Diagram of the computing facility: What's on the floor now

SCD's supercomputer gallery: A history of computing at NCAR from 1963 to the present

The NCAR computing facility is managed and operated by the Scientific Computing Division (SCD).

NCAR computers: Complete user information

This page is a comprehensive guide to the NCAR computing facility: computing systems, allocations, usage, charges, job queues, job submission, remote computing, Mass Storage System, research data archives, models, and support for users.

Before you start using the NCAR computing facility:

1. You need to read and understand the Overview of computing at NCAR, which provides all basic user information.
2. You need to know how to work on UNIX systems. If you need help with this, consult these UNIX references.

Computer security

As computer security needs increase, SCD enhances the protection for all systems at UCAR via software upgrades, stricter password rules, and ongoing network changes. All changes affecting supercomputer users are announced in the Daily Bulletin. More details of the ongoing security changes are available to users who have logons and passwords to access the internal file Computer security enhancements.

Passwords are critical to our computer security. SCD provides Password guidelines and instructions for changing passwords on SCD computers to help you keep your account and SCD's computers secure.

Detailed information about computer security is provided on the website for the UCAR Computer Security Advisory Committee.

Computing systems for users

The Community Computing facility and the Climate Simulation Lab (CSL) share resources on the same supercomputers.

Production supercomputing systems run tasks such as large simulations or models that require massive computing and data-handling resources. Production systems at NCAR include:

• blueice (IBM p575+ cluster running AIX)
• bluevista (IBM p5-575 cluster running AIX)
• lightning (IBM e1350 cluster running Linux)

Data analysis and visualization systems support data preprocessing prior to a job run and data postprocessing after a run. Please see:

Data analysis and visualization Quick Start Guide

Note: The tempest system was decommissioned on April 23, 2008.

System and application testing systems are available to users when they are not being used by Consulting Services or the Supercomputing Systems Group. These computers have the same software as their comparable production systems, but the software versions may be more recent. These systems are also much smaller than the production supercomputers. Resources are provided at no charge because user jobs can be interrupted with little or no warning. These platforms include:

• bluedawn (IBM Cluster 1600 running AIX) - not currently available to users

Diagram of the NCAR computing facility

Obtaining computing allocations

• Obtaining computing services - Details for obtaining an allocation and a login for NCAR systems, plus an overview of available services.

Using NCAR computers

These documents provide user information for all parts of the NCAR computing facility:

• Accessing computing services - Instructions for logging in to NCAR systems.
• Overview of computing at NCAR - All new users: Read and understand this document
• The SCD user documentation catalog - All documents relevant to computing at NCAR
• Getting started on bluevista - IBM Power5 Cluster
• Getting started on blueice - IBM Power5+ Cluster
• Getting started on tempest - SGI Origin 3800
• Getting started with SSH at NCAR - Logging on to SCD production platforms
• Cleanscape FortranLint Source Code Analyzer User's Manual

Programming for parallel systems - The production systems operated by SCD are clusters of symmetric multiprocessor (SMP) computers. While you can run single-processor jobs on these cluster systems, parallel programming techniques make best use of their power, and parallel programming is required for compute-intensive tasks.

If you are not familiar with parallel programming, read SCD's brief introduction, Overview of parallel programming for SCD computers. SCD provides a concise discussion of basic terminology in Parallel programming terminology, which is a part of the fundamental concepts document for IBM systems: Critical concepts for compiling, building, and running jobs.

Charges for NCAR computing resources

• Charges for SCD computing resources
• Charges for CSL projects
• Charging and machine reports
• Reducing GAU charges on IBM systems

• Queues and charging for resource usage on blueice
• Queues and charging for resource usage on bluevista
• Queues and charging for resource usage on lightning
• Queues for batch jobs on tempest

Job submission and remote computing

• SSH: Getting Started with secure shell at NCAR
• Remote computing at NCAR (working from a modem via RAS, NCAR's remote access server)

Mass Storage System and research data archives

NCAR's Mass Storage System (MSS) is well connected to all of the systems in NCAR's computing facility. The MSS stores data used and generated by climate models and other programs that run on NCAR supercomputers. The MSS also supplies users with an extensive, high-quality research data archive.

• Mass Storage System
• MSS documentation
• NCAR research data archives

Models

The following production models are frequently run on NCAR systems. If you use any of these models, you need to study the information provided by these links:

• Climate models page introducing:
    ·   NCAR Community Climate Model (CCM)
    ·   Community Atmosphere Model (CAM)
    ·   Community Climate System Model (CCSM)
    ·   And others
• MM5 Community Model (The Penn. State/NCAR Mesoscale Model) MM5 mirror site
• Weather Research and Forecasting (WRF) Model
• Model for OZone And Related chemical Tracers (MOZART)

Support for users

• SCD Consulting Services
• User support