Development of neural network tools for rapid computer system design

Methodologies are developed for improving design quality and the speed of the design process. The authors present the portion of the work which involves the application and extension of neural networks to solve optimization problems that arise in this design process. Specifically, they discuss the use of neural networks to solve the scheduling and allocation problems that arise when assigning operations to control steps and functional units. The goal is to minimize the number of control steps, number of each type of functional unit, allow multi-step operations, and handle bussing constraints. The application of neural networks to high-level computer design has the potential to provide more flexible, advanced computer system design tools.<
>