NCA - AI Infrastructure and Operations NCA-AIIO Practice Test 70
Question 690
An organization is evaluating an AI deployment: An accelerated server specification lists GPUs and CPUs but fails to allocate suitable network adapters or high-bandwidth storage. What solution MOST directly fits the stated need?
- Identify and protect shared control-plane and storage failure domains with appropriate redundancy.
- Design logical or physical management, storage and workload network separation appropriate to risk.
- Map CPU, GPU, NIC, storage and interconnect components to the complete data path and workload profile.
- Provide shared high-throughput storage and a protected checkpoint repository sized for concurrent jobs.
Explanation
Correct Answer
C:The appropriate response is to map CPU, GPU, NIC, storage and interconnect components to the complete data path and workload profile. This is specific to the case where an accelerated server specification lists GPUs and CPUs but fails to allocate suitable network adapters or high-bandwidth storage. The mechanism is that aI clusters need coordinated compute, communications and storage components rather than GPU devices alone. If ignored, a high-end accelerator server can be idle while missing network or storage resources constrain useful work. Validate the choice using this targeted evidence: A system diagram and benchmark show that every critical data path meets measured workload requirements.
Incorrect Answers
A:This alternative addresses a different requirement: A cluster stores all training metadata and management state on one server without recovery copies. It would be appropriate there to identify and protect shared control-plane and storage failure domains with appropriate redundancy. But the present scenario is about an accelerated server specification lists GPUs and CPUs but fails to allocate suitable network adapters or high-bandwidth storage. For it, the needed control is to map CPU, GPU, NIC, storage and interconnect components to the complete data path and workload profile.
B:This alternative addresses a different requirement: A research environment shares the same traffic interface for administrative SSH, storage recovery and large all-reduce jobs. It would be appropriate there to design logical or physical management, storage and workload network separation appropriate to risk. But the present scenario is about an accelerated server specification lists GPUs and CPUs but fails to allocate suitable network adapters or high-bandwidth storage. For it, the needed control is to map CPU, GPU, NIC, storage and interconnect components to the complete data path and workload profile.
D:This alternative addresses a different requirement: A new AI cluster relies on local workstation disks for shared multi-node training data and model checkpoints. It would be appropriate there to provide shared high-throughput storage and a protected checkpoint repository sized for concurrent jobs. But the present scenario is about an accelerated server specification lists GPUs and CPUs but fails to allocate suitable network adapters or high-bandwidth storage. For it, the needed control is to map CPU, GPU, NIC, storage and interconnect components to the complete data path and workload profile.
Question 691
An architecture review proposes: Assess RAPIDS GPU-accelerated data processing where the data format and operators are compatible. The underlying requirement is: An analytics department runs large tabular transformations on CPUs while its GPUs remain idle between training jobs. What technical rationale MOST strongly supports it?
- NIM packages supported AI inference components as deployable services that simplify integration and operations.
- cuDNN supplies GPU-optimized deep-learning primitives so application teams need not hand-code every operator.
- GPU data-processing libraries can accelerate suitable parallel transformations without redefining the analytical objective.
- The NGC catalog helps distribute optimized AI software artifacts with documented versions and dependencies.
Explanation
Correct Answer
C:The governing principle is that gPU data-processing libraries can accelerate suitable parallel transformations without redefining the analytical objective. It supports the measure to assess RAPIDS GPU-accelerated data processing where the data format and operators are compatible. because an analytics department runs large tabular transformations on CPUs while its GPUs remain idle between training jobs. Without the control, the preparation stage can dominate wall-clock time even when model training itself is highly accelerated. The most useful acceptance evidence is: A parity test confirms equivalent output data and measures end-to-end processing gain including transfers.
Incorrect Answers
A:That principle is relevant to this distinct case: A support team wants to deploy a packaged language-model microservice without maintaining custom inference server wrappers. The associated response would be to select a supported NVIDIA NIM microservice and verify its model, runtime and deployment prerequisites. It does not describe the central reason for this case, which is that gPU data-processing libraries can accelerate suitable parallel transformations without redefining the analytical objective.
