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 21

Question 200

During an NVIDIA AI factory acceptance test, engineers discover: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session. Which THREE elements together establish the appropriate corrective action, independent test evidence and verified successful outcome? Select THREE answers.

  1. Check controller address configuration, switch OOB paths, login records, firmware identity and host-off connectivity.
  2. Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites.
  3. Check supported CDU supply, flow, leak detection and interlock behavior, then demonstrate controlled thermal response before production.
  4. Review DCGM entity discovery, health watches, diagnostic test results, operating load and error details.
  5. Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests.
  6. Required GPU health suites pass on the intended devices without disrupting unapproved production activity.

Explanation

Correct Answers

D:

This is the direct operational evidence to investigate: Review DCGM entity discovery, health watches, diagnostic test results, operating load and error details. Observed scenario: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session. The governing mechanism is: Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination.

E:

This is an appropriate correction for DCGM initial health and diagnostic scope: Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests. Observed scenario: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session. The governing mechanism is: Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination.

F:

This is an observable verified result, not merely a claimed configuration change: Required GPU health suites pass on the intended devices without disrupting unapproved production activity. Observed scenario: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session.

Incorrect Answers

A:

This alternative is relevant to BMC network bring-up and out-of-band separation but does not directly resolve the current DCGM initial health and diagnostic scope issue: Check controller address configuration, switch OOB paths, login records, firmware identity and host-off connectivity. Observed scenario: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session.

B:

This tempting but unsafe shortcut is not a sufficient remediation for DCGM initial health and diagnostic scope: Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites. Observed scenario: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session. The governing mechanism is: Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination. The proportionate action is: Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests. A direct check is: Review DCGM entity discovery, health watches, diagnostic test results, operating load and error details. The acceptance condition is: Required GPU health suites pass on the intended devices without disrupting unapproved production activity.

C:

This alternative is relevant to Liquid cooling supply and leak interlocks but does not directly resolve the current DCGM initial health and diagnostic scope issue: Check supported CDU supply, flow, leak detection and interlock behavior, then demonstrate controlled thermal response before production. Observed scenario: A rack acceptance script runs intrusive high-load diagnostics during an unannounced customer training session.

Question 201

A production GPU cluster review identifies this condition: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites.
  2. Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests.
  3. Calculate peak supported server and fabric load, apply facility margin and verify branch and PDU capacity under representative load.
  4. Provision unique BMC addresses on protected out-of-band management, rotate initial credentials and test access separately from host networking.

Explanation

Correct Answer

B:

This is an appropriate correction for DCGM initial health and diagnostic scope: Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for DCGM initial health and diagnostic scope: Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites. A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

C:

This alternative is relevant to GPU node peak power budgeting but does not directly resolve the current DCGM initial health and diagnostic scope issue: Calculate peak supported server and fabric load, apply facility margin and verify branch and PDU capacity under representative load. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

D:

This alternative is relevant to BMC network bring-up and out-of-band separation but does not directly resolve the current DCGM initial health and diagnostic scope issue: Provision unique BMC addresses on protected out-of-band management, rotate initial credentials and test access separately from host networking. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

Question 202

In a professional AI infrastructure deployment, a technician observes: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. Which underlying infrastructure principle BEST explains the issue? Select ONE best answer.

  1. Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites.
  2. Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination.
  3. Grounding and electrostatic protection limit personnel hazards and component damage; successful boot alone does not establish safe installation.
  4. Liquid cooling integrity and appropriate flow are part of accelerator availability and hardware safety, not optional performance tuning.

Explanation

Correct Answer

B:

This is the governing infrastructure principle: Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for DCGM initial health and diagnostic scope: Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites. A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination.

C:

This alternative is relevant to Rack bonding grounding and ESD handling but does not directly resolve the current DCGM initial health and diagnostic scope issue: Grounding and electrostatic protection limit personnel hazards and component damage; successful boot alone does not establish safe installation. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

D:

That option concerns Liquid cooling supply and leak interlocks. The proportionate action is: Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests.

Question 203

The cluster operations team needs to evaluate this situation: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Review DCGM entity discovery, health watches, diagnostic test results, operating load and error details.
  2. Compare platform power specifications, measured peak draw, PDU branch telemetry, breaker ratings and planned concurrency.
  3. Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites.
  4. Inspect facility grounding tests, rack bonds, wrist-strap practice, ESD storage and chassis service records.

