NCA-AIIO mock exam

Question 1, AI InfrastructureWhich link connects GPUs directly at very high bandwidth inside a server?

• ANVIDIA NVLink
• BNVIDIA BlueField
• CInfiniBand HDR
• DSpectrum Ethernet

Question 2, AI InfrastructureA large training cluster needs very low latency and RDMA between nodes. Which network is the classic choice?

• AInfiniBand
• BA consumer Wi-Fi mesh
• CA VPN over the internet
• DBluetooth

Question 3, AI InfrastructureYou need one physical GPU split into isolated instances for several small jobs. Which feature does this?

• AInfiniBand partitioning
• BRemote Direct Memory Access
• CNVLink Switch System
• DMulti-Instance GPU (MIG)

Question 4, AI InfrastructureWhat is often the biggest facility constraint when adding high-density GPU racks?

• ANumber of USB ports
• BPower and cooling
• CMonitor size
• DKeyboard layout

Question 5, Essential AI KnowledgeWhy do GPUs suit neural network training better than CPUs?

• AThey have larger hard disks
• BThey use less total power per server
• CThey avoid the need for software
• DThey run many parallel matrix operations

Question 6, Essential AI KnowledgeHow does inference typically differ from training?

• AInference always needs more GPUs than training does
• BInference serves requests with a focus on latency
• CInference updates weights continuously from new data
• DInference needs no memory beyond the input batch

Question 7, AI OperationsYou want GPU health, utilization and diagnostics across a cluster. Which NVIDIA tool fits?

• ANVIDIA Canvas
• BNVIDIA DCGM
• CNVIDIA Omniverse
• DNVIDIA Isaac

Question 8, AI OperationsResearchers run long batch training jobs in a shared HPC-style cluster. Which scheduler is common here?

• ANginx
• BGit
• CSlurm
• DCron