B:That principle is relevant to this distinct case: A computer vision team has CUDA kernels but repeatedly rewrites convolution primitives while optimizing its neural network training code. The associated response would be to adopt a supported cuDNN-enabled framework and validate the convolution workload with representative input shapes. It does not describe the central reason for this case, which is that gPU data-processing libraries can accelerate suitable parallel transformations without redefining the analytical objective.
D:That principle is relevant to this distinct case: Data scientists repeatedly download unverified deep-learning container images from personal registries before sharing GPU nodes. The associated response would be to use vetted NGC catalog images and pin approved container versions with tested driver compatibility. It does not describe the central reason for this case, which is that gPU data-processing libraries can accelerate suitable parallel transformations without redefining the analytical objective.
Question 692
A data-center engineer must resolve the following situation: Multiple research teams compete for a shared GPU fleet and one account submits hundreds of low-priority jobs. What is the MOST appropriate response?
- Use gang-style scheduling or coordinated job admission where supported for dependent distributed workers.
- Cordon or drain the node through approved scheduler workflows and coordinate checkpointable tasks.
- Define resource quotas, scheduling priorities and fair-share policy with accountable ownership.
- Standardize supported host driver, container toolkit and GPU Operator lifecycle on eligible worker pools.
Explanation
Correct Answer
C:The appropriate response is to define resource quotas, scheduling priorities and fair-share policy with accountable ownership. This is specific to the case where multiple research teams compete for a shared GPU fleet and one account submits hundreds of low-priority jobs. The mechanism is that multi-tenant accelerator scheduling must balance fairness and business priority rather than first-come volume alone. If ignored, one tenant can monopolize GPUs and starve urgent inference or research commitments. Validate the choice using this targeted evidence: An admission test respects tenant quotas while an authorized high-priority job receives agreed capacity.
Incorrect Answers
A:This alternative addresses a different requirement: A team submits two tightly coupled distributed training pods but one starts long before resources for the other are ready. It would be appropriate there to use gang-style scheduling or coordinated job admission where supported for dependent distributed workers. But the present scenario is about multiple research teams compete for a shared GPU fleet and one account submits hundreds of low-priority jobs. For it, the needed control is to define resource quotas, scheduling priorities and fair-share policy with accountable ownership.
B:This alternative addresses a different requirement: A GPU node is marked for maintenance while the scheduler continues assigning new long-running training jobs to it. It would be appropriate there to cordon or drain the node through approved scheduler workflows and coordinate checkpointable tasks. But the present scenario is about multiple research teams compete for a shared GPU fleet and one account submits hundreds of low-priority jobs. For it, the needed control is to define resource quotas, scheduling priorities and fair-share policy with accountable ownership.
D:This alternative addresses a different requirement: GPU container images work on one host but fail on new workers because drivers and runtime integration differ. It would be appropriate there to standardize supported host driver, container toolkit and GPU Operator lifecycle on eligible worker pools. But the present scenario is about multiple research teams compete for a shared GPU fleet and one account submits hundreds of low-priority jobs. For it, the needed control is to define resource quotas, scheduling priorities and fair-share policy with accountable ownership.
Question 693
A GPU operations plan calls for: Check facility liquid cooling availability, heat exchangers, leak detection and vendor flow requirements. The reason is that a team wants to deploy liquid-cooled GPU systems in a building that has no validated coolant distribution network. Which principle is MOST relevant?
- Data-center maintenance planning must account for the coupling between compute load and cooling availability.
- High-density liquid cooling requires compatible facility infrastructure beyond server procurement.
- Power budgets must include realistic concurrent server demand and approved electrical safety margins.
- Local thermal conditions determine equipment behavior more directly than a remote room thermostat.
Explanation
Correct Answer
B:The governing principle is that high-density liquid cooling requires compatible facility infrastructure beyond server procurement. It supports the measure to check facility liquid cooling availability, heat exchangers, leak detection and vendor flow requirements. because a team wants to deploy liquid-cooled GPU systems in a building that has no validated coolant distribution network. Without the control, cooling interruptions or leaks can make new systems unusable or risk equipment damage. The most useful acceptance evidence is: Commissioning confirms required coolant conditions, monitoring alarms and planned safe isolation steps.
Incorrect Answers
A:That principle is relevant to this distinct case: Facilities schedules maintenance to an air-handling unit while operations keeps every GPU rack at full demand. The associated response would be to coordinate workload reduction and protected shutdown windows against available cooling redundancy. It does not describe the central reason for this case, which is that high-density liquid cooling requires compatible facility infrastructure beyond server procurement.
