, we strongly recommend that you take a snapshot of your instance. This will allow you to roll back if necessary.

Connecting to a Windows OS instance

After the instance has been created, the Windows installation needs to be finalised (sysprep). To achieve this, click on the more options ... button and select Instance details. Switch to the tab VNC console. The console should already display the post-installation interface.

In the first step, decide on your localization settings by selecting a region, a language, and a keyboard layout. Click on Next to proceed.

The second step requires setting up the default "Administrator" account. Enter your passphrase twice and click on Finish to complete the installation process. Use the eye symbol to check if all the characters entered into the fields match the actual layout of your keyboard.

The instance will reboot, and you will be able to log in with these credentials using a remote desktop client.

From Windows

Use the Windows Search if necessary, and open the native "Remote Desktop Connection" client application.

Enter the IPv4 address of your instance and "Administrator" as the user, then type your passphrase. Usually, a warning message will appear, asking to confirm the connection because of an unknown certificate. Click on Yes to log in to the instance.

If you experience any issues with this procedure, verify that remote (RDP) connections are allowed on your device by checking your system settings, firewall rules, and possible network restrictions.

Once you have connected to your instance, you will need to install the NVIDIA driver from the official website.

After installation, the driver will appear in Device Manager > Graphics cards, allowing you to verify that the GPU card is correctly recognized and operational. You can then start using your instance for applications that require GPU acceleration.

Go further

For more information and tutorials, please see our other Public Cloud support guides or explore the guides for other OVHcloud products and services.