Performance-Driven Clustering of Asynchronous Circuits

This paper describes a novel approach for generating asynchronous circuits from HDL specifications by clustering the synthesized gates into asynchronous pipeline stages while preserving liveness, meeting throughput constraints, and minimizing area. The method enables a form of automatic re-pipelining in which the throughput of the resulting design is not limited to the clock frequency or the level of pipelining in the original RTL specification. The method is design-style agnostic and is thus applicable to many asynchronous design styles.