200 - Running Your First Job

This module walks through the full first-job process, from deciding what resources to request to checking that the job finished correctly. It is meant for researchers who can already log into ICE but are not yet comfortable writing and submitting a Slurm job.

In this module

1. 201 - The Four Numbers Before You Submit - estimate CPU cores, GPU VRAM, system RAM, storage, and wall time.
2. 202 - Finding Your Sizing Recipe - choose a starting point based on the type of workload you plan to run.
3. 203 - Choosing Your GPU - select the smallest compatible GPU that fits the job.
4. 204 - Using the VRAM Calculator - estimate GPU memory for LLM inference and understand the calculator’s limits.
5. 205 - Submitting Your First Job - create a Slurm script, submit it, monitor it, and check the output. A shorter command reference is also available.
6. 206 - Further Reading - continue into environments, file transfers, interactive jobs, job arrays, and troubleshooting.

By the end of this module

You should be able to explain the resources your job needs, choose an appropriate CPU or GPU setup, submit the job with sbatch, monitor it with squeue, and confirm that the expected output file was created.