Manually install a Tesla driver on Windows - Elastic GPU Service

In general computing scenarios such as deep learning and AI scenarios, GPU-accelerated instances installed with NVIDIA Tesla drivers can deliver high-performance computing capabilities. When you create GPU-accelerated compute-optimized Windows instances, the instances do not support the automatic installation of Tesla drivers. You can only manually install Tesla drivers on the instances by following the instructions in this topic.

Procedure

Note

This topic is suitable for all GPU-accelerated compute-optimized Windows instances. For more information, see GPU-accelerated compute-optimized instance families.

Step 1: Download a Tesla driver

Connect to your GPU-accelerated compute-optimized Windows instance.
For more information, see Connect to a Windows instance by using a password or key.
On your remote desktop, visit the NVIDIA drivers download page.
Note
For more information about how to install and configure an NVIDIA driver, see NVIDIA CUDA Installation Guide for Linux.

Configure filters and click Search to search for a driver that is suitable for your instance.

驱动搜索.jpg

The following table describes the filters.

Filter	Description	Example value
Product Type Product Series Product	Select values from the Product Type, Product Series, and Product drop-down lists based on the GPU of your GPU-accelerated compute-optimized instance. Note For more information about how to view the details of a GPU-accelerated instance, such as the instance ID, instance type, and OS, see View instance information.	Data Center / Tesla A-Series NVIDIA A10
Operating System	Select a Windows version based on the image of the instance.	Windows 10 64-bit
CUDA Toolkit	Select a CUDA Toolkit version.	11.2
Language	Select a language for the driver.	Chinese (Simplified)

The following table describes the GPUs, Tesla driver versions, and CUDA Toolkit versions that are supported by different GPU-accelerated compute-optimized instance families.

Instance family	gn5	gn5i	gn6v	gn6i	gn6e	gn7	gn7i	gn7e
Product type	Data Center / Tesla	Data Center / Tesla	Data Center / Tesla	Data Center / Tesla	Data Center / Tesla	Data Center / Tesla	Data Center / Tesla	Data Center / Tesla
Product series	P-Series	P-Series	V-Series	T-Series	V-Series	A-Series	A-Series	A-Series
Recommended Tesla driver version	Version 426.00 or later					Version 452.77 or later	Version 462.31 or later	Version 452.77 or later
Recommended CUDA Toolkit version	CUDA Toolkit 10.1 Update 2					CUDA Toolkit 11.0 Update 1	CUDA Toolkit 11.2	CUDA Toolkit 11.0 Update 1

Note

The preceding table describes only the GPUs supported by some popular GPU-accelerated compute-optimized instance families. Instances that use the same GPU have the same GPU information, such as the product type, product series, and product. For example, instances of the ebmgn6e and gn6e instance families use NVIDIA V100 GPUs. Therefore, the product type, product series, and product family of the instances are the same.
When you manually install a Tesla driver and CUDA Toolkit, make sure that the driver version is compatible with the CUDA Toolkit version. For more information, see CUDA Compatibility.

On the search result page, click Show More Versions.
Find the driver that you want to download and click View.
In this example, the Data Center Driver for Windows driver is selected. The driver version is 462.31 and the CUDA Toolkit version is 11.2.
On the driver details page, click Download.

Step 2: Install the Tesla driver

Open the folder where the driver installation package is stored and double-click the installation package.
Follow the on-screen instructions to install the driver.
After the driver is installed, you can restart the GPU-accelerated instance based on your business requirements. For more information, see Restart an instance.
Important
Windows Remote Desktop Protocol (RDP) does not support specific applications, such as DirectX and Open Graphics Library (OpenGL) applications. After a Tesla driver takes effect for a Windows instance, you must install the Virtual Network Computing (VNC) service and client, or use remote access clients that support protocols such as PC over IP (PCoIP) or XenDesktop HDX 3D to support the applications.
In the lower-left corner of your remote desktop, type Device Manager in the search box and then press Enter.
On the Device Manager page, click Display adapters to check whether the driver is installed.
If the information about the Tesla driver appears, such as NVIDIA A10, the Tesla driver is installed.

References

If you purchase a GPU-accelerated compute-optimized Linux instance, you can install a Tesla driver to enhance the performance of the instance in general computing scenarios, such as deep learning and AI scenarios. For more information, see Automatically install or load a Tesla driver when you create a GPU-accelerated instance or Manually install a Tesla driver on a GPU-accelerated compute-optimized Linux instance.
If you no longer require a Tesla driver due to a specific reason, you can uninstall the driver. For more information, see Uninstall an NVIDIA Tesla driver.
If the driver version of your GPU-accelerated instance cannot meet your business requirements, or your GPU-accelerated instance becomes unavailable due to an invalid driver type or version, you can uninstall the driver and then install a new driver. You can also upgrade the driver. For more information, see the "Upgrade an NVIDIA Tesla driver" section of the Upgrade an NVIDIA Tesla or GRID driver topic.