INSIGHTS April 1998 Volume 5

Elements of the High-Performance Computing and Communications Program:
	Computational Aerosciences (CAS) Accelerates the availability of high-performance computing hardware and software to the U.S. aerospace community for its use in design processes.
	Remote Exploration and Experimentation (REE) Extends supercomputing capabilities developed by industry into routine use in outer space, reducing the mass, size and power consumption of computers used in space.
	Earth and Space Sciences (ESS) Builds an assortment of computer simulated models that combine complex Earth and space science disciplines.
	Learning Technologies Project (LT) Offers NASA science and engineering to the educational community across the Internet.
	NASA Research and Education Network (NREN) Extends U.S. Technological leadership in computer communications by research and development advancing leading-edge networking technology and services.

From the program manager

One of the many facets of NASA's High-Performance Computing and Communications Program is researching the efficient use of current high-performance computers, as well as exploring the future in extreme high-end computing.

As described in this issue's lead story, NASA's OVERFLOW code provides a software tool that is used by government and industry users to simulate the air flow surrounding complex aerodynamic shapes. OVERFLOW, a respected computational fluid dynamics code, was originally developed on vector processors. It has evolved and adapted to run on a variety of computer architectures, including multiple processors with distributed memory.

There is, however, a pressing need to begin investigating computing requirements beyond our current focus, the realm of teraflops, a trillion floating-point operations per second. NASA supports far-sighted individuals in academia, industry and government such as Thomas Sterling, the author of In pursuit of a quadrillion operations per second, who is working to push out performance boundaries. Achieving one million billion floating-point operations per second (petaflops) has value for many complex scientific problems such as building a single global model that encompasses the heliosphere. Heliosphere model examines the long arm of the sun shows the complexity of these problems that require orders-of-magnitude increases in computing and memory.

Many applications of the future will exchange data at unprecendented rates. The NASA Research and Education Network (NREN), NASA's high-performance wide area network, addresses these challenges. Even mountains of data can be moved over networks, as we show in High-performance networks provide access to NASA's data treasure-trove.

These applications are why HPCC honors the ideal and common goal of achieving distributed heterogeneous computational resources. This is the spirit behind the Information Power Grid which will support the design process of the future.

Bill Feiereisen