Explanation

Correct Answer

A:

This is the direct operational evidence to investigate: Review DCGM entity discovery, health watches, diagnostic test results, operating load and error details. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

Incorrect Answers

B:

This alternative is relevant to GPU node peak power budgeting but does not directly resolve the current DCGM initial health and diagnostic scope issue: Compare platform power specifications, measured peak draw, PDU branch telemetry, breaker ratings and planned concurrency. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

C:

This tempting but unsafe shortcut is not a sufficient remediation for DCGM initial health and diagnostic scope: Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. The governing mechanism is: Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination. The proportionate action is: Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests.

D:

This alternative is relevant to Rack bonding grounding and ESD handling but does not directly resolve the current DCGM initial health and diagnostic scope issue: Inspect facility grounding tests, rack bonds, wrist-strap practice, ESD storage and chassis service records. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

Question 204

At a multi-node AI infrastructure handoff, the team records: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. After corrective work, which observable result BEST confirms that the requirement is met? Select ONE best answer.

  1. Sustained representative load runs without overcurrent events, while reserve capacity satisfies the approved electrical design.
  2. Required GPU health suites pass on the intended devices without disrupting unapproved production activity.
  3. A planned single-feed isolation leaves required servers within power limits and serving their intended workloads.
  4. Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites.

Explanation

Correct Answer

B:

This is an observable verified result, not merely a claimed configuration change: Required GPU health suites pass on the intended devices without disrupting unapproved production activity. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

Incorrect Answers

A:

This alternative is relevant to GPU node peak power budgeting but does not directly resolve the current DCGM initial health and diagnostic scope issue: Sustained representative load runs without overcurrent events, while reserve capacity satisfies the approved electrical design. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed.

C:

That option concerns Dual-feed rack power and PDU mapping. Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination.

D:

This tempting but unsafe shortcut is not a sufficient remediation for DCGM initial health and diagnostic scope: Claim that reading a DCGM health metric is equivalent to completing all active hardware and fabric diagnostic suites. Observed scenario: A cluster operator assumes that a healthy one-minute GPU metric proves all diagnostic tests were performed. The governing mechanism is: Telemetry observation and active diagnostic stress serve different purposes; intrusive checks need workload coordination. The proportionate action is: Run supported DCGM health checks and appropriately scoped diagnostics during approved windows, then review failed tests. A direct check is: Review DCGM entity discovery, health watches, diagnostic test results, operating load and error details. The acceptance condition is: Required GPU health suites pass on the intended devices without disrupting unapproved production activity.

Question 205

During an NVIDIA AI factory acceptance test, engineers discover: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Use approved authenticated BMC inventory endpoints, distinguish read-only calls from power actions and corroborate returned component state.
  2. Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.
  3. Follow each component's documented firmware update, activation, maintenance and rollback sequence with validation afterwards.
  4. Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness.

Explanation

Correct Answer

D:

This is an appropriate correction for ConnectX NIC detection and firmware inventory: Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Incorrect Answers

A:

This alternative is relevant to Redfish inventory and IPMI management boundaries but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Use approved authenticated BMC inventory endpoints, distinguish read-only calls from power actions and corroborate returned component state. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

B:

This tempting but unsafe shortcut is not a sufficient remediation for ConnectX NIC detection and firmware inventory: Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.

C:

This alternative is relevant to Firmware activation and reset sequencing but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Follow each component's documented firmware update, activation, maintenance and rollback sequence with validation afterwards. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Question 206

A production GPU cluster review identifies this condition: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears. Which underlying infrastructure principle BEST explains the issue? Select ONE best answer.

  1. Grounding and electrostatic protection limit personnel hazards and component damage; successful boot alone does not establish safe installation.
  2. Physical network presence, driver initialization and transport offload availability are distinct layers of the deployment.
  3. Redfish and IPMI expose hardware management data and operations, but endpoint selection and privileges determine their safety.
  4. Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.

Explanation

Correct Answer

B:

This is the governing infrastructure principle: Physical network presence, driver initialization and transport offload availability are distinct layers of the deployment. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Incorrect Answers

A:

That option concerns Rack bonding grounding and ESD handling. The proportionate action is: Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness.

C:

This alternative is relevant to Redfish inventory and IPMI management boundaries but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Redfish and IPMI expose hardware management data and operations, but endpoint selection and privileges determine their safety. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

D:

This tempting but unsafe shortcut is not a sufficient remediation for ConnectX NIC detection and firmware inventory: Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly. Physical network presence, driver initialization and transport offload availability are distinct layers of the deployment.

Question 207

In a professional AI infrastructure deployment, a technician observes: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Inspect management API target paths, access scopes, sensor values, chassis actions and independent host inventory.
  2. Compare platform power specifications, measured peak draw, PDU branch telemetry, breaker ratings and planned concurrency.
  3. Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.
  4. Inspect lspci, device firmware and driver metadata, link state, RDMA devices and host topology configuration.

