Overview
Subscribe to a GPU pool to access shared GPU resources for deploying and running AI models. Configure pool resources and deploy a pod for your AI workloads.
Access Accelerators Tab
Open the Accelerators tab to subscribe to a GPU as a service pool. This displays all available accelerators and management options.
Initiate Pool Subscription
Select "Subscribe to Pool" to begin the subscription process. This action opens the selection of available GPU pools for your team.
Select GPU Pool
Choose GPU Pool
Choose an available GPU pool to view its resource specifications and associated costs. This selection allows you to proceed with configuring your subscription.
Specify Virtual Accelerators
Specify the number of virtual accelerators to subscribe to. This determines the quantity of physically available GPUs allocated from the chosen pool.
If the Pool sharing ratio is 1 a single pod will be deployed, if the sharing ratio is 2 or higher a pod will be deployed for each GPU selected.
Configure Pod Resources
Select the Image
Select the image to deploy for your GPU instance. This defines the software environment for your applications.
Choose Instance Type
Choose an instance type that correlates to the desired sizing of the pod you are deploying. This selection allocates CPU and RAM resources for your environment.
Select Ephemeral Storage
Select an ephemeral storage option to provide temporary disk space for your pod. This storage is non-persistent and is removed when the pod terminates. It also goes back to it's default state upon each reboot.
Add Persistent Storage
Add persistent storage if you have an existing volume created. This ensures your data remains available across pod restarts or redeployments. Persistent storage can be shared across multiple pods in a team.
Review and Subscribe
Review Subscription Details
Review all selected subscription details, including instance type, storage configurations, and container image. This summary allows you to verify your choices and understand the costs involved before finalizing the deployment.
Confirm Subscription
Confirm your subscription to complete the deployment initiation. This action begins the process of provisioning your GPU resources.
Monitor Subscription Status
Observe Subscription Progress
Observe the subscription progress notification. Provisioning your GPU resources takes time.
Subscription Activation Time
After confirmation, your GPUaaS subscription is not instant. Expect a 'Subscribing to pool...' notification for a minute or two while full deployment is proceeding in the background.
View Deployed Pod Details
Once the subscription is complete, view the details of your newly deployed pod. This confirms the successful provisioning and status of your GPU instance.