Issues in the Design and Implementation of a Distributed Design System for Microelectronics

Professor Richard Newton
Dept. of Electrical Engineering and Computer Science
UC Berkeley

Thursday, March 6th, 1997
Hogan Room, 531 Cory Hall
5:00-6:00 p.m.


Since the late 1970's, finding the best way to integrate users, tools, and services to form an integrated environment for electronics system design, verification, and transfer to manufacturing, has been a challenging and evolving problem. As in many engineering disciplines, the underlying assumptions that define an appropriate architecture are always changing. Pre-Unix, integration was accomplished primarily via "footnet." During the 1970's, Unix facilities like YACC and LEX were exploited to build translator networks to couple what became known as "tools" into complex spiderwebs of pipes and file formats. By the early 1980's, shared in-memory data structures and bit-mapped user interfaces began to find their way into engineering design system architectures, while over the last decade, as the network has become an integrated part of the computing environment, an integration architecture that exploits RPC, object-oriented design concepts, and fine-grain transactions has become the norm.

In "tomorrow"s world of the Internet, 3D graphics support, virtual environments and virtual communities, what are the appropriate matches of architecture and technology that best-optimize the quality of the user environment, the user's ability to collaborate effectively, and handle the design of complexity of a modern electronic system design problem? In addition, to what extent can we design a system that can evolve and scale with the design community and changes in the underlying technology constraints?

These and related issues will be addressed in this presentation, where a number of the "invariants" over many generations of EDA integration issues will be identified, and glimpses of what might eventually form a kind of "Zen" of complex engineering software design.