[Sep-2025] NCP-AIO Pre-Exam Practice Tests Exam Questions and Answers for NVIDIA-Certified Professional Study Guide [Q33-Q57]

QUESTION 33
What is the primary purpose of using a container runtime interface (CRI) with BCM and Kubernetes in an AI environment?

To manage the lifecycle of containers (create, start, stop, delete).

To provide a standard interface for Kubernetes to interact with different container runtimes (e.g., Docker, containerd).

To handle networking for containers within the Kubernetes cluster.

To schedule pods onto nodes based on resource availability.

To encrypt container images at rest and in transit.

QUESTION 34
When using GPUDirect RDMA for inter-GPU communication, what component MUST be supported by the network interface card (NIC) to ensure optimal performance?

TCP Offload Engine (TOE)

Remote Direct Memory Access (RDMA)

Quality of Service (QOS)

Ethernet Flow Control

Jumbo Frames

QUESTION 35
Consider the following Dockerfile snippet:

The container will have the NVIDIA drivers pre-installed.

The container will automatically utilize all available GPUs without further configuration.

The NVIDIA Container Toolkit must be installed on the host to enable GPU access within the container.

The environment variable must be set externally to control GPU visibility.

This Dockerfile is incomplete and requires additional configuration to properly enable GPU access.

QUESTION 36
You observe that ‘nvsm’ is consuming a significant amount of CPU resources, even when the system is idle. You suspect that the high CPU usage is due to excessive logging. How can you reduce the logging verbosity of ‘nvsm’?

Edit the ‘nvsm.conf file and set the parameter to ‘ERROR or

Use the ‘nvsm -log-level ERROR command-line option when starting the service.

Modify the system’s syslog configuration to filter out ‘nvsm’ messages.

Uninstall and reinstall ‘nvsm’.

There is no way to control ‘nvsm’ logging verbosity.

QUESTION 37
You are deploying a Deep Learning model as a service using Triton Inference Server within a VMI container on Google Cloud Platform (GCP). You want to leverage TensorRT for optimized inference performance. Which of the following steps are necessary to ensure TensorRT is properly configured and used?

Ensure the Triton Inference Server container image includes the necessary TensorRT libraries and dependencies.

Convert the model to a TensorRT engine using ‘trtexec’ command before loading it into Triton.

Specify the ‘optimization’ parameter as ‘TRT in the model configuration file Cconfig.pbtxt) for Triton.

Configure the GCP instance with an NVIDIA GPU that supports TensorRT.

No special configuration is needed; Triton automatically detects and uses TensorRT if available.

QUESTION 38
You are using CUDA-Aware MPI for a distributed deep learning training job. After implementing CUDA-Aware MPI, you observe no performance improvement compared to regular MPI. What is the MOST likely reason?

The network interconnect is too slow.

The NCCL version is outdated.

The data being transferred is too small to benefit from GPU direct memory access.

The CPU is the bottleneck in the data loading pipeline.

The batch size is too small.

QUESTION 39
You are managing a Kubernetes cluster running GPU-accelerated AI workloads. You need to ensure that sensitive data, such as API keys and database credentials, used by these workloads is securely managed and accessible only to authorized pods. What are the best practices and Kubernetes features for achieving this?

Store the sensitive data directly in the container image.

Store the sensitive data in environment variables within the pod definition.

Use Kubernetes Secrets to store the sensitive data and mount them as files or environment variables in the pods.

Store the sensitive data in a centralized configuration management system (e.g., HashiCorp Vault) and use a sidecar container to retrieve the data and inject it into the pods.

Encrypt the sensitive data using a symmetric key and store the encrypted data in a ConfigMap.

QUESTION 40
You want to upgrade your BCM installation to the latest version. What is the recommended approach for upgrading BCM in a production environment?

Simply run the installation script for the new version, overwriting the existing installation.

Back up the BCM database, then run the upgrade script provided with the new version, following the documented upgrade procedure.

Delete the existing BCM installation and reinstall the new version from scratch, restoring the database from backup.

Stop the BCM service, download the new version, and manually replace the binary files.

Upgrade the NVIDIA drivers on all GPU nodes before upgrading BCM.

QUESTION 41
An organization has multiple containers and wants to view STDIN, STDOUT, and STDERR I/O streams of a specific container.
What command should be used?

docker top CONTAINER-NAME

docker stats CONTAINER-NAME

docker logs CONTAINER-NAME

docker inspect CONTAINER-NAME

QUESTION 42
You are managing an on-premises cluster using NVIDIA Base Command Manager (BCM) and need to extend your computational resources into AWS when your local infrastructure reaches peak capacity.
What is the most effective way to configure cloudbursting in this scenario?

Use BCM’s built-in load balancer to distribute workloads evenly between on-premises and cloud resources without any pre-configuration.

Manually provision additional cloud nodes in AWS when the on-premises cluster reaches its limit.

Set up a standby deployment in AWS and manually switch workloads to the cloud during peak times.

