Exploring Your GPU Memory Clusters and Memory Fabric

Explore how your GPU memory fabric and memory clusters connect using the OCIcommand line reference for Compute.

This diagram shows how GPU memory clusters and memory fabric relate to compute clusters, hosts, and instances.

Relationship diagram that describes how GPU memory clusters and memory fabric relate to compute clusters. In this diagram, a single compute cluster can have multiple GPU memory clusters. Each GPU memory cluster is related to a single GPU memory fabric. A GPU memory fabric can be related to multiple bare metal hosts, and each bare metal host is related to an instance.

To further explore these relationships, use these sample commands:

To list all GPU memory fabrics in your regional dedicated capacity:
Use the compute-gpu-memory-fabric list command and required parameters:
```
oci compute compute-gpu-memory-fabric list --compartment-id ocid1.tenancy.oc1... # use the root compartment / tenancy OCID 
```
To get more details about a specific GPU memory fabric:
Use the compute-gpu-memory-fabric get command and required parameters:
```
oci compute compute-gpu-memory-fabric get --compute-gpu-memory-fabric-id ocid1.computegpumemoryfabric.oc1... 
```
To find the GPU memory fabric and instance associated with a bare metal host:
Use the compute-host get command and required parameters:
```
oci compute compute-host get --compute-host-id ocid1.computebaremetalhost.oc1... 
```
To find all bare metal hosts on a given GPU memory fabric:
Use the compute-host list command and required parameters:
```
oci compute compute-host list --network-resource-id ocid1.computegpumemoryfabric.oc1... 
```
To find the GPU memory cluster an instance belongs to:
Use the instance get command and required parameters:
```
oci compute instance get --instance-id ocid1.instance.oc1... 
```
To find the compute cluster, GPU memory fabric, and instance configuration associated with a GPU memory cluster:
Use the compute-gpu-memory-cluster get command and required parameters:
```
oci compute compute-gpu-memory-cluster get --compute-gpu-memory-cluster-id ocid1.computegpumemorycluster.oc1.... 
```
To list all instances in a GPU memory cluster:
Use the compute-gpu-memory-cluster-instance-summary list-compute-gpu-memory-cluster-instances command and required parameters:
```
oci compute compute-gpu-memory-cluster-instance-summary list-compute-gpu-memory-cluster-instances --compute-gpu-memory-cluster-id ocid1.computegpumemorycluster.oc1.... 
```
To see additional details about a compute cluster:
Use the compute-cluster get command and required parameters:
```
oci compute compute-cluster get --compute-cluster-id ocid1.computecluster.oc1.... 
```

For a complete list of flags and variable options for the Compute service CLI commands, see the command line reference for Compute.

Oracle Cloud Infrastructure Documentation

Exploring Your GPU Memory Clusters and Memory Fabric