Courseiva
Free · No account needed · No credit card

NVIDIA Certified Professional: Generative AI LLMs Practice Test

352 questions with instant explanations, domain breakdown, and wrong-answer analysis. Built for the real exam.

Instant feedback after each answer
Full explanations included
Domain score breakdown
Real exam: 120 min

Sample questions with explanations

This is exactly what you see during practice — question, options, and a full explanation after you answer.

Q1Model Optimizationmedium
Full explanation →

An enterprise deploying a large language model on an NVIDIA A100 GPU experiences high memory bandwidth bottlenecks during autoregressive token generation. Which optimization technique specifically addresses this memory-bound phase by merging element-wise operations and reducing global memory round-trips?

AApplying static INT8 post-training quantization to all linear layers.
Enabling TensorRT-LLM custom kernel fusion for multi-head attention and activation blocks.Correct
CIncreasing the global batch size to maximize arithmetic intensity.
DSwitching from FlashAttention to standard vanilla self-attention mechanisms.

Kernel fusion combines multiple successive GPU operations, such as bias additions and activations, into a single CUDA kernel. This drastically reduces high-latency global memory read and write operations, directly mitigating the memory bandwidth bottleneck characteristic of autor…Read full explanation

Q2Model Optimizationmedium
Full explanation →

When optimizing a model using NVIDIA TensorRT, what is the primary benefit of enabling 'layer fusion' during the optimization process?

AIt increases the overall model weight precision
It reduces global memory access overheadCorrect
CIt automatically prunes redundant parameters
DIt replaces floating-point math with integer math

Layer fusion reduces the overhead of launching multiple small GPU kernels by combining them into a single, optimized kernel. This strategy minimizes the total number of reads and writes to global VRAM, which is often the bottleneck in modern deep learning. By keeping intermediate…Read full explanation

Q3Model Optimizationhard
Full explanation →

Refer to the exhibit. The TensorRT build process fails with a memory limit error. Which configuration adjustment is most likely to resolve this build-time error?

AReducing the number of hidden layers
Increasing the builder's workspace memory limitCorrect
CSwitching to a lower precision inference
DDisabling the TensorRT engine cache

The build error indicates that the workspace memory required for the optimization process exceeds the available VRAM or the limit specified by the build configuration. By explicitly increasing the workspace size in the builder configuration, the engineer provides the necessary sc…Read full explanation

What's covered

Questions are tagged by exam domain. Practice a specific domain or mix them all.

Untimed Practice

Answer at your own pace. Explanation and domain tag shown immediately after each answer.

Timed Practice

Countdown timer starts immediately. Results and domain scores shown at the end — just like the real exam.

Why practice here?

Full explanations on every question

Not just the right answer — you get exactly why each wrong option is wrong, so you learn the concept, not the answer.

Domain score breakdown

After each session see your score by exam domain so you know exactly where to focus study time.

100% free, forever

No subscription, no trial, no email wall. Start a session in under 10 seconds.

Exam-style questions

Scenario-based, precise wording, realistic distractors — written to match what you actually see on exam day.

← All NCP-GENL questionsNCP-GENL exam guideStudy guidePractice by domain