Free NVIDIA-Certified Professional: AI Rack and Interconnect practice — 6 questions on AI Infrastructure Basics, with explanations. No sign-up.
Full 12-question mixed test →
Question 1 of 6 · AI Infrastructure Basics
A data-center team is commissioning a GB200 NVL72 rack with a critical IT load of 132 kW. The facility power design calls for N+1 power-shelf redundancy. Which configuration correctly verifies power redundancy at the rack level?
N+1 redundancy requires installed spare shelf capacity beyond the calculated N shelves needed to carry the full load, verified by an actual failover load test — not just paper capacity math.
Question 2 of 6 · AI Infrastructure Basics
Which statement accurately differentiates the AC and DC power architecture used in an NVIDIA GB200 NVL72 rack?
The GB200 NVL72 reference architecture takes facility 3-phase AC into in-rack power shelves that rectify it to 48V DC, which is then distributed to compute trays via a DC busbar — this is the defining feature that makes it a 'DC rack' at the distribution level.
Question 3 of 6 · AI Infrastructure Basics
In the GB200 NVL72 architecture, what is the primary distinction between the roles of NVLink and InfiniBand?
NVLink (via the NVLink Switch trays) forms the scale-up domain joining all 72 GPUs in the rack, while InfiniBand is the scale-out fabric that links multiple NVL72 racks into a larger cluster.
Question 4 of 6 · AI Infrastructure Basics
During rPDU installation for an AC-powered AI rack, a technician must balance a three-phase 415V feed across intelligent rPDUs feeding the rack's redundant power inputs. Which configuration correctly balances the load?
Even load distribution across all three phases, with rotated phase assignment between A-side and B-side rPDUs, minimizes phase imbalance and ensures a single-phase loss doesn't drop redundant power to any device.
Question 5 of 6 · AI Infrastructure Basics
A liquid-cooled GB200 NVL72 deployment uses a rack-level CDU that isolates the facility water loop from the technology cooling system (TCS) loop. During commissioning, the technician measures a facility loop supply temperature of 32°C and a TCS loop supply temperature of 40°C at the manifold. What does this indicate?
A well-functioning CDU heat exchanger typically has only a small approach temperature (roughly 2–3°C) between facility supply and TCS supply; an 8°C gap points to reduced facility water flow, a fouled plate heat exchanger, or a valve/control issue that needs troubleshooting before production load.
Question 6 of 6 · AI Infrastructure Basics
A GB200 NVL72 rack uses liquid cooling manifolds where stainless steel fittings mate to copper cold plates. During installation, what is the PRIMARY grounding/bonding concern the technician must verify to prevent long-term system failure?
Dissimilar metals in contact within a conductive coolant path create a galvanic corrosion risk; dielectric isolation or proper bonding at these joints, combined with standard rack-to-facility ground continuity, prevents both corrosion and electrical hazards.
Ready for the real thing?
The full course has two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed answer explanations.
