cert
cert-1
cert-2

Pass NVIDIA NCP-AII Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
NCP-AII Exam - Verified By Experts
NCP-AII Premium File

NCP-AII Premium File

$89.99
$98.99
  • Premium File 170 Questions & Answers. Last Update: Nov 06, 2026

Whats Included:

  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates
 
$98.99
$89.99
accept 14 downloads in the last 7 days
block-screenshots
NVIDIA NCP-AII Practice Test 17

Question 160

During an NVIDIA AI factory acceptance test, engineers discover: An eight-GPU server reports only seven devices after service even though the host OS loads normally. Which TWO complementary elements are needed: the correct technical action and the strongest independent diagnostic evidence? Select TWO answers.

  1. Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.
  2. Verify approved protective grounding, ESD procedures and component handling before energizing or servicing rack hardware.
  3. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.
  4. Inspect PDU outlet maps, breaker assignments, dual-cord wiring, load readings and approved failover test logs.
  5. Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout.

Explanation

Correct Answers

A:

This is the direct operational evidence to investigate: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs. Observed scenario: An eight-GPU server reports only seven devices after service even though the host OS loads normally. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

E:

This is an appropriate correction for PCIe topology and accelerator enumeration: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout. Observed scenario: An eight-GPU server reports only seven devices after service even though the host OS loads normally. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

Incorrect Answers

B:

This alternative is relevant to Rack bonding grounding and ESD handling but does not directly resolve the current PCIe topology and accelerator enumeration issue: Verify approved protective grounding, ESD procedures and component handling before energizing or servicing rack hardware. Observed scenario: An eight-GPU server reports only seven devices after service even though the host OS loads normally. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

C:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. An eight-GPU server reports only seven devices after service even though the host OS loads normally. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

D:

This alternative is relevant to Dual-feed rack power and PDU mapping but does not directly resolve the current PCIe topology and accelerator enumeration issue: Inspect PDU outlet maps, breaker assignments, dual-cord wiring, load readings and approved failover test logs. Observed scenario: An eight-GPU server reports only seven devices after service even though the host OS loads normally.

Question 161

A production GPU cluster review identifies this condition: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.
  2. Trace each power supply to independent approved rack feeds and test safe loss of either feed within the supported power budget.
  3. Verify approved protective grounding, ESD procedures and component handling before energizing or servicing rack hardware.
  4. Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout.

Explanation

Correct Answer

D:

This is an appropriate correction for PCIe topology and accelerator enumeration: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

B:

This alternative is relevant to Dual-feed rack power and PDU mapping but does not directly resolve the current PCIe topology and accelerator enumeration issue: Trace each power supply to independent approved rack feeds and test safe loss of either feed within the supported power budget. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

C:

This alternative is relevant to Rack bonding grounding and ESD handling but does not directly resolve the current PCIe topology and accelerator enumeration issue: Verify approved protective grounding, ESD procedures and component handling before energizing or servicing rack hardware. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

Question 162

In a professional AI infrastructure deployment, a technician observes: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. Which underlying infrastructure principle BEST explains the issue? Select ONE best answer.

  1. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.
  2. Liquid cooling integrity and appropriate flow are part of accelerator availability and hardware safety, not optional performance tuning.
  3. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.
  4. Cooling effectiveness depends on local inlet air and airflow separation; a room-average temperature cannot establish adequate GPU cooling.

Explanation

Correct Answer

C:

This is the governing infrastructure principle: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

B:

That option concerns Liquid cooling supply and leak interlocks. A direct check is: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.

D:

That option concerns Cold aisle inlet air and thermal envelope. A direct check is: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.

Question 163

The cluster operations team needs to evaluate this situation: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Inspect TPM ownership and event logs, secure-boot state, signed firmware path and expected PCR measurements.
  2. Check controller address configuration, switch OOB paths, login records, firmware identity and host-off connectivity.
  3. Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.
  4. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.

Explanation

Correct Answer

C:

This is the direct operational evidence to investigate: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

Incorrect Answers

A:

That option concerns TPM state and measured secure boot. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

B:

This alternative is relevant to BMC network bring-up and out-of-band separation but does not directly resolve the current PCIe topology and accelerator enumeration issue: Check controller address configuration, switch OOB paths, login records, firmware identity and host-off connectivity. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

D:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement. The proportionate action is: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout.

Question 164

At a multi-node AI infrastructure handoff, the team records: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. After corrective work, which observable result BEST confirms that the requirement is met? Select ONE best answer.

