SJTU HPCv0.1.0

A graduate researcher needs to submit, monitor, and cancel batch jobs on SJTU HPC clusters like pi or sy. The skill enables SSH-based job operations on the appropriate entry node, ensuring efficient use of computational resources.

A data scientist wants to transfer or manage large datasets on HPC storage systems (lustre or dssg). The skill connects to dedicated data entry nodes (e.g., data.hpc.sjtu.edu.cn) for secure file operations.

An HPC user needs to check their storage quota or update account details (e.g., password, email) without waiting for admin support. The skill uses the HPC REST API to handle these requests directly.

A principal investigator managing allocations across pi, sy, and kp groups needs to select the correct entry node for each cluster. The skill automatically maps partitions and storages to node groups for seamless access.

An admin or power user must refresh or request HPC API tokens securely, handling password uploads and deletion to prevent leaks. The skill guides step-by-step token acquisition and refresh using Python scripts.

Universities or research institutes subscribe to SJTU HPC services for their members. The skill automates user self-service tasks, reducing support costs and improving user satisfaction.

External clients pay for compute time on specific partitions. The skill enables them to submit jobs and manage data via a managed interface, abstracting complex SSH and API interactions.

Consultants offer training on HPC usage and integration of this skill into existing workflows. They help clients optimize job submissions and data management for better efficiency.