NCP-AIO Practice Test 4 — 10 Questions

Question 1

An administrator is deploying the NVIDIA GPU Operator on a Kubernetes cluster in an air-gapped environment. The cluster nodes have no internet access, and all container images must be pulled from a private registry. Which two actions are required to ensure a successful deployment? (Choose two.)

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Configure the GPU Operator to use the private registry by setting the image repository in the Helm chart values.. Air-gapped deployments require that all container images are available in a private registry, and the GPU Operator must be configured to pull from that registry. Mirroring the images ensures availability, and setting the image repository in the Helm values directs the Operator to the correct location. These two actions are essential for a successful deployment without internet access.

Answer

Ensure that the private registry supports manifest lists for multi-architecture images.

Answer

Disable the driver container and install drivers manually on each node.

Answer

Set the environment variable 'AIRGAP=true' in the operator's deployment.

Question 2

When configuring a node for NVIDIA AI Enterprise in a Kubernetes environment, what is the primary function of the NVIDIA Container Toolkit?

Accepted Answer

Exposing GPUs to containers. The NVIDIA Container Toolkit allows container engines, such as Docker or containerd, to interface with physical GPUs. It enables the exposure of GPUs inside containers, ensuring that applications can access CUDA libraries and hardware acceleration. Without this toolkit, the container runtime cannot bridge the gap between the host's GPU hardware and the containerized process, effectively rendering the GPU unusable for AI workloads.

Answer

Provisioning virtual machines

Answer

Optimizing neural network layers

Answer

Managing Kubernetes network policies

Question 3

A platform team runs an NVIDIA AI cluster with the GPU Operator deployed. Users submit jobs directly with kubectl and frequently request whole GPUs even when their notebooks only need a fraction of one. The team wants Kubernetes itself to admit and queue jobs based on GPU demand without users changing their manifests. Which component should the team deploy to meet this requirement?

Accepted Answer

NVIDIA KAI Scheduler. The requirement is for Kubernetes to make admission and queueing decisions based on GPU demand while users submit normal manifests. KAI Scheduler is the NVIDIA component that provides queue-based, gang-aware scheduling for AI workloads and understands fractional GPU requests, so jobs wait rather than fail. The GPU Operator, MIG Manager, and DCGM Exporter each address installation, partitioning, or observability rather than scheduling admission.

Answer

NVIDIA MIG Manager configured for every GPU

Answer

NVIDIA GPU Operator with the device plugin enabled

Answer

NVIDIA DCGM Exporter with Prometheus alerting

Question 4

Which strategy is most effective for managing heterogeneous GPU clusters containing both older architectures (e.g., V100) and newer architectures (e.g., H100)?

Accepted Answer

Using Node Selectors and Taints/Tolerations.. Using Kubernetes Node Selectors and Taints/Tolerations allows administrators to route specific workloads to compatible hardware. This is essential because newer GPUs support features like MIG or specific precision formats that older hardware lacks. By properly tagging nodes based on architecture, AI Ops prevents 'Incompatible Device' errors and ensures that high-performance training jobs are scheduled on hardware that meets their specific computational requirements, optimizing cluster-wide performance and efficiency.

Answer

Enabling uniform scheduling across all nodes.

Answer

Configuring the NVIDIA Device Plugin to hide older GPUs.

Answer

Deploying a single unified container image for all jobs.

Question 5

A platform team is preparing a bare-metal Kubernetes cluster to run AI workloads. They want the GPU Operator to install and manage the NVIDIA driver automatically on each node. Which prerequisite must be satisfied on every worker node before the GPU Operator can succeed?

Accepted Answer

The node must have a supported Linux kernel with matching kernel headers available for the driver container to build against.. The driver container builds the NVIDIA kernel module at runtime against the host kernel, so matching kernel headers or development packages must be present on each node. The GPU Operator manages the container toolkit, DCGM, and device plugin itself, so those do not need to be preinstalled. vGPU Manager applies to virtualized deployments, not bare-metal driver automation.

Answer

The node must have the NVIDIA Container Toolkit preinstalled and configured.

