Quantitative design of a scalable microsystem using ALMA: the example of the dictionary machine

Abstract

A quantitative method for designing application specific integrated (sub)systems is presented. The method uses a general-purpose silicon assembler-ALMA (Approche d'un Langage de Micro-Architecture). We illustrate this approach by the example of a dictionary machine that consists of n processing elements working in parallel. An analysis of the effect of several generic parameters on the performance and the size of the silicon parts exemplifies the quantitative approach to cost/performance tradeoffs that is offered by ALMA.<>