SLURM Compatible CLI Tool (SCC) provides AI developers with a CLI tool that is compatible with the SLURM syntax, helping AI developers use Slurm or Slurm-like command lines to perform actions on the SenseCore AI Compute Pool such as initiating and managing tasks and viewing node resources without the need of writing yaml files or configuring complex parameters when users are accustomed to minimal changes, so that the migration cost of users is minimized, and researchers can smoothly complete the transition of large-scale cluster resource management systems and fully enjoy the advantages of the optimized industrial-level large-scale computing power management system.
Command line compatible with the SLURM syntax
Support users to directly use slurm-native or slurm-like command lines, such as srun, scontrol, scancel, squeue, sinfo, and other commands to initiate and manage jobs
Enjoy the unique advantages of container technology
Support user-initiated jobs to enjoy the ease of deployment, environment consistency, and observability brought by the containerized distributed architecture, and significantly improve resource isolation and utilization based on the modern large-scale cluster resource management system
Quick application without the need of complex configuration
Automatically integrate built-in command line tools in the AI Cloud Lab, enabling users to easily complete installation and update in other application scenarios, and automatically trigger the encapsulation request when the command line interface is used to initiate related requests
Simple and easy to use
Free users from the need of learning kubectrl or other related command line syntaxes based on the k8s architecture and the need of writing complex parameters in the yaml file, and enable them to complete script writing and request through simple parameter configuration with the Slurm-like command line
Support users to initiate training tasks of different frameworks based on the Srun command line, configure the required resource specifications, and return the results in a specified way;
Support users to perform actions on submitted tasks based on scontrol, scancel, and other command lines, for example, view details and cancel a submitted task;
Support users to comprehensively view the available cluster workspace resources and the current user's task queuing condition based on sinfo, squeue, and other command lines.
Build commands on the CLI frontend to verify user options and parameters.
Transform, encapsulate, and build the request body on the CLI backend and send it, and format the output.
Authenticate identity through authentication mechanism to realize resource permission control.
Help you achieve new breakthroughs in business with professional AI solutions and advanced AI products