Contents

• 1. Introduction
  • 1.1. System Requirements
  • 1.2. OS Support Policy
  • 1.3. About This Document
• 2. Pre-installation Actions
  • 2.1. Verify You Have a CUDA-Capable GPU
  • 2.2. Verify You Have a Supported Version of Linux
  • 2.3. Verify the System Has gcc Installed
  • 2.4. Verify the System has the Correct Kernel Headers and Development Packages Installed
  • 2.5. Install MLNX_OFED
  • 2.6. Choose an Installation Method
  • 2.7. Download the NVIDIA CUDA Toolkit
  • 2.8. Address Custom xorg.conf, If Applicable
  • 2.9. Handle Conflicting Installation Methods
• 3. Package Manager Installation
  • 3.1. Overview
  • 3.2. RHEL 7 / CentOS 7
    • 3.2.1. Prepare RHEL 7 / CentOS 7
    • 3.2.2. Local Repo Installation for RHEL 7 / CentOS 7
    • 3.2.3. Network Repo Installation for RHEL 7 / CentOS 7
    • 3.2.4. Common Installation Intructions for RHEL 7 / CentOS 7
    • 3.2.5. Installing a Previous NVIDIA Driver Branch on RHEL 7
  • 3.3. RHEL 8 / Rocky 8
    • 3.3.1. Prepare RHEL 8 / Rocky 8
    • 3.3.2. Local Repo Installation for RHEL 8 / Rocky 8
    • 3.3.3. Network Repo Installation for RHEL 8 / Rocky 8
    • 3.3.4. Common Instructions for RHEL 8 / Rocky 8
  • 3.4. RHEL 9 / Rocky 9
    • 3.4.1. Prepare RHEL 9 / Rocky 9
    • 3.4.2. Local Repo Installation for RHEL 9 / Rocky 9
    • 3.4.3. Network Repo Installation for RHEL 9 / Rocky 9
    • 3.4.4. Common Instructions for RHEL 9 / Rocky 9
  • 3.5. KylinOS 10
    • 3.5.1. Prepare KylinOS 10
    • 3.5.2. Local Repo Installation for KylinOS
    • 3.5.3. Network Repo Installation for KylinOS
    • 3.5.4. Common Instructions for KylinOS 10
  • 3.6. Fedora
    • 3.6.1. Prepare Fedora
    • 3.6.2. Local Repo Installation for Fedora
    • 3.6.3. Network Repo Installation for Fedora
    • 3.6.4. Common Installation Intructions for Fedora
  • 3.7. SLES
    • 3.7.1. Prepare SLES
    • 3.7.2. Local Repo Installation for SLES
    • 3.7.3. Network Repo Installation for SLES
    • 3.7.4. Common Installation Instructions for SLES
  • 3.8. OpenSUSE
    • 3.8.1. Prepare OpenSUSE
    • 3.8.2. Local Repo Installation for OpenSUSE
    • 3.8.3. Network Repo Installation for OpenSUSE
    • 3.8.4. Common Installation Instructions for OpenSUSE
  • 3.9. WSL
    • 3.9.1. Prepare WSL
    • 3.9.2. Local Repo Installation for WSL
    • 3.9.3. Network Repo Installation for WSL
    • 3.9.4. Common Installation Instructions for WSL
  • 3.10. Ubuntu
    • 3.10.1. Prepare Ubuntu
    • 3.10.2. Local Repo Installation for Ubuntu
    • 3.10.3. Network Repo Installation for Ubuntu
    • 3.10.4. Common Installation Instructions for Ubuntu
  • 3.11. Debian
    • 3.11.1. Prepare Debian
    • 3.11.2. Local Repo Installation for Debian
    • 3.11.3. Network Repo Installation for Debian
    • 3.11.4. Common Installation Instructions for Debian
  • 3.12. Additional Package Manager Capabilities
    • 3.12.1. Available Packages
    • 3.12.2. Meta Packages
    • 3.12.3. Optional 32-bit Packages for Linux x86_64 .deb/.rpm
    • 3.12.4. Package Upgrades
• 4. Driver Installation
• 5. NVIDIA Open GPU Kernel Modules
  • 5.1. CUDA Runfile
  • 5.2. Debian
  • 5.3. Fedora
  • 5.4. KylinOS 10
