NCP-AIO Installation and Deployment Practice Test 2 — 15 Questions

Question 1

Which TWO actions should be performed to optimize GPU performance for a multi-tenant AI workload?

Accepted Answer

Enable MIG (Multi-Instance GPU). To optimize performance in a multi-tenant environment, isolation and scheduling efficiency are key. MIG allows for hardware-level isolation, preventing one tenant from impacting another, while proper node scheduling ensures that workloads are placed on the most appropriate hardware. Together, these strategies maximize hardware efficiency and protect the performance of critical workloads from noisy neighbors in a shared cluster environment.

Answer

Reduce the Kubernetes API polling interval

Answer

Disable the NVIDIA Container Toolkit

Answer

Use the standard Linux kernel scheduler

Question 2

An administrator is deploying NVIDIA AI Enterprise on a bare-metal server. Which component is strictly required to enable GPU virtualization for multiple virtual machine instances?

Accepted Answer

NVIDIA vGPU Manager. NVIDIA AI Enterprise relies on the NVIDIA vGPU software stack to slice a single physical GPU into multiple virtual instances. Installing the vGPU manager on the hypervisor is the foundational step for orchestration, allowing VMs to access GPU resources via the guest driver. Without this virtualization layer, instances would lack the necessary hardware acceleration APIs required for intensive AI inference or training workloads in multi-tenant environments.

Answer

NVIDIA Base Command

Answer

NVIDIA Collective Communications Library (NCCL)

Answer

CUDA Toolkit

Question 3

Which environment variable must be set in a container's configuration to ensure the application uses a specific GPU when multiple are available?

Accepted Answer

NVIDIA_VISIBLE_DEVICES. The CUDA_VISIBLE_DEVICES environment variable is the standard mechanism for controlling GPU visibility in applications. By setting this variable, an administrator can restrict a container process to only see and use specific physical GPUs. This is fundamental for managing resource assignment in multi-GPU nodes, ensuring that workload scheduling is enforced and that processes do not interfere with each other by grabbing unassigned devices.

Answer

CUDA_DEVICE_ORDER

Answer

GPU_DEVICE_INDEX

Answer

KUBERNETES_GPU_LIMIT

Question 4

Which NVIDIA container runtime configuration is mandatory for ensuring that containers can successfully interact with physical GPU hardware within a standard Docker or Kubernetes environment?

Accepted Answer

nvidia-container-runtime. The NVIDIA container runtime acts as a bridge between the containerized application and the host's NVIDIA driver. It intercepts the container creation process to inject the necessary GPU device nodes and libraries (like CUDA) into the container's environment. Without this runtime, the container lacks the low-level visibility required to interface with the GPU, leading to runtime errors when the application attempts to execute GPU-accelerated code or libraries.

Answer

runc

Answer

crun

Answer

kata-runtime

Question 5

An administrator is deploying NVIDIA AI Enterprise on a bare-metal server. Which prerequisite must be verified to ensure the NVIDIA AI Enterprise suite operates correctly with the underlying hardware acceleration?

Accepted Answer

The host kernel version must be verified against the NVIDIA Driver compatibility matrix.. Verifying that the installed GPU driver version matches the specific NVIDIA AI Enterprise release requirements is critical. Mismatched drivers lead to kernel module failures or inability to access GPU compute resources, preventing container orchestration platforms like Kubernetes from scheduling AI workloads effectively. This verification step ensures the software stack correctly interfaces with the NVIDIA GPU hardware, maintaining the performance and stability required for enterprise-grade AI production environments.

Answer

The server BIOS must support UEFI Secure Boot for all GPU firmware updates.

Answer

The server must have an active connection to the NVIDIA NGC private registry.

Answer

The GPU must be partitioned into at least four Multi-Instance GPU (MIG) slices.

Question 6

Refer to the exhibit. An administrator is deploying the NVIDIA GPU Operator using a ClusterPolicy. The administrator notices that GPU pods are failing to initialize. Based on the provided manifest, what is the most likely cause?

Accepted Answer

The sandbox feature is enabled without proper host-level support for GPU isolation.. The configuration includes 'sandbox: enabled: true', which indicates that the operator is attempting to use GPU virtualization or confidential computing features without potentially having the required host-level infrastructure support. Enabling sandbox mode requires specific firmware and kernel settings that, if missing, cause the driver to fail during container initialization. This configuration is critical because it fundamentally alters how the container runtime interacts with the host GPU resources.

Answer

The driver version is incompatible with the requested sandbox mode configuration.