C:That principle is relevant to this distinct case: A proposed rack contains high-density GPU servers but its power distribution units cannot deliver full simultaneous load. The associated response would be to calculate worst-case electrical demand and verify PDU, branch-circuit and feed ratings with facilities staff. It does not describe the central reason for this case, which is that high-density liquid cooling requires compatible facility infrastructure beyond server procurement.
D:That principle is relevant to this distinct case: A rack runs multiple powerful accelerators but operators monitor only average room temperature, not intake air. The associated response would be to collect rack inlet and component thermal measurements during representative peak workloads. It does not describe the central reason for this case, which is that high-density liquid cooling requires compatible facility infrastructure beyond server procurement.
Question 694
An AI infrastructure team faces this requirement: A team ports a highly parallel matrix-multiplication workload to a CPU-only fleet despite abundant accelerator capacity. Which action BEST addresses the operational constraint?
- Select GPUs for appropriate data-parallel arithmetic while retaining CPU responsibility for orchestration.
- Profile kernel launch overhead, batching and occupancy before purchasing additional GPUs.
- Keep control-flow-heavy sequential orchestration on CPUs and map suitable parallel kernels to GPUs.
- Benchmark supported FP16, BF16 or other lower-precision modes against the application quality threshold.
Explanation
Correct Answer
A:The appropriate response is to select GPUs for appropriate data-parallel arithmetic while retaining CPU responsibility for orchestration. This is specific to the case where a team ports a highly parallel matrix-multiplication workload to a CPU-only fleet despite abundant accelerator capacity. The mechanism is that gPU architectures emphasize many parallel execution units, making them suitable for selected matrix operations. If ignored, the calculation may finish far slower than required despite allocating many general-purpose CPU cores. Validate the choice using this targeted evidence: An equivalent matrix workload is compared for numerical correctness and throughput across CPU and GPU.
Incorrect Answers
B:This alternative addresses a different requirement: A data scientist reports low GPU utilization even though thousands of tiny kernels launch every second. It would be appropriate there to profile kernel launch overhead, batching and occupancy before purchasing additional GPUs. But the present scenario is about a team ports a highly parallel matrix-multiplication workload to a CPU-only fleet despite abundant accelerator capacity. For it, the needed control is to select GPUs for appropriate data-parallel arithmetic while retaining CPU responsibility for orchestration.
C:This alternative addresses a different requirement: A data-center technician assumes one GPU is a universal replacement for a CPU in every application. It would be appropriate there to keep control-flow-heavy sequential orchestration on CPUs and map suitable parallel kernels to GPUs. But the present scenario is about a team ports a highly parallel matrix-multiplication workload to a CPU-only fleet despite abundant accelerator capacity. For it, the needed control is to select GPUs for appropriate data-parallel arithmetic while retaining CPU responsibility for orchestration.
D:This alternative addresses a different requirement: A team reduces model precision to boost GPU throughput without assessing the resulting numerical behavior. It would be appropriate there to benchmark supported FP16, BF16 or other lower-precision modes against the application quality threshold. But the present scenario is about a team ports a highly parallel matrix-multiplication workload to a CPU-only fleet despite abundant accelerator capacity. For it, the needed control is to select GPUs for appropriate data-parallel arithmetic while retaining CPU responsibility for orchestration.
Question 695
An AI infrastructure team faces this requirement: A hospital uses a deployed vision model to classify imaging requests within a fixed response time during business hours. Which action BEST addresses the operational constraint?
- Size online inference for bounded tail latency, concurrent sessions and availability at the actual model size.
- Preload supported model artifacts or maintain warm capacity based on startup latency and expected demand.
- Evaluate model parallelism with appropriate tensor or pipeline partitioning and high-bandwidth interconnect needs.
- Evaluate dynamic batching within the permitted queueing delay and model-specific latency budget.
