Application Scenarios of the Cluster Scheduling Components

The cluster scheduling components are mainly used in data centers to provide basic functions such as training and inference job scheduling and NPU device discovery. The user interface at the upper layer and user service logic are excluded and can be used only after an integration party performs secondary development. By using the cluster scheduling components, developers do not need to directly obtain NPU resources, and select and schedule processors. They only need to focus on the type and number of NPUs, number of CPUs, and memory size required by a job. The cluster scheduling components are recommended in the following application scenarios:

• A data center performs training and inference.
• A device contains Huawei NPUs.
• Deployment is based on containerization technologies.
• The cluster scheduling components run on the Kubernetes, which is used as the basic platform.