Quick start¶

This guide walks you through the minimum steps needed to start submitting jobs on the DIPC HPC clusters. By the end you will have an account, a working SSH connection, and a simple job running on the scheduler. Everything you need to begin using the clusters for your research.

No prior HPC experience is required. Each step includes only the essential commands; links to the full documentation are provided if you want to dive deeper into any topic.

Step 1: Request an account¶

If you don't have an HPC account yet, follow the instructions on the Accounts and access page to request one.

Once your account is approved you will receive a username and temporary password by email.

Step 2: Set up your SSH key¶

SSH keys are required to access the clusters. Open a terminal and generate a key pair:

ssh-keygen -t ed25519

ssh-keygen -t ed25519

Press Enter to accept the default file location and optionally set a passphrase.

Then upload your public key to the authentication server:

ssh your_username@account-setup.sw.ehu.es

A setup script will run automatically. Paste the contents of ~/.ssh/id_ed25519.pub when prompted.

Full details: Account Setup

Step 3: Connect to the cluster¶

Once your key is registered, connect via SSH:

ssh your_username@hyperion.sw.ehu.es

Replace your_username with your assigned username. If everything is set up correctly, you will land on a login node.

Host hyperion
    HostName hyperion.sw.ehu.es
    User your_username

Full details: How to connect

Step 4: Transfer your data¶

Every cluster provides a /scratch filesystem designed for job I/O. A personal directory /scratch/<your_username> is created automatically with your account. This is where you should store input files and run your jobs.

Upload files to the cluster with scp:

scp my_input.dat your_username@hyperion.sw.ehu.es:/scratch/your_username/

scp my_input.dat your_username@hyperion.sw.ehu.es:/scratch/your_username/

For larger or repeated transfers, use rsync (Linux/macOS) or a GUI client like WinSCP on Windows.

Full details: Data transfer

Storage and quotas

Step 5: Run your first job¶

All computation runs through the SLURM job scheduler. First, move to your scratch directory:

cd /scratch/your_username

Create a batch script called hello.slurm:

#!/bin/bash
#SBATCH --job-name=hello
#SBATCH --partition=general
#SBATCH --qos=regular
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=1gb
#SBATCH --time=00:05:00
#SBATCH --output=%x.%j.out
#SBATCH --error=%x.%j.err

echo "Hello from $(hostname) on $(date)"

Submit it and check the status:

sbatch hello.slurm
squeue -u your_username

When the job finishes, view the output:

cat hello.*.out

Full details: Job submission (SLURM)

What's next?¶