How do you track GPU cost by workload in Kubernetes?
I'm trying to get a clearer picture of our GPU spending. We can see total infrastructure cost, but it's difficult to determine which namespaces, workloads, or individual jobs…
idle GPU capacity
How are you tracking GPU costs by Kubernetes workload?
We're trying to get a clearer picture of our GPU spending. We can see total infrastructure costs, but it's difficult to identify which namespaces, workloads, or batch jobs…