NCP-AII Question 115
Select 3You are tasked with deploying an AI application inside a containerized environment, and you need to ensure the NVIDIA GPU resources are accessible within the container. While installing the NVIDIA Container Toolkit on an Ubuntu-based system, which of the following steps must you complete?
- A
Add the NVIDIA Container Toolkit repository to your system.
- B
Install the NVIDIA Container Toolkit package using the package manager.
- C
Ensure the Docker service is stopped before installing the NVIDIA Container Toolkit.
- D
Verify the installation by running a container that uses GPU resources, such as nvidia/cuda.
- E
Recompile the Linux kernel to include GPU support.
Show answer and explanation
Correct answers: A, B, D
Explanation
Installing the NVIDIA Container Toolkit involves adding the appropriate repository, installing the toolkit package, and verifying functionality. These steps ensure that GPUs can be accessed from containers. Stopping the Docker service or recompiling the kernel are unnecessary steps for this process.
- A. Correct.
Correct: Adding the NVIDIA Container Toolkit repository is necessary to download the toolkit from the correct source.
- B. Correct.
Correct: Installing the NVIDIA Container Toolkit package is required to enable GPU support in containerized environments.
- C. Incorrect.
Incorrect: It is not required to stop the Docker service before installing the NVIDIA Container Toolkit. The toolkit integrates with Docker without requiring a service restart.
- D. Correct.
Correct: Verifying the installation by running a GPU-enabled container (e.g., nvidia/cuda) ensures that the toolkit is properly installed and functional.
- E. Incorrect.
Incorrect: Recompiling the Linux kernel is not necessary for installing or using the NVIDIA Container Toolkit. The toolkit works with precompiled kernel modules provided by the NVIDIA driver.