NVIDIA DGX Explained - Network Operator#
Network Operator
- Scope
- Why NVIDIA Network Operator
- NVIDIA Network Operator Overview
- Key Concepts
- Network Operator Installation using Base Command Manager
- Validation
- Check Subnet Manager (SM) is enabled on the InfiniBand switches
- Check Virtualization is enabled for the Subnet Manager on the InfiniBand switches
- Check the installed NVIDIA Network Operator version
- Verify NVIDIA Network Operator is installed
- Check values.yaml is properly applied by Helm
- Verify NIC Cluster Policy is applied
- Verify IP Pools are applied
- Verify SR-IOV setup
- Verify NVIDIA Network Operator is operational
- Use NVIDIA Collective Communications Library (NCCL) to verify inter-node GPU-to-GPU performance
- Use Cases
- Common Issues and Troubleshooting Tips
- Appendix
- Reference Links
Notices