  1. A planned single-feed isolation leaves required servers within power limits and serving their intended workloads.
  2. Authorized operators can manage an offline host through the protected BMC while unauthorized network access is denied.
  3. The correct devices are enumerated in the intended slots with healthy PCIe links and no degraded bandwidth warnings.
  4. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.

Explanation

Correct Answer

C:

This is an observable verified result, not merely a claimed configuration change: The correct devices are enumerated in the intended slots with healthy PCIe links and no degraded bandwidth warnings. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

Incorrect Answers

A:

This alternative is relevant to Dual-feed rack power and PDU mapping but does not directly resolve the current PCIe topology and accelerator enumeration issue: A planned single-feed isolation leaves required servers within power limits and serving their intended workloads. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

B:

This alternative is relevant to BMC network bring-up and out-of-band separation but does not directly resolve the current PCIe topology and accelerator enumeration issue: Authorized operators can manage an offline host through the protected BMC while unauthorized network access is denied. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down.

D:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. Observed scenario: A replacement network adapter trains at an unexpectedly narrow PCIe width and collective workloads slow down. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement. The proportionate action is: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout. A direct check is: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs. The acceptance condition is: The correct devices are enumerated in the intended slots with healthy PCIe links and no degraded bandwidth warnings.

Question 165

During an NVIDIA AI factory acceptance test, engineers discover: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Provision unique BMC addresses on protected out-of-band management, rotate initial credentials and test access separately from host networking.
  2. Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout.
  3. Calculate peak supported server and fabric load, apply facility margin and verify branch and PDU capacity under representative load.
  4. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.

Explanation

Correct Answer

B:

This is an appropriate correction for PCIe topology and accelerator enumeration: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

Incorrect Answers

A:

This alternative is relevant to BMC network bring-up and out-of-band separation but does not directly resolve the current PCIe topology and accelerator enumeration issue: Provision unique BMC addresses on protected out-of-band management, rotate initial credentials and test access separately from host networking. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

C:

That option concerns GPU node peak power budgeting. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

D:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

Question 166

A production GPU cluster review identifies this condition: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. Which underlying infrastructure principle BEST explains the issue? Select ONE best answer.

  1. Power redundancy depends on independent supply paths and properly sized capacity, not merely on the number of power cords.
  2. Redfish and IPMI expose hardware management data and operations, but endpoint selection and privileges determine their safety.
  3. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.
  4. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

Explanation

Correct Answer

D:

This is the governing infrastructure principle: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

Incorrect Answers

A:

That option concerns Dual-feed rack power and PDU mapping. A direct check is: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.

B:

That option concerns Redfish inventory and IPMI management boundaries. A direct check is: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.

C:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

Question 167

In a professional AI infrastructure deployment, a technician observes: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Inspect CDU telemetry, coolant specifications, alarm contacts, flow and pressure records and documented fail-safe test results.
  2. Inspect management API target paths, access scopes, sensor values, chassis actions and independent host inventory.
  3. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.
  4. Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs.

Explanation

Correct Answer

D:

This is the direct operational evidence to investigate: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

Incorrect Answers

A:

That option concerns Liquid cooling supply and leak interlocks. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

B:

That option concerns Redfish inventory and IPMI management boundaries. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

C:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement. The proportionate action is: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout.

Question 168

The cluster operations team needs to evaluate this situation: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. After corrective work, which observable result BEST confirms that the requirement is met? Select ONE best answer.

  1. Under representative load each node stays within supported inlet limits without avoidable thermal throttling or recirculation.
  2. The correct devices are enumerated in the intended slots with healthy PCIe links and no degraded bandwidth warnings.
  3. Ignore a missing accelerator because the CUDA application still sees at least one working GPU.
  4. All nodes are securely mounted in compatible rack positions and mechanical inspection passes without blocked service access.

Explanation

Correct Answer

B:

This is an observable verified result, not merely a claimed configuration change: The correct devices are enumerated in the intended slots with healthy PCIe links and no degraded bandwidth warnings. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

Incorrect Answers

A:

That option concerns Cold aisle inlet air and thermal envelope. An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement.

C:

This tempting but unsafe shortcut is not a sufficient remediation for PCIe topology and accelerator enumeration: Ignore a missing accelerator because the CUDA application still sees at least one working GPU. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide. The governing mechanism is: An AI node's compute and I/O performance depend on physical PCIe connectivity and supported device placement. The proportionate action is: Compare PCIe inventory, negotiated link width and speed, slot placement and GPU topology with the certified system layout. A direct check is: Inspect lspci topology, link capabilities versus negotiated state, nvidia-smi topology output and hardware event logs. The acceptance condition is: The correct devices are enumerated in the intended slots with healthy PCIe links and no degraded bandwidth warnings.

