Distributed versus GPU-accelerated frameworks: Practice Questions — Foundations of Accelerated Data Science (NVIDIA-Certified Associate: Accelerated Data Science)
Practice Questions: Distributed versus GPU-Accelerated Frameworks These multiple-choice questions are designed to help you prepare for the...
Practice Questions: Distributed versus GPU-Accelerated Frameworks
These multiple-choice questions are designed to help you prepare for the Distributed versus GPU-Accelerated Frameworks section of the NVIDIA-Certified Associate: Accelerated Data Science exam. Each question includes four options, the correct answer, and a brief explanation.
Which of the following best describes a key advantage of GPU-accelerated frameworks over traditional distributed CPU frameworks?
- A. They eliminate the need for data preprocessing.
- B. They provide massive parallelism for compute-intensive tasks.
- C. They require less memory bandwidth than CPU frameworks.
- D. They are optimized for sequential task execution.
Answer: B
Explanation: GPU-accelerated frameworks leverage thousands of cores to perform parallel computations efficiently, which is ideal for compute-intensive data science workloads. This parallelism is a primary advantage over CPU-based distributed frameworks.
In the context of data science workflows, what is a common challenge when transferring data between CPU and GPU memory?
- A. Data format incompatibility.
- B. High latency and bandwidth limitations during memory transfer.
- C. GPUs cannot access CPU memory directly.
- D. GPUs automatically cache all CPU data.
Answer: B
Explanation: Transferring data between CPU and GPU memory can introduce latency and bandwidth bottlenecks, which can reduce overall performance if not managed carefully.
Which framework is an example of a distributed computing system optimized for CPU clusters rather than GPU acceleration?
- A. RAPIDS cuDF
- B. Apache Spark
- C. Dask with CUDA support
- D. NVIDIA Triton Inference Server
Answer: B
Explanation: Apache Spark is a widely used distributed computing framework designed primarily for CPU clusters. While it can be extended to support GPUs, its core design is CPU-centric.
When scaling a data science workload, which scenario favors using a distributed CPU framework over a GPU-accelerated framework?
- A. When the workload is highly parallel and compute-bound.
- B. When the dataset is too large to fit into GPU memory and requires horizontal scaling.
- C. When low-latency inference is required.
- D. When using deep learning models optimized for CUDA.
Answer: B
Explanation: Distributed CPU frameworks can scale horizontally across many nodes and handle datasets larger than GPU memory, making them suitable when data size exceeds GPU capacity.
Which of the following best describes the role of Dask in GPU-accelerated data science workflows?
- A. It is a GPU-only framework that replaces all CPU operations.
- B. It provides a flexible parallel computing library that can integrate with GPUs via RAPIDS.
- C. It is a proprietary NVIDIA framework for data visualization.
- D. It is used exclusively for model deployment on GPUs.
Answer: B
Explanation: Dask is a parallel computing library that supports distributed computing on CPUs and can be integrated with GPU-accelerated libraries like RAPIDS to accelerate data science workflows.
What is a primary benefit of using GPU-accelerated frameworks in the end-to-end data science workflow?
- A. They reduce the need for data cleaning.
- B. They accelerate data preparation, model training, and inference by leveraging GPU parallelism.
- C. They automatically distribute workloads across multiple CPU nodes.
- D. They eliminate the need for Python programming.
Answer: B
Explanation: GPU-accelerated frameworks speed up various stages of the data science workflow by utilizing GPU parallelism, improving performance for data preparation, training, and inference.
Which statement correctly contrasts distributed frameworks and GPU-accelerated frameworks?
- A. Distributed frameworks always outperform GPU-accelerated frameworks.
- B. GPU-accelerated frameworks are limited to single-node execution.
- C. Distributed frameworks scale across multiple nodes, while GPU-accelerated frameworks focus on parallelism within a node.
- D. GPU-accelerated frameworks cannot be used with Python.
Answer: C
Explanation: Distributed frameworks enable scaling across multiple machines (nodes), while GPU-accelerated frameworks exploit parallelism within a single node's GPU(s). Both approaches can be complementary.
More in this topic
Ready to test your knowledge?
Put what you've learned into practice with a quick quiz and track your progress.
Test your knowledge →