NCP-AIO Workload Management Practice Question
An administrator must run a batch inference job that requires exactly two NVIDIA GPUs on a Kubernetes cluster managed by the NVIDIA GPU Operator. Which pod specification field should be used to request those GPUs?
⚠ Common exam trap
The trap here is treating nvidia.com/gpu like CPU or memory and specifying it only under resources.requests, which the API server rejects for extended resources.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
resources.limits with the key nvidia.com/gpu set to 2
GPUs exposed by the NVIDIA device plugin appear to Kubernetes as the extended resource nvidia.com/gpu. Extended resources must be declared in resources.limits, and setting that limit to 2 makes the scheduler reserve two GPUs on a suitable node and pass the devices into the container.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
resources.requests with the key nvidia.com/gpu set to 2 and no limit
Why it's wrong here
Extended resources such as nvidia.com/gpu cannot be requested without an equal limit; the Kubernetes API rejects a pod that specifies only a request for an extended resource. The scheduler would never admit this pod, so the batch inference job would remain pending indefinitely rather than starting with two GPUs.
- ✗
annotations with the key nvidia.com/gpu.count set to 2
Why it's wrong here
Annotations are free-form metadata and are not interpreted by the scheduler or the device plugin for resource allocation. Setting an annotation value of 2 changes nothing about scheduling or device injection, so the pod would run without any GPU and the inference job would fail or fall back to CPU.
- ✗
nodeSelector with the key nvidia.com/gpu.count set to 2
Why it's wrong here
nodeSelector matches node labels, not resource quantities or counts. There is no standard label expressing a GPU count of two, so this selector would either match nothing or match arbitrary nodes, and it would not allocate any GPU device to the container, leaving the job without accelerators.
- ✓
resources.limits with the key nvidia.com/gpu set to 2
Why this is correct
The NVIDIA device plugin advertises GPUs as the extended resource nvidia.com/gpu, and extended resources must be requested through resources.limits. Setting the limit to 2 causes the scheduler to place the pod on a node with two allocatable GPUs and injects those devices into the container, which meets the exact requirement.
About these practice questions
One of 309 original NCP-AIO practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official NVIDIA exam blueprint
This NCP-AIO practice question is part of Courseiva's free NVIDIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the NCP-AIO exam.