Use BCM’s Cluster Extension feature to automatically provision AWS resources when local resources are exhausted.

QUESTION 43
Describe a scenario where using MIG (Multi-lnstance GPU) on an NVIDIAAI 00 GPU within a Kubernetes cluster would be most beneficial. Explain why MIG is advantageous in that specific use case.

Running a single, large deep learning training job that requires the full resources of the A100 GPU.

Hosting a large number of small, independent AI inference services, each with modest GPU requirements.

Executing a computationally intensive scientific simulation that benefits from high GPU memory bandwidth.

Running a virtual desktop infrastructure (VDI) environment where each user requires dedicated GPU resources.

Distributing a single large training job across multiple GPUs on different nodes.

QUESTION 44
You are deploying a stateful application to your Kubernetes cluster running on NVIDIA hardware provisioned through BCM. This application requires direct access to a persistent volume on a high-performance NVMe drive. Which of the following methods is MOST appropriate for providing this access while ensuring high performance and data consistency?

Using a ‘hostPatW volume, directly mapping the NVMe drive’s path on the host node to the container.

Creating a PersistentVolumeClaim (PVC) backed by a cloud-based block storage service (e.g., AWS EBS, Azure Disk).

Using a Network File System (NFS) share mounted on the host and exposed to the container via a PersistentVolume.

Leveraging a local Persistent Volume with ‘volumeBindingMode: WaitForFirstConsumer’ and node affinity to ensure the pod is scheduled on the node with the NVMe drive.

Configuring a standard Persistent Volume Claim backed by a software-defined storage solution like Ceph or Rook.

QUESTION 45
When deploying BCM using Helm, which Helm chart parameter controls the resource limits for the BCM pods (CPU and Memory)?

bcm.resources.limits’

‘resources.bcmLimits’

resources.limits’

‘bcmResources.limits’

‘deployment.resources.limits’

QUESTION 46
You are tasked with deploying a deep learning framework container from NVIDIA NGC on a stand-alone GPU-enabled server.
What must you complete before pulling the container? (Choose two.)

Install Docker and the NVIDIA Container Toolkit on the server.

Set up a Kubernetes cluster to manage the container.

Install TensorFlow or PyTorch manually on the server before pulling the container.

Generate an NGC API key and log in to the NGC container registry using docker login.

QUESTION 47
A user reports that their AI training job running on a BCM-managed cluster is experiencing slow 1/0 performance. What steps would you take to diagnose and resolve the issue, considering the potential involvement of storage?

Check the network bandwidth between the compute nodes and the storage system using *iperf or similar tools.

Monitor the storage system’s performance metrics (IOPS, latency, throughput) using the storage vendor’s monitoring tools.

Verify that the correct storage class is being used for the persistent volumes used by the training job.

Inspect the pod’s logs for any I/O related errors or warnings.

Increase the pod’s CPU and memory limits to improve 1/0 performance.

QUESTION 48
A data scientist reports that a Run.ai job is consistently crashing with a ‘SIGKILL’ signal. After verifying that the job is not exceeding its resource limits (CPU, memory, GPU), what is the MOST likely reason for this signal, and how can you diagnose it further within the Run.ai environment?

The job is experiencing a kernel panic on the node. Check the system logs on the node for kernel-related errors.

The Kubernetes liveness probe is failing, causing the pod to be restarted. Inspect the pod’s events and liveness probe configuration.

The Run.ai agent is terminating the job due to exceeding a pre-defined time limit. Check the job’s configuration for any time limits or deadlines.

The job is being preempted due to higher priority tasks, resulting in SIGKILL. Check the job’s priority, quota, and resource usage history with ACM.

The process is consuming a lot of disk I/O.

QUESTION 49
You are responsible for deploying a deep learning model for real-time inference using Triton Inference Server from NGC. Latency is a critical requirement. Which of the following optimization techniques can you employ to minimize inference latency?

Increase the number of instances of the model deployed on the Triton server.

Use dynamic batching to aggregate multiple inference requests into a single batch.

Reduce the model’s precision (e.g., from FP32 to FP16 or INT8).

Disable CUDA graphs to improve CPU utilization.

Optimize the model’s architecture and code for GPU execution.

QUESTION 50
You are using BCM to manage a Kubernetes cluster with multiple GPU nodes. You need to enable GPU monitoring using Prometheus and the NVIDIA DCGM exporter. Outline the steps required to accomplish this. Choose the correct sequence:

0 1. Deploy the NVIDIA DCGM exporter as a DaemonSet in your Kubernetes cluster. 2. Configure Prometheus to scrape metrics from the DCGM exporter endpoints. 3. Install Prometheus in your Kubernetes cluster. 4. Verify GPU metrics are available in Prometheus.

0 1. Install Prometheus in your Kubernetes cluster. 2. Deploy the NVIDIA DCGM exporter as a DaemonSet in your Kubernetes cluster. 3. Configure Prometheus to scrape metrics from the DCGM exporter endpoints. 4. Verify GPU metrics are available in Prometheus.