  • 5.5. RHEL 9 and Rocky 9
  • 5.6. RHEL 8 and Rocky 8
  • 5.7. RHEL 7 and CentOS 7
  • 5.8. OpenSUSE and SLES
  • 5.9. Ubuntu
• 6. Precompiled Streams
  • 6.1. Precompiled Streams Support Matrix
  • 6.2. Modularity Profiles
• 7. Kickstart Installation
  • 7.1. RHEL 8 / Rocky Linux 8
  • 7.2. RHEL 9 / Rocky Linux 9
• 8. Runfile Installation
  • 8.1. Runfile Overview
  • 8.2. Installation
  • 8.3. Disabling Nouveau
    • 8.3.1. Fedora
    • 8.3.2. RHEL / Rocky and KylinOS
    • 8.3.3. OpenSUSE
    • 8.3.4. SLES
    • 8.3.5. WSL
    • 8.3.6. Ubuntu
    • 8.3.7. Debian
  • 8.4. Device Node Verification
  • 8.5. Advanced Options
  • 8.6. Uninstallation
• 9. Conda Installation
  • 9.1. Conda Overview
  • 9.2. Installing CUDA Using Conda
  • 9.3. Uninstalling CUDA Using Conda
  • 9.4. Installing Previous CUDA Releases
  • 9.5. Upgrading from cudatoolkit Package
• 10. Pip Wheels
• 11. Tarball and Zip Archive Deliverables
  • 11.1. Parsing Redistrib JSON
  • 11.2. Importing Tarballs into CMake
  • 11.3. Importing Tarballs into Bazel
• 12. CUDA Cross-Platform Environment
  • 12.1. CUDA Cross-Platform Installation
  • 12.2. CUDA Cross-Platform Samples
• 13. Post-installation Actions
  • 13.1. Mandatory Actions
    • 13.1.1. Environment Setup
    • 13.1.2. POWER9 Setup
  • 13.2. Recommended Actions
    • 13.2.1. Install Persistence Daemon
    • 13.2.2. Install Writable Samples
    • 13.2.3. Verify the Installation
      • 13.2.3.1. Verify the Driver Version
      • 13.2.3.2. Running the Binaries
    • 13.2.4. Install Nsight Eclipse Plugins
    • 13.2.5. Local Repo Removal
  • 13.3. Optional Actions
    • 13.3.1. Install Third-party Libraries
    • 13.3.2. Install the Source Code for cuda-gdb
    • 13.3.3. Select the Active Version of CUDA
• 14. Advanced Setup
• 15. Frequently Asked Questions
  • 15.1. How do I install the Toolkit in a different location?
  • 15.2. Why do I see “nvcc: No such file or directory” when I try to build a CUDA application?
  • 15.3. Why do I see “error while loading shared libraries: <lib name>: cannot open shared object file: No such file or directory” when I try to run a CUDA application that uses a CUDA library?
  • 15.4. Why do I see multiple “404 Not Found” errors when updating my repository meta-data on Ubuntu?
  • 15.5. How can I tell X to ignore a GPU for compute-only use?
  • 15.6. Why doesn’t the cuda-repo package install the CUDA Toolkit and Drivers?
  • 15.7. How do I get CUDA to work on a laptop with an iGPU and a dGPU running Ubuntu14.04?
  • 15.8. What do I do if the display does not load, or CUDA does not work, after performing a system update?
  • 15.9. How do I install a CUDA driver with a version less than 367 using a network repo?
  • 15.10. How do I install an older CUDA version using a network repo?
  • 15.11. Why does the installation on SUSE install the Mesa-dri-nouveau dependency?
  • 15.12. How do I handle “Errors were encountered while processing: glx-diversions”?
• 16. Additional Considerations
• 17. Switching between Driver Module Flavors
• 18. Removing CUDA Toolkit and Driver
• 19. Notices
  • 19.1. Notice
  • 19.2. OpenCL
  • 19.3. Trademarks
• 20. Copyright