User Manual

Welcome to the GridUnesp User Manual! This section gathers all the information you need to access the cluster, install applications, create submission scripts, monitor your jobs, and optimize your processing.

Note

If you are new to GridUnesp, we recommend starting with the Quick Start Guide for an overview of the process.

Chapter Overview

Getting Started:

• Accessing the Cluster - How to log in to GridUnesp

• Transferring Files - How to transfer data to/from the cluster

Preparing the Environment:

• Installing Applications - How to install software and libraries

• Authentication with SSH Keys - Advanced SSH configuration

Job Submission:

• Running Simulations - Basic submission concepts

• Improving the Submission Script - SBATCH directives

• Job Array - Submitting multiple jobs

• Partitions and Resource Limits - System partitions and limits

Parallelism and Optimization:

• Shared Memory - OpenMP and threads

• Distributed Memory - MPI and distributed processes

• GPU Usage - Using GPUs

• Optimizing Performance - Optimization techniques

Advanced Topics:

• Containers - Using containers (Singularity)

• Installed Applications - Examples with specific applications

• job-nanny Script - Wrapper script for jobs

Reference:

• Complete Storage Guide - Details about the file systems

• Complete SLURM Directive Reference - SLURM reference commands

• Priority Policy - How job priority works

Typical Workflow

A typical workflow on GridUnesp follows these steps:

1. Access the cluster via SSH (Accessing the Cluster)

2. Transfer your data and scripts (Transferring Files)

3. Prepare the environment by loading modules or installing applications (Installing Applications)

4. Create a submission script with the appropriate directives (Running Simulations)

5. Submit the job with sbatch

6. Monitor the execution with squeue and scontrol (Monitoring Jobs)

7. Retrieve the results after completion

8. Transfer the results to your local machine

Where to Find Help

Resource	Description
Frequently Asked Questions	Answers to the most common questions
Troubleshooting	Diagnosis and resolution of problems
Best Practices	Recommendations for responsible use
support.ncc@unesp.br	Direct support from the team

Important

Before requesting support, check the Frequently Asked Questions and Troubleshooting sections. Most questions have already been answered there.

Practical Examples

Throughout the manual, you will find practical examples for different scenarios:

• Basic scripts: Simple serial jobs

• Scripts with modules: Loading software

• Parallel jobs: OpenMP and MPI

• Job arrays: Batch processing

• GPU jobs: Using accelerators

• Containers: Running with Apptainer

Tip

Adapt the examples to your specific needs. The scripts are provided as a starting point.

Next Steps

We recommend reading the first chapters in sequence for beginners. Advanced users can navigate directly to the topics of interest.

Start with: Accessing the Cluster

See also

• Infrastructure - Details about the available hardware

• Statistics - Cluster usage data

• Contact - Support channels