Hi, Jay: Included are the remaining sections of the draft(NPAC's input). Please modify it as you like, as it is not well elorabated due to the short time and my English. If you think the resume part is not long enough, please let me know. --- Gang 3.2 NPAC's relevant experience Through its technology transfer ACTION program which is funded by New York State, NPAC has being actively engaged in developing and applying core technologies of parallel computing software and computational science into industry applications. Major research and application projects, funded by either federal agencies or private industry partners such as DARPA, NSF, NASA, ONR, ASAS, DEC,INTEL, IBM, nCUBE, Thinking Machines, GE, Niagara Mohawk, TRW, include portable and scalable parallel languages(Fortran90D/HPF), parallel databases, integration software and visualization techniques, financial and economic modeling, power system simulation, weather and climate modeling, four-dimensional data assimilation, environmental modeling, electromagnatics simulation, signal processing and health applications. NPAC has smaller-scale, but state-of-the-art MPP computing facilities suitable for development work (Phase 1&2) and production run(Phase 1): 32-node Connection Machine-5(32MBytes/node, 4 vector-unit/node) with 10 Gbytes of DataVault. 16-node Intel iPSC/860(16MBytes/node, 8KBytes-cache/node) 8-node IBM SP-1(available from September, 1993) In an applications benchmark set for Fortran-D/HPF, a project funded by DARPA, and working closely with Dr. Harrington's group at Syracuse University, NPAC researchers developed a MPP implementation of an electromagnetic scattering application using moment method[14][15]. Data parallel and message passing algorithms of this application were developed on Connection Machine CM2, CM5 and Intel iPSC/860[20]. Using an application visualization system(AVS), an interactive remote visualization environment for this CEM application is further developed to visualize the real-time simulation on a hertergeneous distributed computing system[16][21] in which four workstations and CM5 is used. In the same DARPA-funded project, block parallel algorithms of a LAPACK subset for distributed-memory architectures, including LU-factorization, QR factorization and Cholesky factorization, was developed on CM5 and iPSC/860. Scalability and perofrmance issues were investigated for the parallel linear solvers on those MPP machines. Through our collaboration with CRPC and working closely with other national research partners, we are able to access to the most advanced MPP computing facilities and to share and exchange newest MPP research and application advances with other groups work. For example, NPAC has very close relationships with J. Dongarra's linear algebra group at ORNL, L. Johnsson's CMSSL group at TMC and T. Cwik's CEM group at JPL. Our major federal project is funded by DARPA and through this we have good relations with major DARPA funded work on HPCC libraries, including Illinois and Minnesota efforts. We have excellent relations with all MPP vendors and these will be used as necessary to support the proposal, e.g., to collaborate with TMC, Intel and IBM for access to best solvers on their machines. In addition, NPAC has initiated the Multidisciplinary Analusis and Design Industrial Consortium(MADIC) with efforts to apply high-performance computing technologies to the solution of large-scale, "grand Challenge" problems in the fields of electromagnetics, fluids and structures. Resumes: Geoffrey Fox is an internationally recognized expert in the use of parallel architectures and the development of concurrent algorithms. He is also a leading proponent for the development of computational science as an academic discipline and a scientific method. His research on parallel computing has focused on development and use of MPP technology to solve large-scale computational problems. Fox has expertise in computational physics, parallel linear algebra algorithms, parallel software/hardware systems and MPP programming environments. Fox will supervise NPAC's contribution to this proposal, including issues of parallel software and algorithm development, evaluation of the set of application codes, and collaboration and utilization with other research groups work such as those at ORNL,Thinking Machines, Intel, JPL. Gang Cheng is a Research Scientist for developing scientific and industrial applications of parallel computing at the NPAC. Previous application experience includes developing an electromagnetic scattering simulation on a high performance computing and visualization system, parallel models for financial modeling applications, techniques and methodologies of integrating visualization and MPP in a hetergeneous distributed computing system. Current work related to this proposal involves parallel algorithm design and implementation of the CEM model on proposed MPP systems, evaluation of the existing fortran77 code and resulting MPP software performance. Cheng will be responsible for linking the computer science to the CEM in our NPAC/SRC team to apply parallel software and algorithms to the set of SRC's CEM models. More references: [20] Y. Lu, T. Haupt, G. Fox and R. F. Harrington, "Implementation of Simulation of Electromagnetic Scattering from Slots on Massively Parallel Computers," SCCS Technical Report, April, 1993. [21] G. Cheng, C. Faigle, G. Fox, W. Fumanski, B. Li and K. Mills, "Exploring AVS for HPDC Software Integration: Case Studies Towards Parallel Support for GIS,", AVS Conference AVS'93, Lake Buena Vista, FL, May 24-26, 1993. [22] G. C. Fox, "Parallel Computers and Complex Systems," Complex Systems '92: From Biology to Computation, Innaugural Australian National Conference on Complex Systems, December 1992. Editors: Bossomaier, David G. Green. CRPC-TR92266. [23] A. Choudhary, G.C. Fox, S. Hiranandani, K. Kennedy, C. Koelbel, S. Ranka and J. Saltz, "A Classification of Irregular Loosely Synchronous Problems and Their Support in Scalable Parallel Software Systems," DARPA Software Technology Conference 1992 Proceedings, pp. 138-149, April, 1992. [24] G.C. Fox, "Applications of Parallel Supercomputers: Scientific Results and Computer Science Lessons," 4, 47-90, Natural and Artificial Parallel Computationa, ed., M. Arbib, J.A. Robinson, The MIT Press, Cambridge, MA, 1990.