Last update: 05/07/2008
This introduction explains how the various types of systems in the NCAR computing facility provide various types of functionality for the user community. Different types of systems are suited to different tasks. This section will help you decide which systems best meet your needs.
Production supercomputers are generally used for computationally intensive algorithms in which data are used repeatedly to perform the calculations. These calculations and their data demands are so intensive that only this class of machine can perform them in a reasonable amount of time.
The most common supercomputer applications run at NCAR are time simulations of coupled systems of partial differential equations. Researchers use them to simulate fluid field changes over time in closed thermodynamic systems for planetary and stellar investigations. Typically such simulations approximate various thermodynamic and fluid states defined on a Eulerian grid governed by the Navier-Stokes equations or by derivations from these equations.
Data produced by simulations must be condensed for effective analysis and presentation. These data processing and visualization tasks typically require such a small fraction of the computational power of a supercomputer. So NCAR provides specialized platforms for analysis and visualization. This allows the supercomputers to be dedicated to computationally intensive tasks.
NCAR's production supercomputing systems provide a broad range of integrated resources to support applications of all sizes. Primary services on each system include:
- Compilers that make efficient use of the platform's resources by generating high performance executables.
- Libraries that support a wide variety of math, science, and engineering applications. These include parallelized math libraries optimized for the system architecture.
- A large allocation of disk space where codes can write temporary data.
- High-bandwidth access to a central, large-scale archive facility (the Mass Storage System) that stores data used and generated by climate models and other programs.
- High-bandwidth access to an extensive collection of high-quality research data.
- Tools that help software engineers develop large simulations. Tools are also provided for debugging codes and for analyzing code performance on the system.
- Operating systems that are configured to deliver dedicated high performance service to a few running codes. This allows users with large job runs to get a high percentage of the available resources. High performance codes can therefore receive better throughput.
NCAR's production supercomputing systems include bluevista, lightning.
Data analysis and visualization systems perform tasks that require the data movement capacity of a supercomputer, but not the computational power. Data analysis and visualization computers may require access to more data storage than a supercomputer.
These systems provide resources for preprocessing and postprocessing the data output by simulations running on supercomputers. Preprocessing involves obtaining and formatting input data. Postprocessing involves examining, editing, and processing data for use on another system.
Data analysis and visualization systems have to be well connected to high-bandwidth networks to move data quickly enough to avoid restricting data flow between computers and storage.
These systems require large amounts of memory and high memory bandwidth combined with massive data movement capacity to run large applications, sort data, move data between arrays, and move data into and out of storage.
To perform common data-processing and visualization tasks at NCAR, these systems are supplied with third-party vendor applications and libraries including:
NCAR's data analysis and visualization systems include a number of Linux clusters with fast processors and large memory. Please see the Data Analysis Cluster Quick Start Guide if you want to learn more about them.
NCAR's Mass Storage System is a central, large-scale data archive that stores data used and generated by climate models and other programs that run on NCAR supercomputers. Most of the data stored on the MSS is generated by global climate-simulation models, mesoscale weather models, and other earth-science models.
The MSS has high-speed access to all of the systems in NCAR's computing facility.
The CISL Research Data Archive is an ever-growing store of observed data and verified data (analyzed observations) that is used for major national and international research projects.
Overview of computing at NCAR - Table of contents
If you have questions about this document, please contact us via any of the methods (phone, email, ticket, or in person) described here: CISL Customer Support.
© Copyright 2002-2007. University Corporation for Atmospheric Research (UCAR). All Rights Reserved.
© Copyright 2003-2008. University Corporation for Atmospheric Research (UCAR). All Rights Reserved.
Address of this page: http://www.cisl.ucar.edu/docs/access/purpose.html