NCP-AII Question 112
Select 4You are tasked with setting up a containerized AI workload on a server equipped with NVIDIA GPUs running Ubuntu 20.04. To enable GPU acceleration in your containers, you need to install the NVIDIA Container Toolkit. Which of the following steps must you perform to successfully install the NVIDIA Container Toolkit?
- A
Add the NVIDIA package repository and GPG key to your system.
- B
Install the 'nvidia-container-runtime' package using the package manager.
- C
Enable GPU support by editing the Docker configuration file to include the NVIDIA runtime.
- D
Remove the default Docker runtime to avoid conflicts with the NVIDIA runtime.
- E
Restart the Docker daemon after updating the configuration.
Show answer and explanation
Correct answers: A, B, C, E
Explanation
To successfully install the NVIDIA Container Toolkit and enable GPU acceleration in containers, you need to add the NVIDIA package repository and install the necessary runtime. After installation, you must configure Docker to use the NVIDIA runtime and restart the daemon to apply the changes. Removing the default Docker runtime is unnecessary as it does not conflict with the NVIDIA runtime.
- A. Correct.
Correct: Adding the NVIDIA package repository and GPG key is necessary to ensure the system can download and verify the NVIDIA Container Toolkit packages.
- B. Correct.
Correct: The 'nvidia-container-runtime' package provides the essential runtime support for enabling GPU acceleration in containers.
- C. Correct.
Correct: Editing the Docker configuration file to include the NVIDIA runtime ensures that Docker can recognize and utilize the NVIDIA GPUs.
- D. Incorrect.
Incorrect: Removing the default Docker runtime is not required. The NVIDIA runtime can coexist with the default runtime, and Docker can be configured to use the NVIDIA runtime for GPU-enabled workloads.
- E. Correct.
Correct: Restarting the Docker daemon is required to apply the changes made to the Docker configuration file.