Explanation
Correct Answer
A:The appropriate response is to size online inference for bounded tail latency, concurrent sessions and availability at the actual model size. This is specific to the case where a hospital uses a deployed vision model to classify imaging requests within a fixed response time during business hours. The mechanism is that inference runs a trained model for predictions and is often governed by response-time service-level objectives. If ignored, a deployment optimized only for bulk throughput can violate urgent per-request latency requirements. Validate the choice using this targeted evidence: A simulated concurrent request load stays below its p95 deadline and preserves agreed model output accuracy.
Incorrect Answers
B:This alternative addresses a different requirement: An autoscaled text-generation endpoint starts new model replicas only after a demand spike and sees long cold starts. It would be appropriate there to preload supported model artifacts or maintain warm capacity based on startup latency and expected demand. But the present scenario is about a hospital uses a deployed vision model to classify imaging requests within a fixed response time during business hours. For it, the needed control is to size online inference for bounded tail latency, concurrent sessions and availability at the actual model size.
C:This alternative addresses a different requirement: A foundation-model training plan exceeds the memory of a single accelerator even after supported precision reductions. It would be appropriate there to evaluate model parallelism with appropriate tensor or pipeline partitioning and high-bandwidth interconnect needs. But the present scenario is about a hospital uses a deployed vision model to classify imaging requests within a fixed response time during business hours. For it, the needed control is to size online inference for bounded tail latency, concurrent sessions and availability at the actual model size.
D:This alternative addresses a different requirement: An inference service receives many short requests but operates each GPU one request at a time with poor utilization. It would be appropriate there to evaluate dynamic batching within the permitted queueing delay and model-specific latency budget. But the present scenario is about a hospital uses a deployed vision model to classify imaging requests within a fixed response time during business hours. For it, the needed control is to size online inference for bounded tail latency, concurrent sessions and availability at the actual model size.
Question 696
A team omits this measure: Apply least-privilege administration and protected policy management for DPU-hosted infrastructure controls. In this environment: A data-center design uses a DPU for security enforcement but permits every workload administrator to reconfigure its policies. Which resulting problem is MOST likely?
- A compromised host may gain inappropriate control of shared infrastructure functions if boundaries are absent.
- Invisible offload failures may be mistaken for model or GPU faults and prolong repair.
- A tenant administrator may bypass network policies or alter access outside the assigned workload.
- The project may buy or configure unsuitable hardware and fail to improve neural-network compute speed.
Explanation
Correct Answer
C:The primary exposure is that a tenant administrator may bypass network policies or alter access outside the assigned workload. This follows from a data-center design uses a DPU for security enforcement but permits every workload administrator to reconfigure its policies. where the omitted measure was to apply least-privilege administration and protected policy management for DPU-hosted infrastructure controls. The technical reason is that infrastructure isolation benefits disappear when untrusted tenants can modify offloaded enforcement rules. A good operational validation after repair would be: A controlled unauthorized-change attempt is denied and records an auditable policy event.
Incorrect Answers
A:This exposure arises in a different situation: A security architect wants infrastructure service isolation from host applications that may be compromised. There, the corresponding safeguard is to evaluate BlueField DPU-supported trusted infrastructure processing and its administrative trust boundaries. In the present case, the more direct risk is that a tenant administrator may bypass network policies or alter access outside the assigned workload.
B:This exposure arises in a different situation: A cluster sees intermittent network errors but its DPU ports and offload services are missing from health dashboards. There, the corresponding safeguard is to monitor supported BlueField interfaces, service state and relevant packet or error counters. In the present case, the more direct risk is that a tenant administrator may bypass network policies or alter access outside the assigned workload.
D:This exposure arises in a different situation: A platform team installs a DPU and assumes it will automatically accelerate every CUDA compute kernel. There, the corresponding safeguard is to assign appropriate network, security or storage infrastructure tasks to the DPU rather than model tensor computation. In the present case, the more direct risk is that a tenant administrator may bypass network policies or alter access outside the assigned workload.
Question 697
The design team faces this case: A data-center technician assumes one GPU is a universal replacement for a CPU in every application. It adopts: Keep control-flow-heavy sequential orchestration on CPUs and map suitable parallel kernels to GPUs. What evidence BEST validates the intended outcome?
- An equivalent matrix workload is compared for numerical correctness and throughput across CPU and GPU.
- A load run records peak allocated memory, bandwidth utilization and sustained task throughput.
- Profiling confirms eligible Tensor Core kernels run while accuracy remains within agreed tolerance.