Answer

The node must be running the NVIDIA Data Center GPU Manager (DCGM) as a systemd service before joining the cluster.

Answer

The node must have the NVIDIA vGPU Manager installed so the physical GPU can be partitioned before driver installation.

Question 6

A DevOps engineer needs to monitor GPU health metrics in real-time for workload management. Which tool provides the most granular visibility into GPU utilization and power consumption for individual containers?

Accepted Answer

NVIDIA DCGM Exporter.. The NVIDIA DCGM (Data Center GPU Manager) Exporter is the industry-standard tool for collecting fine-grained GPU telemetry. By integrating with Prometheus and Grafana, it allows administrators to track metrics like power usage, temperature, and utilization at the container level. This granular data is vital for proactive workload management, allowing for autoscaling based on actual hardware demand rather than generic CPU metrics.

Answer

Kubernetes Horizontal Pod Autoscaler (HPA) with CPU metrics.

Answer

Standard Linux 'top' command.

Answer

Docker stats command.

Question 7

When configuring a multi-tenant environment on an NVIDIA DGX system, how are MIG (Multi-Instance GPU) instances best provisioned?

Accepted Answer

Using the nvidia-smi command to define the desired MIG profiles.. MIG allows a single GPU to be partitioned into multiple isolated instances, each with dedicated compute and memory resources. Configuring MIG via the nvidia-smi tool at the OS level ensures that these partitions are visible to the container runtime. This is crucial for AI operations, as it enables safe, secure multi-tenancy where different users can run jobs simultaneously without resource contention or cross-tenant interference.

Answer

Through the BIOS settings, creating physical partitions on the GPU die.

Answer

By editing the /etc/nvidia/mig-config.json file directly.

Answer

Via the OS kernel boot parameters in the GRUB configuration.

Question 8

A Kubernetes cluster running the NVIDIA GPU Operator is shared by an inference team and a research team. The research team's training pods repeatedly evict the inference pods from GPUs, causing latency spikes in production. The administrator wants to guarantee that inference pods always get GPU access first. Which Kubernetes scheduling mechanism should be configured?

Accepted Answer

Assign a higher PriorityClass to the inference pods and enable preemption on the scheduler.. Priority and preemption are the native Kubernetes controls that decide which pods win when GPU resources are contended. Giving inference pods a higher PriorityClass lets the scheduler admit them first and evict lower-priority training pods when necessary, which directly satisfies the requirement that production inference always obtains GPU access.

Answer

Enable the NVIDIA MIG feature on all GPUs and dedicate a MIG instance to each inference pod.

Answer

Configure a taint on the GPU nodes and add the corresponding toleration only to the inference pods.

Answer

Create a PodDisruptionBudget for the inference deployment and set maxUnavailable to zero.

Question 9

Refer to the exhibit. The system has two GPUs. What is the most likely cause of the observed performance discrepancy?

Accepted Answer

Only one GPU is receiving the workload due to improper data distribution.. The exhibit shows one GPU heavily utilized while the second is idling despite similar memory consumption. This indicates a data parallelism imbalance, where one process is doing the bulk of the work. This is a common issue in multi-GPU setups where workload distribution is not correctly handled, leading to massive inefficiencies where expensive hardware is under-utilized, significantly increasing the time required for model training or inference tasks.

Answer

The GPUs have different amounts of physical memory.

Answer

The system is limited by the PCIe bus bandwidth.

Answer

The model is too small to be parallelized across two GPUs.

Question 10

Refer to the exhibit. The training job is showing intermittent "thermal throttling" warnings. Which configuration change is the most appropriate adjustment?

Accepted Answer

Lower the "power_limit_watts" value.. Thermal throttling occurs when the GPU reaches its maximum operating temperature and lowers its clock speed to prevent physical damage. While power limits can be adjusted to reduce heat, simply lowering the limit may negatively impact training performance. Adjusting the power limit to a value that balances thermal overhead with compute requirements is a necessary trade-off to ensure stable, consistent performance during long training runs without hardware damage.

Answer

Set "compute_mode" to "exclusive_process".

Answer

Disable ECC mode to lower the power consumption.

Answer

Set "persistence" to "disabled".