Funds to support activities in High End Computing and Computation for FY 1997 total approximately $454 million, compared to the FY 1996 Estimated budget of $451 million, an increase of 0.6%.
Effective in FY 1997, NSF has discontinued reporting as HPCC Program activities approximately $9 million in the Supercomputer centers' budgets and approximately $16 million in activities in theoretical computer science--these actions were recommended in the National Research Council's Computer Science and Telecommunications Board report reviewing the HPCC Program. However, new programs in Computing Systems make up for this decrease, emphasizing fundamental research in the design and implementation of computing systems. NSF will initiate the Partnerships for Advanced Computational Infrastructure (PACI) to provide access to high performance computing for the academic research community at a level of one to two orders of magnitude greater than that typically available at a major research university. NSF will begin a new, interdisciplinary initiative to apply massively parallel computing using biological materials such as DNA. NSF plans to support research to determine how formal, automated design techniques can be used to enable rapid advances in design for nanomanufacturing.
In DOE, new activities, such as the new Advanced Computational Testing and Simulation (ACTS) project, are designed to provide simulations of difficult (even impossible) or dangerous experimental phenomena. DOE will re-compete both the HPCRC and the Grand Challenges projects for a second phase. They will begin development of software tools that are designed around the ACTS Toolkit interface definitions and integrate most critical existing software tools into the ACTS framework. DOE plans to deliver high performance data management and archival storage software to users of DOE high performance computer centers by integrating file systems of supercomputers in access centers using the High Performance Storage System. DOE will also begin development of ACTS-enabled applications development in compressible fluid dynamics, combustion, environmental chemistry, and materials sciences. This effort is coordinated with parallel development in DOE Defense Programs of ACTS-enabled applications in weapons hydrodynamics and materials characterization and aging.
DARPA is developing new operating systems capabilities that include object management, high assurance, and support for real time distributed applications. They are continuing their high performance language and compiler research and developing advanced software environments. DARPA will provide scalable versions of widely used commercial engineering software, including MSC NASTRAN, leveraging scalable software library technology available to the defense community.
NASA plans to support long-term HECC projects in high performance computing systems research and high performance systems software and technology, in coordination with DARPA, NSF, and DOE. NASA's Earth and Space Sciences (ESS) scalable testbed will support the achievement of 50 gigaFLOPS sustained on ESS investigator codes. NASA is also interested in an end-to-end reduction in cost and time to solution for aerospace design applications on distributed workstation clusters at 25% of the capital cost of a 1994 comparison base. In support of this effort, NASA will develop commercially available systems software that provides high availability and portability demonstrated in a large-scale production environment.
NIH will continue use of high performance computer resources accessible through high speed networks to generate biomedical images, support structure-based drug design, and perform large-scale computer simulations to elucidate the structure and function of biological macromolecules including assemblies of macromolecules such as cell membranes. The intent of the NIH program is to expand activities in high performance computing applications to determine and predict the structure of proteins, apply computational chemistry to understanding of drug interactions with enzymes and nucleic acids as a basis for discovery of new drugs, and develop new positron emission tomography imaging methods.
NSA will begin a new initiative in materials, devices, and optoelectronics for very high communication rates and computing. They will continue with the design of a uniform memory access supercomputer without a backplane, seeking to improve processor-to-memory latency and bandwidth by an order of magnitude. They will continue development of a superconducting crossbar switch and associated electronics.
NIST will stabilize and release a scalable parallel multigrid solver for unstructured adaptive grids and demonstrate high-performance object-oriented libraries for sparse linear algebra on challenging applications.
NOAA plans to implement its ETA regional model ("ETA" refers to the type of vertical coordinate used in the model) on one of its massively parallel systems. They will evaluate the model's performance at several grid resolutions and assess its potential for use in operational forecasting and for updating real-time experimental forecast models to support nesting. NOAA's National centers for Environmental Prediction (formerly the National Meteorological center) plan to use scalable high performance computing systems to enable future improvements in operational weather forecasting and to initiate algorithm development on scalable systems with the goal of achieving 5-10 kilometer resolution in mesoscale atmospheric models.
 |
 |