Citation[]
High-End Computing Revitalization Task Force (HETRTF), Federal Plan for High-End Computing (May 10, 2004) (full-text).
Overview[]
This Plan offers a vision for a proactive Federal effort that advances high-end computing technology to address many of society’s most challenging large-scale computational problems and, in doing so, strengthens the nation’s global leadership in the sciences, engineering, and technology. The Task Force focused its scope on technology directly needed for high-end computing. Consequently, a number of important technological components essential to science and engineering, such as visualization, networking, grid computing, security, and applications-specific software were not considered by this study. The Plan has three primary components: HEC Research and Development (R&D), HEC Resources, and Procurement.
- HEC Research and Development: The Task Force recommends first and foremost a coordinated, sustained research program over 10-15 years to overcome major technology barriers that limit effective use of high-end computer systems. Today, poor system reliability, the increasing cost and risk of software development, and architectural features (such as the growing imbalance between processor and memory performance) all greatly restrict the ability to achieve high levels of performance for science, engineering, and national security applications. To address these barriers, the HECRTF Plan outlines a comprehensive technology strategy involving basic research, advanced development, engineering and prototype development, and test and evaluation. The Plan presents technology roadmaps for hardware, software, and systems, comparing the current program to a robust revitalization plan. The outcome of the HECRTF [[R&D] Plan will be a robust diversity of tools, technologies, and systems that minimize time to solution for the most challenging computational problems. The end result will be a secure leadership position in high-end computing, and the scientific and technological advances such a position enables, for decades to come.
- HEC Resources: Providing high-end computing resources across the full scope of critical Federal missions raises three major issues. First, some agencies have a science and technology mission but lack access to high-end computers. Second, high-end computing has been so successful in contributing to research in science and technology that current resources to meet overall Federal science and engineering demands are heavily oversubscribed. Third, Federal HEC resources do not include systems powerful enough to solve many important large-scale problems. The Task Force provides an interagency collaborative strategy to address all three of these issues.
- Procurement: The HECRTF Plan proposes several pilot projects for improving the efficiency of Federal procurement processes, benefiting both government and industry. These pilot projects involve benchmarking (i.e., using software to measure the performance of systems), development of models for total cost of ownership, and approaches to sharing procurement processes across agencies. The intent of the pilot projects is to build teams that span agencies and increase visibility on issues critical to HEC procurement. The Task Force expects that these projects will improve the information flow to assist in the prioritization of future HEC research, development, and engineering investments. Moreover, coordinated procurement of HEC resources will provide more leverage in working with industry vendors to address the needs of the HEC applications communities.
The Plan proposes alternative approaches and planning strategies to carry out these activities. The Task Force analyzes the likely outcomes five years out in the absence of a revitalization effort. The current program allows for some evolutionary advances in high-end computing. However, the Task Force believes that, even with management improvements, the current program will neither maintain U.S. leadership in the face of serious competition nor keep pace with the accelerating growth of demand for high-end computing resources to meet Federal agency needs.