The purpose of this project is to develop tools and techniques for performing computer assisted analyses of parallel programs and to use these analyses to gain insight into the design of optimally efficient parallel programs. The primary focus of the project is on the microanalysis of programs executing on families of massively parallel MIMD and SIMD computers (e.g., multi-dimensional mesh connected computers) with special attention given to frequently-used, automatically analyzable classes of programs (e.g., divide and conquer). The research has both theoretical and experimental components.