- A profile shows sequential coordination remains CPU-resident while parallel compute is offloaded appropriately.
Explanation
Correct Answer
D:The strongest proof is: A profile shows sequential coordination remains CPU-resident while parallel compute is offloaded appropriately. It directly tests the safeguard to keep control-flow-heavy sequential orchestration on CPUs and map suitable parallel kernels to GPUs. The control addresses this specific context: A data-center technician assumes one GPU is a universal replacement for a CPU in every application. It matters because cPUs are strong general-purpose control processors while GPUs excel at many parallel numerical operations. Without it, small serial tasks can incur overhead or run inefficiently when forced onto accelerators.
Incorrect Answers
A:This observation would validate a different control: Select GPUs for appropriate data-parallel arithmetic while retaining CPU responsibility for orchestration. It fits the case where a team ports a highly parallel matrix-multiplication workload to a CPU-only fleet despite abundant accelerator capacity. It cannot prove success here; instead the required evidence is a profile shows sequential coordination remains CPU-resident while parallel compute is offloaded appropriately.
B:This observation would validate a different control: Compare usable memory capacity, bandwidth and software support alongside compute performance. It fits the case where a team chooses an accelerator based only on peak compute rating although the model nearly fills GPU memory. It cannot prove success here; instead the required evidence is a profile shows sequential coordination remains CPU-resident while parallel compute is offloaded appropriately.
C:This observation would validate a different control: Evaluate supported Tensor Core execution and numerical precision settings with quality tests. It fits the case where a transformer inference plan overlooks special hardware paths for accelerated matrix operations. It cannot prove success here; instead the required evidence is a profile shows sequential coordination remains CPU-resident while parallel compute is offloaded appropriately.
Question 698
An organization is evaluating an AI deployment: A GPU server repeatedly runs out of host memory when constructing large mini-batches from many source files. What solution MOST directly fits the stated need?
- Size host RAM and CPU data-loader workers for preprocessing, buffering and model-feed requirements.
- Include checkpoint size, frequency, retention and restore throughput in capacity and storage planning.
- Inspect CPU NUMA, PCIe and GPU-to-NIC affinity before assigning communication-heavy training jobs.
- Model total training-state memory and evaluate supported sharding or memory-saving strategies.
Explanation
Correct Answer
A:The appropriate response is to size host RAM and CPU data-loader workers for preprocessing, buffering and model-feed requirements. This is specific to the case where a GPU server repeatedly runs out of host memory when constructing large mini-batches from many source files. The mechanism is that gPU provisioning does not replace adequate CPU and system-memory resources for data preparation. If ignored, accelerators can remain idle while host preprocessing stalls or the operating system swaps heavily. Validate the choice using this targeted evidence: The data-loader benchmark maintains intended batch supply without swap pressure or CPU queue buildup.
Incorrect Answers
B:This alternative addresses a different requirement: A training program uses frequent multi-terabyte checkpoints but reserves storage for only its original dataset. It would be appropriate there to include checkpoint size, frequency, retention and restore throughput in capacity and storage planning. But the present scenario is about a GPU server repeatedly runs out of host memory when constructing large mini-batches from many source files. For it, the needed control is to size host RAM and CPU data-loader workers for preprocessing, buffering and model-feed requirements.
C:This alternative addresses a different requirement: Two GPUs share a host but one is connected through a constrained PCIe path to the preferred network adapter. It would be appropriate there to inspect CPU NUMA, PCIe and GPU-to-NIC affinity before assigning communication-heavy training jobs. But the present scenario is about a GPU server repeatedly runs out of host memory when constructing large mini-batches from many source files. For it, the needed control is to size host RAM and CPU data-loader workers for preprocessing, buffering and model-feed requirements.
D:This alternative addresses a different requirement: A team estimates LLM training memory from the file size of weights and omits optimizer and gradient storage. It would be appropriate there to model total training-state memory and evaluate supported sharding or memory-saving strategies. But the present scenario is about a GPU server repeatedly runs out of host memory when constructing large mini-batches from many source files. For it, the needed control is to size host RAM and CPU data-loader workers for preprocessing, buffering and model-feed requirements.
