Session 6 keynote address

Abstract

The designer's ability to create new compact memory storage cells and dense layouts had served as a goal of the industry until recently, when new challenges began to appear. Today, the focus is on narrower market segments, where the profit margins can be significant, without requiring the lowest manufacturing costs. Up graded circuits in this new design era include those used in telecommunications, analog/digital interfaces and other application specific functions. Designer skills needed today are dramatically different today and will continue to change with changing technologies. The value of the chip is now in the implemented algorithms and architectures, and on the time to market. The answer to this need has been to use such techniques, as gate arrays and standard cell designs: two approaches that almost eliminate IC designers from the design process. The traditional designer's contribution must evolve into one that is primarily centered on determining new applications for ICs and new architectures that will make these applications possible. In circuits of 10's of thousands of transistors or larger, it is no longer rational to perform significant optimization at the transistor level, because architectural decisions can have considerably more impact. We have seen this in performance achieved by RISC microprocessors, and by recently developed image processing circuits performing hundreds of millions of operations/sec. The problem of enormous design complexity has frequently been attacked by the inefficient application of more manpower, rather than by use of efficient software tools. These tools should not replace the designer, rather they should relieve him of documentation and other mundane tasks. He can contribute and introduce innovations at the applications and architectural levels. However, the designer must be comfortable with software development, because only he will know what tools are required for unburdening him of the details, while leaving him in control of the design process. Enter - a new methodology- the reorganization of the traditional centralized CAD group and separate design groups into a more homogeneous mix of designers and tool developers. Relieving the designer of low-level tasks, permitting him to focus on identifying new applications, will require a greater systems level of orientation and the increasing use of diversified talent. The IC designer will become more of a facilitator between those who have applications and those who provide the circuit implementation. Also, the integration of analog circuit functions is becoming increasingly critical for many applications such as telecommunications. The design style used is now in the process of undergoing a major upheaval. The traditional techniques based on precision components are being replaced by techniques that are more compatible with the integration of systems on a chip. In summary, IC design of the future must be carefully reassessed.