Explanation

Correct Answer

D:

This is the direct operational evidence to investigate: Inspect lspci, device firmware and driver metadata, link state, RDMA devices and host topology configuration. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Incorrect Answers

A:

This alternative is relevant to Redfish inventory and IPMI management boundaries but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Inspect management API target paths, access scopes, sensor values, chassis actions and independent host inventory. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

B:

This alternative is relevant to GPU node peak power budgeting but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Compare platform power specifications, measured peak draw, PDU branch telemetry, breaker ratings and planned concurrency. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

C:

This tempting but unsafe shortcut is not a sufficient remediation for ConnectX NIC detection and firmware inventory: Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly. A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Question 208

The cluster operations team needs to evaluate this situation: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears. After corrective work, which observable result BEST confirms that the requirement is met? Select ONE best answer.

  1. Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.
  2. Sustained representative load runs without overcurrent events, while reserve capacity satisfies the approved electrical design.
  3. Inventory lists the intended devices accurately, while privileged resets require explicit approval and are logged.
  4. Every required ConnectX adapter enumerates with approved versions, expected ports and validated transport capability.

Explanation

Correct Answer

D:

This is an observable verified result, not merely a claimed configuration change: Every required ConnectX adapter enumerates with approved versions, expected ports and validated transport capability. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for ConnectX NIC detection and firmware inventory: Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly. A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears. Physical network presence, driver initialization and transport offload availability are distinct layers of the deployment.

B:

This alternative is relevant to GPU node peak power budgeting but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Sustained representative load runs without overcurrent events, while reserve capacity satisfies the approved electrical design. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

C:

This alternative is relevant to Redfish inventory and IPMI management boundaries but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Inventory lists the intended devices accurately, while privileged resets require explicit approval and are logged. Observed scenario: A GPU server boots with the intended compute devices but only one of its two high-speed network adapters appears.

Question 209

At a multi-node AI infrastructure handoff, the team records: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.
  2. Check rack dimensions, floor load and rail compatibility, then place approved chassis using the vendor's supported installation sequence.
  3. Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness.
  4. Capture component-specific firmware inventories and compare the approved system compatibility matrix before changes.

Explanation

Correct Answer

C:

This is an appropriate correction for ConnectX NIC detection and firmware inventory: Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for ConnectX NIC detection and firmware inventory: Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly. A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

B:

This alternative is relevant to DGX rack load and rail placement but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Check rack dimensions, floor load and rail compatibility, then place approved chassis using the vendor's supported installation sequence. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

D:

This alternative is relevant to Platform firmware component inventory but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Capture component-specific firmware inventories and compare the approved system compatibility matrix before changes. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

Question 210

During an NVIDIA AI factory acceptance test, engineers discover: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes. Which TWO complementary elements are needed: the correct technical action and the strongest independent diagnostic evidence? Select TWO answers.

  1. Check target versions before and after, update logs, activation requirements, node drain records and recovery options.
  2. Inspect lspci, device firmware and driver metadata, link state, RDMA devices and host topology configuration.
  3. Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly.
  4. Calculate peak supported server and fabric load, apply facility margin and verify branch and PDU capacity under representative load.
  5. Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness.

Explanation

Correct Answers

B:

This is the direct operational evidence to investigate: Inspect lspci, device firmware and driver metadata, link state, RDMA devices and host topology configuration. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

E:

This is an appropriate correction for ConnectX NIC detection and firmware inventory: Inventory NIC hardware, PCIe binding, supported firmware and host drivers before validating network-service readiness. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes. The governing mechanism is: Physical network presence, driver initialization and transport offload availability are distinct layers of the deployment.

Incorrect Answers

A:

This alternative is relevant to Firmware activation and reset sequencing but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Check target versions before and after, update logs, activation requirements, node drain records and recovery options. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

C:

This tempting but unsafe shortcut is not a sufficient remediation for ConnectX NIC detection and firmware inventory: Treat a lit transceiver LED as proof that the Linux RDMA stack and firmware are configured correctly. A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes. Physical network presence, driver initialization and transport offload availability are distinct layers of the deployment.

D:

This alternative is relevant to GPU node peak power budgeting but does not directly resolve the current ConnectX NIC detection and firmware inventory issue: Calculate peak supported server and fabric load, apply facility margin and verify branch and PDU capacity under representative load. Observed scenario: A replacement ConnectX card runs a different firmware and driver combination from adjacent cluster nodes.

NCP-AI Infrastructure (NCP-AII) Practice Tests