Why monitor GPU power as well as utilization?
Why A is the answer
It shows thermal and capacity limits
Power affects throttling and facility capacity.
In detail
- Power affects throttling and facility capacity.
- Power affects facility capacity too.
Use case
An illustrative scenario, not a real incident.
A GPU throttles under load, and power data shows it at its limit.
See it in different ways
flowchart TD A["DCGM"] --> B["Utilization"] A --> C["Memory"] A --> D["Temperature"] A --> E["Power"] A --> F["ECC and Xid errors"]
mindmap
root((Operations))
Monitoring
nvidia-smi
DCGM
Scheduling
Slurm
Kubernetes
Provisioning
Base Command Manager
Sharing
MIG
vGPUmindmap
root(("GPU power"))
n1["Device"]
n2["Throttling"]
n3["Facility"]
n4["Capacity"]Key point: Power shows thermal and capacity limits.
Why the other options are wrong
- B. It shows which user wrote the code
- It does not show who wrote code.
- C. It lists the versions of each model
- It does not list models.
- D. It replaces the need for a scheduler
- You still need a scheduler.