Question 16
UnclassifiedYou need to debug a CUDA out-of-memory error interactively inside a container on a compute node. You want to start a shell inside the NGC container with GPU access and inspect memory usage live.
Correct answer: B
Explanation
Enroot is the right tool because it starts lightweight container shells on HPC systems with GPU access, letting you attach interactively to the NGC image on the compute node. That makes it suitable for live debugging of a CUDA out-of-memory issue while inspecting memory usage inside the container.
Why each option is right or wrong
A. docker
B. enroot
Enroot is the NVIDIA-supported container runtime for HPC that can launch an interactive shell from an NGC image directly on a compute node with GPU passthrough, which is exactly what is needed to reproduce and inspect a CUDA OOM condition in situ. Unlike a batch-only launcher, it supports lightweight, rootless execution and interactive debugging inside the container, so you can watch live GPU memory behavior while the job is running.
C. pyxis