Manually install the Tesla driver on a GPU compute-optimized instance (Windows)

更新时间:
复制 MD 格式

For general-purpose computing workloads such as deep learning and AI, a GPU requires the Tesla driver to use its high-performance computing capabilities. GPU-accelerated compute-optimized instances that run Windows do not support automatic installation of the Tesla driver during instance creation. Follow the steps in this topic to manually install the Tesla driver.

Procedure

Note

This topic applies to all GPU-accelerated compute-optimized Windows instances. For more information, see GPU-accelerated compute-optimized instance families (gn, ebm, and scc).

Step 1: Download the Tesla driver

  1. Connect to your Windows GPU instance.

    For more information, see Log on to a Windows instance by using Workbench.

  2. On your remote desktop, go to the NVIDIA driver download page.

    Note

    For more information about how to install and configure NVIDIA drivers, see the NVIDIA Driver Installation Quickstart Guide.

  3. Set the filters and click Search to find a compatible driver.

    The following table describes the filters.

    Parameter

    Description

    Example

    • Product type

    • Product series

    • Product

    Select the product type, series, and product that match your instance type's GPU model.

    Note

    To view details about your GPU instance, such as the instance ID, instance type, and operating system, see View instance information.

    • Data Center / Tesla

    • A-Series

    • NVIDIA A10

    Operating system

    Select the Windows operating system version that matches your instance's image.

    Windows 10 64-bit

    CUDA Toolkit

    Select a CUDA Toolkit version.

    11.2

    Language

    Select a language for the driver.

    Chinese (Simplified)

    GPU specifications, supported Tesla driver, and CUDA versions

    Item

    gn8is

    gn7i

    gn7e

    gn7

    gn6e

    gn6i

    gn6v

    gn5i

    gn5

    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

    Data Center / Tesla

    Product series

    L-Series

    A-Series

    A-Series

    A-Series

    V-Series

    T-Series

    V-Series

    P-Series

    P-Series

    Recommended Tesla driver version

    538.67 or later

    462.31 or later

    452.77 or later

    426.00 or later

    Recommended CUDA Toolkit version

    CUDA Toolkit 12.2 Update 2

    CUDA Toolkit 11.2

    CUDA Toolkit 11.0 Update 1

    CUDA Toolkit 10.1 Update 2

    Note
    • This table lists GPU information for only some common GPU-accelerated compute-optimized instance types. Instances that use the same GPU model, such as ebmgn6e and gn6e that both use the NVIDIA V100, share the same product type, product series, and product.

    • When you manually install the Tesla driver and CUDA Toolkit, ensure that their versions are compatible. For more information, see CUDA Compatibility.

  4. On the driver search results page, click Show More Versions.

  5. Find the driver you want and click View next to it.

    For example, select Data Center Driver for Windows with driver version 462.31 and CUDA Toolkit 11.2.

  6. On the driver details page, click Download.

    This page shows detailed driver information, such as the Driver Version, Release Date, Operating System, CUDA Toolkit version, Language, and File Size. Verify the details are correct, and then click Download.

Step 2: Install the Tesla driver

  1. Open the folder containing the downloaded driver and double-click the installation file.

  2. Follow the on-screen instructions to complete the installation.

    After the driver is installed, restart the GPU instance if prompted. For more information, see Restart an instance.

    Important

    On Windows, Remote Desktop Protocol (RDP) does not support applications that use DirectX or OpenGL after the Tesla driver is installed. To use these applications, install a VNC service and client, or use another remote connection client that supports protocols such as PCoIP or XenDesktop HDX 3D.

  3. In the Windows search box, type Device Manager and press Enter.

  4. In Device Manager, expand Display adapters to verify that the driver is installed correctly.

    If the Tesla driver information appears (for example, NVIDIA A10), the installation was successful.

Related documents