PPPL Logo            PPPL Research Computing              


beowulf cluster at PPPL

Introduction

PPPL's Research Computing Center provides both batch and interactive scientific computing for PPPL, Princeton's graduate school programs located at PPPL, and the fusion community. It processes parallel computing, single CPU, as well as shared memory batch jobs in a variety of sub clusters.

The design goals of the center are high performance, ease of use, and support for the mid to small size jobs not favored by the large leadership computing sites.

Overview
Hardware
Software
Job queues

Using PPPL Clusters

    **** NEW JOB SECHEDULING PROTOCOL STARTING January 07 2017 ****



Getting Started
A sample SLURM batch script to submit a job
Graphical Menu Interface for Submitting Jobs
A Users Guide for the PPPL cluster
Other usage examples
Interactive use
Access, usage, and filesystem information
NX (NoMachine) - Virtual Desktop instructions
GPU Programming

Monitoring the Cluster (or your job)

Job scheduler information
Cluster usage statistics by queue * Internal only
Graphical monitoring of cluster node performance * Internal only
The Transp Grid Monitor (for running jobs)
The Transp Grid Monitor (for stopped jobs)

Recover your files

How to recover your files if you accidentally deleted them

Data Transfer using Globus Online

How to move large data sets from trusted labs to your project area at PPPL

Support

The research clusters are supported by the Systems and Network division, part of the Information Technology Department.

To open a support request, please submit a help desk ticket.

More Information

A simple MPICH-2 User's Guide
Kruskal Infiniband cluster User's Guide
Glossary of Terms
Frequently Asked Questions (FAQs)
Photos
Presentations

Additional Resources

There are additional resources on Princeton's main campus that are available for PPPL users.

CUDA/GPU programming
Hecate SGI System
Knights Landing Test Systems