Automated Synthesis of Speed-Independent Circuits using Distributed Finite State Machines

Abstract

Asynchronous circuits are a promising, yet intricate, type of digital circuit that offers higher performance and lower power consumption than their synchronous coun terpart. However, asynchronous circuits still see limited usage in today’s commercial products, which often is linked to the adaptation challenges that are posed by in dustry, e.g. the time required for developing new tools and training circuit designers versus existing synchronous-based tools for a faster production line. Several formal models were introduced to aid with asynchronous circuit design. In particular, the ‘legacy’ approach of Burst-Mode (BM) Specifications and the ‘disrup tive’ approach of Signal Transition Graphs (STGs). On one hand, BM specifications resemble synchronous Finite State Machines (FSMs) allowing circuit designers to eas ily adapt and use them, but there is no longer support provided for their tools. On the other hand, STGs have access to state-of-the-art tools that produce well-optimised circuits, yet they are seen as too complicated compared to FSMs. In this thesis, a new model called Burst Automaton (BA) is proposed. BA is a generic FSM-based model that acts as a framework for enabling interoperability between many models including BM specifications and STGs. BA offers a new de sign path that bridges the gap between ‘legacy’ and ‘disruptive’ approaches, granting circuit designers access to state-of-the-art tools for higher quality implementations without costing their familiarity with FSMs. Thus, removing any adaptation require ments. This design path is implemented as a new Workcraft plugin that supports the design automation of BAs, and is evaluated on some case studies.