Designing chips that work

The complexity of integrated circuits continues to grow, and chips with over 10 8 transistors will be in widespread use by the late 1990s. These chips will combine general purpose processors with subsystems for communications and other specialized tasks. They will be far too complex for the design to be tested, and manufacturing volumes will be far too high for the design to be wrong! Mathematical techniques have already been applied to the design of parts of VLSI chips. Most of this work is experimental, and requires an unusual combination of engineering, mathematical and programming skills. Sometimes new theoretical work is needed, and specialized tools may have to be constructed. Despite these difficulties, mathematical techniques are playing an important role in the design of microprocessors at Inmos, and techniques suitable for incorporation in standard computer-aided design systems are emerging.