D:

This alternative is relevant to DGX rack load and rail placement but does not directly resolve the current PCIe topology and accelerator enumeration issue: All nodes are securely mounted in compatible rack positions and mechanical inspection passes without blocked service access. Observed scenario: An installer moves expansion devices into slots that share constrained lanes contrary to the platform placement guide.

Question 169

At a multi-node AI infrastructure handoff, the team records: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Restore planned airflow direction, containment and intake clearance, and validate inlet thermal conditions at loaded nodes.
  2. Provision unique BMC addresses on protected out-of-band management, rotate initial credentials and test access separately from host networking.
  3. Treat eight visible GPUs as conclusive proof that all NVLink or NVSwitch data paths are functional.
  4. Inspect the intra-node GPU peer topology and link state, then run supported fabric diagnostics and collective tests.

Explanation

Correct Answer

D:

This is an appropriate correction for NVLink and NVSwitch internal fabric validation: Inspect the intra-node GPU peer topology and link state, then run supported fabric diagnostics and collective tests. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled.

Incorrect Answers

A:

This alternative is relevant to Cold aisle inlet air and thermal envelope but does not directly resolve the current NVLink and NVSwitch internal fabric validation issue: Restore planned airflow direction, containment and intake clearance, and validate inlet thermal conditions at loaded nodes. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled.

B:

This alternative is relevant to BMC network bring-up and out-of-band separation but does not directly resolve the current NVLink and NVSwitch internal fabric validation issue: Provision unique BMC addresses on protected out-of-band management, rotate initial credentials and test access separately from host networking. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled.

C:

This tempting but unsafe shortcut is not a sufficient remediation for NVLink and NVSwitch internal fabric validation: Treat eight visible GPUs as conclusive proof that all NVLink or NVSwitch data paths are functional.

Question 170

During an NVIDIA AI factory acceptance test, engineers discover: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled. Which TWO complementary elements are needed: the correct technical action and the strongest independent diagnostic evidence? Select TWO answers.

  1. Inspect PDU outlet maps, breaker assignments, dual-cord wiring, load readings and approved failover test logs.
  2. Inspect the intra-node GPU peer topology and link state, then run supported fabric diagnostics and collective tests.
  3. Treat eight visible GPUs as conclusive proof that all NVLink or NVSwitch data paths are functional.
  4. Restore planned airflow direction, containment and intake clearance, and validate inlet thermal conditions at loaded nodes.
  5. Review nvidia-smi topology, NVLink health counters, NVSwitch diagnostics and single-node NCCL measurements.

Explanation

Correct Answers

B:

This is an appropriate correction for NVLink and NVSwitch internal fabric validation: Inspect the intra-node GPU peer topology and link state, then run supported fabric diagnostics and collective tests. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled. The governing mechanism is: GPU discovery, local CUDA computation and high-speed peer connectivity are separate readiness properties.

E:

This is the direct operational evidence to investigate: Review nvidia-smi topology, NVLink health counters, NVSwitch diagnostics and single-node NCCL measurements. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled.

Incorrect Answers

A:

This alternative is relevant to Dual-feed rack power and PDU mapping but does not directly resolve the current NVLink and NVSwitch internal fabric validation issue: Inspect PDU outlet maps, breaker assignments, dual-cord wiring, load readings and approved failover test logs. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled.

C:

This tempting but unsafe shortcut is not a sufficient remediation for NVLink and NVSwitch internal fabric validation: Treat eight visible GPUs as conclusive proof that all NVLink or NVSwitch data paths are functional. GPU discovery, local CUDA computation and high-speed peer connectivity are separate readiness properties.

D:

This alternative is relevant to Cold aisle inlet air and thermal envelope but does not directly resolve the current NVLink and NVSwitch internal fabric validation issue: Restore planned airflow direction, containment and intake clearance, and validate inlet thermal conditions at loaded nodes. Observed scenario: Eight visible accelerators achieve poor all-reduce bandwidth after one internal NVLink path is disabled. The governing mechanism is: GPU discovery, local CUDA computation and high-speed peer connectivity are separate readiness properties.

NCP-AI Infrastructure (NCP-AII) Practice Tests