Answer

The default runtime is set to 'container-runtime' instead of 'nvidia'.

Answer

The driver repository URL is formatted incorrectly for the NVIDIA NGC registry.

Question 7

An enterprise environment requires strict security for AI deployments. Which THREE of the following configurations are required to implement NVIDIA Confidential Computing on supported hardware?

Accepted Answer

Enable AMD SEV or Intel TDX at the host BIOS level.. NVIDIA Confidential Computing secures data in use by encrypting the GPU memory and protecting the execution environment. Implementing this requires specific hardware (H100 or later), a compatible software stack including the GPU operator with confidential computing enabled, and the underlying system firmware configuration. These components form a chain of trust that ensures the GPU can operate in an isolated, encrypted memory space, which is essential for sensitive workloads.

Answer

Disable the NVIDIA Container Toolkit to bypass runtime restrictions.

Answer

Set the GPU persistence mode to 'off' to ensure security.

Question 8

An AI operations engineer is deploying NVIDIA AI Enterprise on a Kubernetes cluster using the NVIDIA GPU Operator. During the installation phase, the cluster nodes fail to discover the available NVIDIA GPUs, and the device plugin pod remains in a CrashLoopBackOff state. Upon inspecting the logs, the engineer notices an error indicating a version mismatch between the host NVIDIA driver and the container runtime interface requirements. Which installation step was most likely overlooked?

Accepted Answer

Configuring the NVIDIA container runtime as the default or an available low-level runtime in the containerd configuration file on each node.. Ensuring the proper installation and matching of the NVIDIA container toolkit alongside the appropriate driver branch is critical for the GPU Operator. When node components fail to initialize or enter crash loops, it frequently stems from missing or misconfigured container toolkit integrations that allow the runtime to interface correctly with the underlying hardware drivers.

Answer

Increasing the memory limits on the Kubernetes API server deployment manifests to accommodate larger cluster state caches.

Answer

Enabling the kubelet dynamic resource allocation feature gate across all control plane nodes in the cluster specification.

Answer

Installing the latest upstream Linux kernel headers directly from the distribution package manager without rebooting.

Question 9

An AI operations team is deploying a Kubernetes cluster that will run mixed workloads: some pods need exclusive access to entire GPUs, while others must share a single GPU among multiple containers. The cluster nodes use data center GPUs that support Multi-Instance GPU (MIG). The team wants the GPU Operator to manage driver and device plugin lifecycle, and they want the scheduler to be aware of both whole-GPU and MIG-partitioned resources. They have already installed the NVIDIA GPU Operator with default settings. What must they do to enable MIG-backed scheduling on the nodes?

Accepted Answer

Set the node label nvidia.com/mig.config to the desired MIG profile and ensure the GPU Operator's MIG Manager is enabled.. MIG-backed scheduling in a GPU Operator-managed cluster is enabled by labeling nodes with nvidia.com/mig.config set to a supported profile. The MIG Manager component then applies the profile and exposes MIG devices to the device plugin, making them visible to the Kubernetes scheduler. Manual MIG commands, runtime environment variables, or non-existent installation flags do not provide the integrated, operator-driven configuration required.

Answer

Install the NVIDIA GPU Operator with the mig.strategy=mixed flag and annotate each pod with nvidia.com/mig-profile.

Answer

Run nvidia-smi mig -cgi on each node manually before scheduling any pods, then restart the kubelet.

Answer

Enable the NVIDIA_VISIBLE_DEVICES=all environment variable in the device plugin DaemonSet and set the nvidia.com/gpu resource limit to 0.

Question 10

An AI operations engineer is deploying the NVIDIA GPU Operator on a Kubernetes cluster that must schedule GPU workloads across nodes with different GPU models (A100 and H100). The cluster uses containerd as the container runtime. Which component of the GPU Operator is responsible for labeling nodes with GPU model and count so that the scheduler can place pods correctly?

Accepted Answer

NVIDIA k8s-device-plugin. The k8s-device-plugin is the component that discovers GPUs and exposes them as Kubernetes resources, while also applying node labels that describe GPU model and count. This allows the scheduler to place pods based on GPU requirements. Other components like NFD and the Container Toolkit support the overall stack but do not directly advertise GPU resources for scheduling.

Answer

NVIDIA GPU Operator node-feature-discovery

Answer

NVIDIA GPU Operator operator-lifecycle-manager

Answer

NVIDIA Container Toolkit

Question 11

An AI operations team is deploying NVIDIA AI Enterprise on a VMware vSphere cluster with NVIDIA vGPU. They need to ensure that vMotion is supported for VMs using vGPU. Which vGPU mode must be configured to allow vMotion?

