The HPCCPT-1 Cooperative Research Announcement

Objective: Highly parallel computer technology is very rapidly changing at the present time. The acquisition of a highly parallel computer system for use in NASA-sponsored research has thus presented a significant challenge: How can state-of-the-art systems be acquired and utilized in an effective, timely fashion?

Approach: With the assistance of NASA headquarters, a new procurement vehicle was devised: a ''cooperative research agreement''(CRA). This combines the placement of high performance computing equipment together with a solicitation for applied research to be performed on the equipment. Prospective proposers were invited to form consortia consisting of HPC vendors, aerospace companies, universities, government laboratories, and independent software vendors, all with the mission of advancing U.S. leadership in high performance computing. NASA does not acquire computers under this vehicle ╤ such equipment is to be placed by the proposer in NASA facilities and removed at the end of the performance period.

Accomplishment: The request for proposals for the HPCC Testbed-1 solicitation was released in July 1993, with a due date of December 1. The Technical Review Committee, consisting of scientists from all three NASA research centers, completed their evaluations of the proposals in late December. Based on these findings, semi-finalists were selected, and on-site performance evaluations were conducted with them in early January. In late January, two finalists were selected, and negotiations were conducted with them until mid-April. The selection of a consortium headed by IBM Corp. was announced from Wesley Harris' office at NASA headquarters. The IBM proposal includes three SP-2 parallel computer systems, plus a research plan that includes Boeing, Lockheed, RPI, Rice University, Intelligent Aerodynamics, and Centric Engineering. The 160-node SP-2 that is now being installed at NASA Ames is expected to out-perform the Cray C-90 supercomputer now at NAS, at considerably less cost.

Significance: The HPCCPT-1 solicitation was the first time a CRA has been used at NASA. It was conducted on schedule until the negotiations phase, which took considerably longer than anticipated, partly because a draft agreement document was not available ahead of time for review. In spite of these difficulties, however, the acquisition was completed from start to finish in less than one year. This experience has paved the way for future solicitations of this form.

Future Plans: The Cooperative Agreement is now in place. The first two 64-node subsystems of the Ames system have been delivered and have passed their acceptance test. An additional 32 nodes will be delivered before the end of the year. Langley's 48-node system has been delivered and is now in acceptance test. The research program has commenced, and some initial results should be available soon.

Point of Contact:


Return to the Table of Contents



curator: Larry Picha