Aegis-GPUNCP-AII learning guides
One short guide per official objective, grouped by exam domain. Each point is backed by a quote from an NVIDIA source. Terms link to the glossary the first time they appear.
System and Server Bring-up
- 1.1 Deployment and validation sequence
- 1.2 Network topologies for AI factories
- 1.3 BMC, OOB and TPM setup
- 1.4 Firmware upgrades and fault detection
- 1.5 Power and cooling
- 1.6 Installing GPU servers
- 1.7 Validating installed hardware
- 1.8 Cable types and transceivers
- 1.9 Installing physical GPUs
- 1.10 Validating hardware for workloads
- 1.11 Third-party storage parameters
Physical Layer Management
Control Plane Installation and Configuration
Cluster Test and Verification
- 4.1 Single-node stress test
- 4.2 Running HPL
- 4.3 Single-node NCCL and NVLink Switch
- 4.4 Signal quality on cables
- 4.5 Confirming cabling
- 4.6 Switch firmware and software
- 4.7 BlueField-3 firmware and software
- 4.8 Transceiver firmware
- 4.9 ClusterKit node assessment
- 4.10 NCCL east-west bandwidth
- 4.11 NCCL burn-in
- 4.12 HPL burn-in
- 4.13 NeMo burn-in
- 4.14 Testing storage