You want a quick look at GPU processes on one host. What do you run?
Why C is the answer
nvidia-smi
nvidia-smi lists GPU processes.
In detail
- nvidia-smi lists GPU processes.
- It covers one host.
Use case
An illustrative scenario, not a real incident.
A quick nvidia-smi shows which processes hold a GPU.
See it in different ways
flowchart TD
A{"What do you need?"} -->|"Quick view of a host"| B["nvidia-smi"]
A -->|"Fleet health and diagnostics"| C["DCGM"]
A -->|"DGX system health"| D["NVSM"]mindmap
root((GPU tools))
One host
nvidia-smi
A fleet
DCGM
dcgmi diag
DGX
NVSMflowchart TD
Q{"What is the scope?"}
Q -->|"One host"| A0["nvidia-smi"]
Q -->|"Many GPUs"| A1["DCGM"]Key point: nvidia-smi lists processes.
Why the other options are wrong
- A. helm list
- That lists releases.
- B. kubectl top pods
- That shows pod CPU and memory.
- D. sacct
- That shows job history.