NASA Testbeds Budget Code: 509-10,509-20,509-30,234,536
The objective of this activity is to encourage and accelerate U.S. commercial development of high performance computing systems that will support the Grand Challenges. To accomplish this, NASA will fund and encourage the commitment to early acquisition, access, and placement of such advanced systems by: acquiring advanced prototype and early production model high performance computing systems for use and evaluation; providing network access to understand execution of full Grand Challenge applications on TeraFLOPS systems; providing a testbed control environment to assist in the collection of data about testbed operations that are needed to project the eventual performance of Grand Challenge applications on TeraFLOPS computing systems; developing a set of parallel benchmark codes based directly on the Grand Challenge applications to evaluate disparate architectures; and performing research and development activities in ground and flight-based testbeds to be used for demonstration, evaluation, and validation of performance and scalability of both high performance and ultra low power prototypes in a controlled environment for use in spacecraft.

To compare different approaches to TeraFLOPS systems on a common basis, NASA will develop these parallel benchmarks to reflect the computational demands of the various Grand Challenge areas. All benchmarks will be scalable and used on HPCC testbeds.

NASA also will develop tools and techniques to project the execution performance of the Grand Challenge applications. High performance computing research facilities will be established to accelerate transition to new generations of high performance computing technology by enabling researchers to explore applications of these new technologies. This will provide early access to advanced computing facilities for advanced application experiments and facilitate the creation and evaluation of new computing technology. Advanced high performance processors will be integrated into a prototype TeraFLOPS computing facility that will include access to the National Research and Education Network, early systems or advanced prototypes of important storage hierarchy subsystems, and sufficient advanced visualization facilities to allow system scalability experiments to be conducted. This will provide a spectrum of experiments for scalability studies as well as access to the largest possible advanced system by the Grand Challenge researchers.
Budget ($ M)
FY 95 Act 25.12
FY 96 Pres 14.81
FY 96 Est 12.91
FY 97 Rqst 15.39
Program Component Areas
  FY 96 FY 97
HECC 12.91 15.39
LSN    
HCS    
HuCS    
ETHR    
Agency Ties
DARPA Partner
NSF Partner
DOE User
NASA User
NIH  
NSA  
NIST  
NOAA  
EPA  
ED  
AHCPR  
VA  
Milestone Changes  
FY 1995 Actual Milestones FY 1996 Estimated Milestones FY 1997 Agency Requested Milestones
Grand Challenge Investigator Teams were provided access to the following testbeds for evaluation of applications: 128 node CRAY T3D at JPL, 16,384 node MasPar MP-2 and 8 processor Convex SPP-1 at NASA Goddard, 528 node Intel Delta at Caltech, Thinking Machines CM-5 scalable machines at NASA Ames and Naval Research Lab, Kendall Square Research scalable system at University of Washington, IBM SP-1s at NASA Lewis and Argonne National Lab, and IBM SP-2 at NASA Ames.

A new SGI Power Challenge L Cluster Machine was installed at NASA Ames.

NASA conducted a workshop on "Future Miniature Spacecraft Technologies" which introduced REE to potential customers, and validated the REE concept and vision. This led to the formulation of a detailed Implementation Plan in FY 96. 		To accommodate for diminishing outyear budgets, NASA exploited the existing infrastructure of design workstations, creating a seamless uniform computing environment to integrate all resources.

During FY 96, NASA will also:
  • Make testbeds available to investigators for applications development and porting, and for system software development and evaluation.
  • Demonstrate interoperation of the network infrastructure at NASA Goddard and JPL with NREN at 622 Mbps.
  • Demonstrate portability and scalability of software components and tools to TeraFLOPS systems.
  • Select and award a contract for development and validation of a high performance flight prototype
  • Install a third generation 50-100 GigaFLOPS sustained scalable ESS testbed.
  • Demonstrate cost effective high performance computing at performance and reliability levels equivalent to 1994 Vector Supercomputers at 25% of the capital cost.
NASA's ESS 50-100 GigaFLOPS/s sustained scalable testbed will support the achievement of 50 GigaFLOPS/s sustained on ESS Investigator codes.

NASA will provide a production systems software environment that integrates distributed workstations with scalable TeraFLOPS machines.