Free NCP-ARI AI Infrastructure Basics Practice Questions
The AI Infrastructure Basics domain makes up 20% of the NVIDIA NCP-ARI exam. Practice free questions covering AI cluster building blocks — GPU nodes, NVLink, InfiniBand and Ethernet fabrics, rail-optimized topologies, east-west traffic patterns, and how compute, storage, and management networks fit together — each with the correct answer and a detailed explanation.
NCP-ARI AI Infrastructure Basics Sample Questions with Answers
Sample Question 1 — AI Infrastructure Basics
A deployment team is preparing to install a high-density AI rack in an existing data hall. Which readiness check should be completed before the rack is populated?
- A. Confirm the rack color, aisle signage, and label font match the existing data hall standard.
- B. Verify available power capacity, cooling capacity, airflow layout, grounding, and redundancy for the planned rack load. (Correct answer)
- C. Install the network cables first so actual power draw can be measured under production traffic.
- D. Use the average draw from nearby racks because every rack in the same row shares completely identical capacity ratings.
Correct answer: B
Explanation: B is best because Domain 2 focuses on verifying power, cooling, redundancy, and grounding before deployment. A is useful housekeeping but not readiness. C puts installation ahead of capacity validation. D assumes nearby rack load proves this rack is supported.
Sample Question 2 — AI Infrastructure Basics
An AC-powered AI rack has two rPDUs fed from independent power paths. Several servers are connected only to one rPDU. What is the best concern to raise?
- A. The rack may look unbalanced, but redundancy is unaffected as long as the total drawn power stays within limits.
- B. The rack should use DC distribution because AC racks cannot provide redundancy.
- C. A single power-path failure could drop equipment that is not connected across the intended redundant feeds. (Correct answer)
- D. The condition only matters after the rack has reached full production utilization.
Correct answer: C
Explanation: C is best because redundancy depends on connecting equipment across independent feeds as designed. A ignores single-path dependency. B falsely claims AC racks cannot be redundant. D delays correction until after risk is already present.
Sample Question 3 — AI Infrastructure Basics
A three-phase rPDU shows one phase near its planning limit while the other two phases are lightly loaded. What should the technician do before adding more equipment?
- A. Redistribute loads according to the rack power plan so phase loading is balanced within allowed limits. (Correct answer)
- B. Add equipment to the lightest physical side of the rack because physical side balance equals phase balance.
- C. Continue installation as long as total rack power stays below the combined rPDU nameplate.
- D. Move all high-draw devices to the same phase so monitoring is easier during burn-in.
Correct answer: A
Explanation: A is best because balancing rPDUs and phases prevents overload on one leg even when total capacity looks acceptable. B confuses physical placement with electrical phase. C ignores per-phase limits. D deliberately worsens imbalance.
Sample Question 4 — AI Infrastructure Basics
A rack has adequate electrical capacity on paper, but the cold aisle temperature rises quickly when adjacent racks run workloads. What is the best next action?
- A. Proceed because electrical capacity is the limiting factor for AI racks.
- B. Open rear doors during peak periods to increase warm air escape.
- C. Reduce the cable labeling work so the required airflow checks can be skipped during deployment.
- D. Reassess cooling capacity and airflow containment before adding the new AI rack load. (Correct answer)
Correct answer: D
Explanation: D is best because power and cooling capacity must both be verified. A ignores thermal readiness. B can disrupt containment and airflow strategy. C removes a quality step without solving the cooling issue.
Sample Question 5 — AI Infrastructure Basics
Which statement best describes a key difference between AC-powered and DC-powered AI racks?
- A. AC racks do not require rPDUs because facility power arrives pre-conditioned, while DC racks always rely on standard 20-amp branch circuits wired inside the rack.
- B. AC racks receive alternating-current facility power through rack distribution, while DC racks use direct-current distribution such as busbar architectures. (Correct answer)
- C. DC racks eliminate the need to verify grounding because the voltage does not alternate.
- D. AC racks are always lower power than DC racks and do not need cooling validation.
Correct answer: B
Explanation: B is best because the distinction is the type of power distribution feeding the rack. A reverses common AC rack distribution practice. C is unsafe because grounding still matters. D overgeneralizes power and ignores cooling requirements.
Sample Question 6 — AI Infrastructure Basics
A team confirms that both A and B feeds are present in a rack, but both originate from the same upstream panel. What should be concluded?
- A. The rack may not have true power-path redundancy and the upstream design should be verified. (Correct answer)
- B. The rack is redundant because two receptacles are visible at the rack level.
- C. The rack is redundant only if the rPDU display shows equal current on both feeds.
- D. The condition is acceptable for AI racks because server PSUs alone provide all of the needed redundancy.
Correct answer: A
Explanation: A is best because redundancy must be verified through independent upstream paths, not just two local feeds. B equates visible cabling with redundancy. C checks balance, not independence. D overstates PSU redundancy without upstream diversity.
Sample Question 7 — AI Infrastructure Basics
A liquid-cooled rack is being staged. Which role should the technician associate with a CDU?
- A. It converts single-mode fiber into multimode fiber for the rack-level optical links between the switch pairs.
- B. It balances rPDU phases by moving IT load between electrical feeds.
- C. It distributes and manages coolant flow between the facility cooling loop and the rack or IT equipment loop. (Correct answer)
- D. It provides cable strain relief for high-density copper bundles.
Correct answer: C
Explanation: C is best because a CDU supports liquid cooling by managing coolant distribution and heat exchange between loops. A is an optical function that a CDU does not perform. B is electrical load management. D is cable management.
Sample Question 8 — AI Infrastructure Basics
During installation, a technician notices that rack bonding hardware is missing from one cabinet. What is the best response?
- A. Continue the installation and note the missing hardware for the final report.
- B. Use cable shields as a temporary bonding path until the hardware arrives.
- C. Rely on the rPDU ground conductor alone because separate rack bonding hardware is never required.
- D. Pause or escalate until grounding and bonding requirements are verified and corrected. (Correct answer)
Correct answer: D
Explanation: D is best because grounding must be verified before high-density infrastructure is placed into service. A documents but does not fix the issue. B creates an improper workaround. C makes an absolute claim that may violate site requirements.
Keep Practicing
Take the 10-question NCP-ARI quick-start test across all 7 domains, or return to the NCP-ARI practice test hub for 270+ questions, exam details, and a study plan. Drill the other domains: High-Density Cabling Installation, AI Infrastructure Basics, Testing, Verification, and Documentation, Pre-Deployment Planning and Site Assessment, Rack Infrastructure Preparation, Safety, Standards, and Compliance, and Cable Support Systems and Weight Management.