slurm

What problem does it solve?

This Skill enables teams to submit, manage, and monitor GPU-accelerated workloads on SRP's Slurm clusters using Apptainer containers, simplifying complex HPC workflows.

Core Features & Use Cases

• Submit & manage jobs: create and control Slurm batch scripts with SBATCH, srun, and apptainer-based executions for reproducible runs.
• Monitor & debug: inspect job status, queue, resource usage, and logs; access dashboards for overview and per-job metrics.
• Real-world use cases: train large models, run inference pipelines, and batch data processing with containerized environments on SRP clusters.

Quick Start

To run a quick GPU test, create a simple SBATCH script and submit it with sbatch, then monitor with squeue and sacct. For example: sbatch --partition=h100 --gres=gpu:1 --wrap "echo 'Hello Slurm'"