Question 699
The team recommends this safeguard: Assess GPU Multi-Instance GPU partitions on supported hardware with compatible workload sizes. It applies because a trusted internal platform needs to run several smaller inference services on one MIG-capable GPU with reserved resources for each workload. Which concept justifies the design?
- Virtual GPU technology can expose allocated accelerator capabilities to separate virtual-machine workloads.
- GPU virtualization requires isolation at management, device allocation and data access layers.
- MIG creates hardware-partitioned accelerator instances on supported GPUs for eligible workloads.
- Accelerated virtual infrastructure needs supported drivers and entitlements as well as physical GPU capacity.
Explanation
Correct Answer
C:The governing principle is that mIG creates hardware-partitioned accelerator instances on supported GPUs for eligible workloads. It supports the measure to assess GPU Multi-Instance GPU partitions on supported hardware with compatible workload sizes. because a trusted internal platform needs to run several smaller inference services on one MIG-capable GPU with reserved resources for each workload. Without the control, without appropriate partitioning, one service may consume memory needed by other workloads. The most useful acceptance evidence is: Two supported partitioned workloads run within their assigned memory and compute resources without interference.
Incorrect Answers
A:That principle is relevant to this distinct case: A virtual-desktop deployment requires accelerated graphics for several isolated users on shared server GPUs. The associated response would be to evaluate supported NVIDIA vGPU capabilities, profiles and licensing against the application requirements. It does not describe the central reason for this case, which is that mIG creates hardware-partitioned accelerator instances on supported GPUs for eligible workloads.
B:That principle is relevant to this distinct case: An organization consolidates multiple AI tenants but has no process to constrain access to virtual GPU devices and images. The associated response would be to enforce hypervisor, guest-driver and tenant resource permissions for virtualized accelerated workloads. It does not describe the central reason for this case, which is that mIG creates hardware-partitioned accelerator instances on supported GPUs for eligible workloads.
D:That principle is relevant to this distinct case: An IT department deploys vGPU workloads but ignores required enterprise license and guest-driver version alignment. The associated response would be to verify supported vGPU software version compatibility and the applicable licensing lifecycle. It does not describe the central reason for this case, which is that mIG creates hardware-partitioned accelerator instances on supported GPUs for eligible workloads.
Question 700
A team omits this measure: Cordon or drain the node through approved scheduler workflows and coordinate checkpointable tasks. In this environment: A GPU node is marked for maintenance while the scheduler continues assigning new long-running training jobs to it. Which resulting problem is MOST likely?
- One tenant can monopolize GPUs and starve urgent inference or research commitments.
- Priority enforcement can waste GPU hours when jobs cannot resume from checkpoint state.
- A planned repair can unexpectedly cancel recently admitted training jobs.
- A large training queue can consume resources needed for customer inference requests.
Explanation
Correct Answer
C:The primary exposure is that a planned repair can unexpectedly cancel recently admitted training jobs. This follows from a GPU node is marked for maintenance while the scheduler continues assigning new long-running training jobs to it. where the omitted measure was to cordon or drain the node through approved scheduler workflows and coordinate checkpointable tasks. The technical reason is that workload lifecycle coordination prevents maintenance actions from racing with new GPU job placement. A good operational validation after repair would be: After drain begins, no new jobs land on the node and existing work follows the approved exit path.
Incorrect Answers
A:This exposure arises in a different situation: Multiple research teams compete for a shared GPU fleet and one account submits hundreds of low-priority jobs. There, the corresponding safeguard is to define resource quotas, scheduling priorities and fair-share policy with accountable ownership. In the present case, the more direct risk is that a planned repair can unexpectedly cancel recently admitted training jobs.
B:This exposure arises in a different situation: Training jobs are repeatedly preempted by higher-priority work but restart from scratch because the scheduler lacks recovery hooks. There, the corresponding safeguard is to coordinate preemption policy with checkpoint and requeue support for eligible training jobs. In the present case, the more direct risk is that a planned repair can unexpectedly cancel recently admitted training jobs.
D:This exposure arises in a different situation: An AI platform runs model-serving jobs and distributed training under one global namespace with no capacity boundaries. There, the corresponding safeguard is to separate workload classes with appropriate quotas, priorities and isolation rules in the scheduler. In the present case, the more direct risk is that a planned repair can unexpectedly cancel recently admitted training jobs.