0 1. Configure Prometheus to scrape metrics from the DCGM exporter endpoints. 2. Install Prometheus in your Kubernetes cluster. 3. Deploy the NVIDIA DCGM exporter as a DaemonSet in your Kubernetes cluster. 4. Verify GPU metrics are available in Prometheus.

0 1. Deploy the NVIDIA DCGM exporter as a Deployment in your Kubernetes cluster. 2. Configure Prometheus to scrape metrics from the DCGM exporter endpoints. 3. Install Prometheus in your Kubernetes cluster. 4. Verify GPU metrics are available in Prometheus.

0 1. Deploy the NVIDIA DCGM exporter as a DaemonSet in your Kubernetes cluster. 2. Configure the NVIDIA DCGM exporter endpoints. 3. Install Prometheus in your Kubernetes cluster. 4. Verify GPU metrics are available in Prometheus.

QUESTION 51
You need to configure BCM to send alerts when a GPU’s temperature exceeds a critical threshold. Where would you configure this alerting policy within BCM?

In the ‘bcm_config.yaml’ file.

Within the DCGM configuration files on the GPU nodes.

Through the BCM web interface, in the ‘Alerting Policies’ section.

By creating a custom Prometheus rule and integrating it with BCM.

Using the ‘nvidia-smi’ command-line tool to set temperature thresholds and trigger alerts.

QUESTION 52
You need to monitor the GPU utilization of individual MIG instances on your NVIDIAA100 GPU. Which of the following tools or methods can provide granular monitoring data for each MIG instance?

nvidia-smi’ alone, without any specific flags, provides per-MIG instance utilization.

The ‘top’ command in Linux provides GPU utilization information.

DCGM (Data Center GPU Manager) provides detailed monitoring metrics for individual MIG instances.

The ‘free command in Linux provides GPU memory usage information.

Use the Windows Task Manager to view GPU utilization.

QUESTION 53
You are deploying a PyTorch container from NGC that utilizes Tensor Cores. How can you verify that Tensor Cores are being effectively used during inference?

Check the container logs for messages indicating Tensor Core usage.

Use the ‘nvidia-smi’ command to monitor GPU utilization and check for high Tensor Core activity.

Analyze the training loss curve; a steep decline indicates Tensor Core usage.

Examine the CUDA code within the container to confirm explicit Tensor Core API calls.

Use the NVIDIA Nsight Systems profiler to analyze GPU kernel execution and identify Tensor Core operations.

QUESTION 54
You’re using Kubernetes with persistent volumes (PVs) backed by a network file system (NFS) to store your AI model checkpoints. You’ve noticed that checkpoint saving operations are slow and intermittently fail. After investigation, you suspect that the issue might be related to NFS locking. How can you diagnose and potentially resolve this issue?

Examine the logs of the NFS server and client for NFS lock-related errors or warnings (e.g., ‘NFS: v4 operation not supported).

Try disabling NFS locking on the client side by using the ‘nolock’ mount option (with caution, as this can lead to data corruption if multiple clients write to the same file concurrently).

Ensure that the NFS server supports NFSv4 locking and that the client is configured to use NFSv4.

Switch to using a different storage backend that doesn’t rely on NFS locking, such as a block storage device or an object storage system.

Increase the frequency of checkpoints to reduce the impact of individual checkpoint failures.

QUESTION 55
You are tasked with deploying a TensorFlow container from NGC on a Kubernetes cluster. The container requires specific NVIDIA drivers and libraries. Which of the following steps are essential to ensure successful deployment and GPU utilization?

Ensure the NVIDIA Container Toolkit is installed and configured on all worker nodes.

Deploy the container without specifying any resource limits or requests to allow it to utilize all available GPUs.

Verify that the NVIDIA drivers on the host machines match the versions required by the container.

Create a Kubernetes DaemonSet to automatically deploy and manage the NVIDIA device plugin on all nodes.

Bypass the NVIDIA Container Toolkit and directly use Docker to deploy the container.

QUESTION 56
You have a Run.ai cluster integrated with NVIDIA’s Cluster Manager (ACM). A data scientist reports that their job is being preempted frequently, even though they have a high-priority quot a. What are the MOST likely reasons for this preemption, assuming ACM is configured correctly?

The job is exceeding its memory limits.

Another job with a higher guaranteed quota and higher priority needs the resources.

The node where the job is running is being drained for maintenance.

The job is using an outdated CUDA driver version.

A higher priority job is requesting resources and preemption is enabled for that queue.

QUESTION 57
After updating the NVIDIA drivers on your NVSwitch-connected GPU server, ‘nvsm’ fails to start. The log file shows the following error: ‘Failed to initialize NVML’. Which of the following actions is MOST likely to resolve the issue?

Reinstall the operating system.

Downgrade to the previous version of the NVIDIA drivers.

Ensure that the NVIDIA kernel modules are correctly loaded and that the CUDA toolkit is installed and configured properly.

Disable SELinux.

Increase the allocated memory to the ‘nvsm’ process.