Generating a GPU report

Hitachi iQ Studio User Guide

Version
1.0.x
Audience
anonymous
Part Number
MK-26HIQS001-00
ft:lastEdition
2026-03-31

Generate a GPU report that summarizes GPU usage, allocation, and performance across your environment. The report includes the following sections:

GPU overview: Shows overall GPU capacity, usage, and health across the environment.
Component Description
GPU nodes Number of cluster nodes with GPUs available.
Total full GPUs Total physical GPUs detected across all nodes.
GPU utilization Overall GPU utilization percentage across the cluster.
MIG overview: Summarizes Multi‑Instance GPU (MIG) usage and configuration.
Component Description
MIG type Number of types with MIG enabled.
Total MIG devices Total MIG instances provisioned across the cluster.
MIG utilization Aggregate utilization across MIG instances.
GPU or Node and MIG or GPU summary by node: Provides a per‑node summary of GPU and MIG assignments.
Component Description
Node Kubernetes node name.
GPU model GPU model installed on the node (for example, A100, H100).
Allocatable GPUs GPUs allocatable to workloads on the node.
Available GPUs GPUs currently free on the node.
Used GPUs GPUs currently in use by workloads on the node.
Total GPUs Total GPUs on the node.
MIG device summary by node: Lists MIG devices on each node, including their configuration and status.
Component Description
Node Kubernetes node name.
MIG type MIG profile configured on the GPU (for example, 1g.10gb and 2g.20gb).
Available MIGs Number of MIG devices of this type currently free on the node.
Allocated MIGs MIGs allocated to workloads on the node.
Total GPUs Total MIGs on the node.
Detailed GPU and MIG resource status: Provides node‑level status for all GPU and MIG resources, including availability, allocation, and current usage.
Component Description
Resource type GPU or MIG resource type.
Resource name Device label (for example, GPU 0 and MIG‑2).
Instance ID Hardware or logical device identifier.
Status Indicates whether the GPU or MIG device is available or in use.
Occupied by Pod or namespace currently using the device (if applicable).
Node‑wise GPU and MIG status: Shows detailed GPU and MIG information for the selected node, including resource type, instance ID, device status, and the pod using the resource.
Component Description
Resource type Indicates whether the resource is a full GPU or a MIG device.
Resource name GPU or MIG profile name (for example, GPU 0 and MIG‑2).
Instance ID Hardware or logical device identifier.
Status Indicates whether the GPU or MIG device is available or in use.
Occupied by The workload or model currently using the resource.
GPU and MIG allocated pods: Displays which pods are using each GPU and MIG device.
Component Description
Namespace Kubernetes namespace of the pod.
Pod Pod name.
Node Node where the pod is scheduled.
Resource type GPU or MIG resource requested by the pod.
Resource name Specific GPU or MIG device allocated.
Count Number of devices allocated to the pod.