Tutorials for AMD Resource Manager

The tutorials here introduce you to core concepts of AMD Resource Manager. For shared vocabulary used in these tutorials and in Project settings, see Glossary: Projects and tutorials.

Projects and quotas Create a project, set GPU and compute quotas, configure storage (secrets) and users, and obtain the kubeconfig for your cluster. First tutorial in the getting-started series. See Projects and quotas.

Monitoring workloads and resource utilization Deploy a workload to your project via kubectl, observe it in Project Details, and view allocation on the Clusters, Cluster Details, and Dashboard pages. Part of the getting-started series. See Monitoring workloads and resource utilization.

Resource sharing and pre-emption Try out how projects borrow idle resources and how workloads are pre-empted when the quota-holding project needs them back. Uses two projects and sample workloads. Part of the getting-started series. See Resource sharing and pre-emption.

Resource utilization of AI workloads Observe resource utilization of your AI workloads (Projects page, Dashboard, Project dashboard) and adjust project quota when workloads are pending. Standalone; use any workload (e.g. fine-tuning). See Resource utilization of AI workloads.