CAD Tools for Third Generation Custom VLSI Design

In this contribution the future of CAD tools for custom design of system VLSI chips will be discussed. In 7 propositions and 7 problem statements it will be shown that a complete separation of system and silicon design is necessary, whereby silicon reusability is of prime concern in a continuously evolving technology environment. Knowledge based synthesis techniques implemented on expert systems will substitute for much of the simulation and verification presently dominating manual design. Silicon module layout and assembly as wall as system design will evolve into programming techniques in an advanced software environment. This will be based on very high level languages such as LISP, PROLOG and SMALLTALK in combination with more traditional algorithmic oriented languages such as C. Workstations with hardware dedicated to this environment will replace the actual emerging 32 bit UNIX-C oriented workstations within a time span of 4 years... Design of CAD tools will then be indistinguisible from the chip design itself... Both will be exclusively programming activities. Some of the major bottlenecks on the road towards this era are : lack of good silicon compilers for analog circuit design, acceptability of new software environments by the designer community, knowledge acquisition for VLSI, portability of module generator environments, chip performance optimization and standardization. All of the latter present a challenge to Europe which, by tradition, has a lot of experience in advanced software. Will it use these capabilities by merging software and VLSI design knowledge ?