Visual Software Development for Parallel Machines

PSI's CAD approach to simulation/software development has cut large scale life cycle costs by an order of magnitude. This approach, based upon significant departures in technical concepts, provides ease of control and reuse of complex software modules. It has broken barriers to building complex systems, making significant upgrades easy to evolve, and cutting support costs dramatically. This is done by achieving module independence through graphically enforced architectural design rules and a fully automated user-friendly environment. This success has resulted in huge simulations that meet customer validity constraints, but now exceed single processor computer power by one to two orders of magnitude. Hardware designers have produced parallel and distributed processor computers with theoretical speeds well into gigaflop ranges. However, their practical use on all but very special problems has been extremely limited due to complex software interfaces they present. The proposed research is to confirm the hypothesis that the same graphical design rules that achieve software understandability and independence also support automatic allocation of independent processes to highly parallel processors. Fusion of this CAD approach to software with parallel processors can result in multiple orders of magnitude of speed improvement while making development easier on parallel processor machines versus single processor machines, due to concurrent memory access and management. Communication network equipment design, manufacturing plant design, process control system design, local, national, and international econometric modeling, business market modeling, world-wide commodity and currency stock and flow predictions, etc