DGX Software for Red Hat Enterprise Linux 8 - Installation Guide

• Introduction
  • Related Documentation
  • Prerequisites
    • Red Hat Subscription
    • Access to Repositories
      • NVIDIA Repositories
      • Red Hat Repositories
    • Network File System
    • BMC Password
• Installing Red Hat Enterprise Linux 8
  • Obtaining Red Hat Enterprise Linux 8
  • Booting Red Hat Enterprise Linux 8 ISO Locally
  • Booting the Red Hat Enterprise Linux 8 ISO Remotely
    • Booting the ISO Image on the DGX-1 Remotely
    • Booting the ISO Image on the DGX-2, DGX A100/A800, or DGX H100 Remotely
  • Installing Red Hat Enterprise Linux
    • Installing with Kickstart
      • Install using Kickstart
    • Disk Partitioning for DGX-1, DGX Station, DGX Station A100, and DGX Station A800
    • Disk Partitioning with Encryption for DGX-1, DGX Station, DGX Station A100, and DGX Station A800
    • Disk Partitioning for DGX-2, DGX A100/A800, and DGX H100
    • Disk Partitioning with Encryption for DGX-2, DGX A100/A800, and DGX H100
• Installing the DGX Software
  • Configuring a System Proxy
  • Enabling the DGX Software Repository
  • Installing Required Components
  • Installing Optional Components
  • Installing the Optional NVIDIA Desktop Theme
• Upgrading the OS and DGX Software
  • Installing or Upgrading to a Newer CUDA Toolkit Release
  • Checking the Currently Installed CUDA Toolkit Release
  • Determining the New Available CUDA Toolkit Releases
  • Installing the CUDA Toolkit or Upgrading Your CUDA Toolkit to a Newer Release
  • Installing GPUDirect Storage Support
    • Prerequisites to Installing GPUDirect Storage Support
    • Installing nvidia-gds
• Installing NVIDIA MLNX_OFED
  • Prerequisites
  • Installing and Configuring MLNX_OFED
  • Updating NVIDIA MLNX_OFED
• Running Containers
• Configuring Storage - NFS Mount and Cache
• Installing Software on Air-Gapped NVIDIA DGX Systems
  • Registering Your System
  • Creating a Local Mirror of the NVIDIA Repository
  • Installing Docker Containers
• Installing Software Using Local Repositories
  • Installing the Base Red Hat Enterprise Linux System
  • Downloading the Local Repositories
  • Installing the Local Repositories
  • Installing the NVIDIA CUDA Driver from the Local Repository
• Changing the BMC Login
  • Changing the BMC Login on the DGX-1
  • Changing the BMC Login on the DGX-2 or DGX A100/A800
• Installing ConnectX-7 Firmware
• Installing the Mellanox ConnectX-6 Firmware
• Installing NVIDIA DOCA OFED
  • DOCA-Host Installation Profiles
  • Prerequisites
  • Installation Steps
    • DGX Systems with NVIDIA ConnectX-7 Adapter Cards
    • DGX systems with BlueField-3 DPU in NIC Mode (Optional)
• Installing BlueField-3 NIC Firmware
  • Alternative Methods
• Changing NVIDIA Driver Branches