Accepted Answer

vGPU with vMotion support enabled. To support vMotion with NVIDIA vGPU, the vGPU type must be one that includes vMotion capability. NVIDIA offers vGPU types that are vMotion-enabled, allowing live migration of VMs with vGPU. This requires vSphere 6.7 or later and compatible NVIDIA drivers. The other options either do not support vMotion or are not applicable. The correct configuration is to use a vGPU type that explicitly supports vMotion, which is typically indicated in the vGPU profile name or documentation.

Answer

vGPU with MIG partitioning

Answer

vGPU with time-sliced scheduling

Answer

vGPU with SR-IOV

Question 12

A DevOps engineer is deploying the NVIDIA GPU Operator on a Kubernetes cluster to manage GPU resources. The cluster nodes are running a compatible Linux distribution with NVIDIA GPUs. Which TWO prerequisites must be met before installing the GPU Operator? (Choose two.)

Accepted Answer

The nodes must have the NVIDIA driver installed if the Operator is configured to skip driver installation.. The GPU Operator can manage the driver installation, but if configured to skip it, the driver must be pre-installed. Also, if the Operator builds the driver, kernel headers are required on the nodes. These are key prerequisites. Other options like pre-installing the device plugin or container runtime are not needed because the Operator handles them. StorageClass is unrelated.

Answer

The cluster must have a default StorageClass defined for persistent volumes.

Answer

The Kubernetes cluster must have the NVIDIA device plugin already installed.

Answer

The container runtime must be configured to use the NVIDIA container runtime.

Question 13

A platform engineer is preparing a Kubernetes cluster for NVIDIA GPU Operator installation. The cluster nodes run Ubuntu 22.04 with containerd as the container runtime, and the engineer wants the GPU Operator to manage the NVIDIA driver lifecycle. Which prerequisite must be satisfied before the GPU Operator can successfully deploy the driver container?

Accepted Answer

The node must have the Linux kernel headers and build tools available for the driver container to compile the kernel modules.. The GPU Operator's driver container compiles NVIDIA kernel modules on the host, so the host must have kernel headers and build tools available. This allows the driver to match the running kernel. Without these prerequisites, the driver container cannot build the modules, and the Operator will not be able to load the driver, causing GPU resources to be unavailable.

Answer

The node must have the NVIDIA Container Toolkit installed and configured for containerd.

Answer

The node must have the nvidia-persistenced daemon running to maintain GPU state.

Answer

The node must have the NVIDIA driver pre-installed and loaded.

Question 14

A company is deploying NVIDIA AI Enterprise on a Kubernetes cluster using the NVIDIA GPU Operator. The security team requires that GPU workloads run with the highest level of isolation and that the operator itself follows the principle of least privilege. Which configuration should the engineer apply to the GPU Operator to meet these requirements?

Accepted Answer

Use the 'nvidia-gpu-operator' with 'sandbox' mode and 'secure' boot enabled. For high isolation and least privilege, the GPU Operator should be configured with sandbox mode (e.g., Kata Containers) and secure boot. Sandbox mode provides VM-level isolation for GPU workloads, while secure boot ensures only trusted drivers are loaded. This meets the security team's requirements without compromising functionality. Other options either do not address privilege or break essential operations.

Answer

Deploy the GPU Operator with 'hostNetwork' disabled and 'privileged' set to 'false'

Answer

Enable the 'mig' strategy and set 'migManager' to 'true'

Answer

Set 'driver.enabled' to 'false' and install drivers manually

Question 15

A DevOps engineer is installing the NVIDIA GPU Operator on a Kubernetes cluster using Helm. The cluster has nodes with different GPU models, and the engineer wants to ensure that the Operator only deploys on nodes with NVIDIA GPUs. Which action should the engineer take during installation?

Accepted Answer

Apply a node label to GPU nodes and configure the GPU Operator to use a node selector.. To ensure the GPU Operator only deploys on nodes with NVIDIA GPUs, the engineer should label the GPU nodes and configure a node selector in the Helm chart. This restricts the Operator's pods to those nodes. The other options either misuse parameters or rely on non-existent automatic detection, and none achieve the goal of limiting deployment to GPU nodes.

Answer

Set the operator.defaultRuntime to nvidia and rely on automatic node detection.

Answer

Enable the driver.enabled=false flag to prevent the Operator from installing drivers on non-GPU nodes.

Answer

Use the NVIDIA GPU Operator's built-in affinity rules that automatically detect GPU nodes.