Decomposition Methods For Library Binding Of Speed-independent Asynchronous Designs

Abstract

We describe methods for decomposing gates within a speed-independent asynchronous design. The decomposition step is an essential part of the library binding process, and is used both to increase the granularity of the design for higher quality mapping and to ensure that the design can be implemented. We present algorithms for simple hazard-free gate decomposition, and show results which indicate that we can decompose most of the gates in our benchmark set by this simple method. We then extend these algorithms to work for those cases in which no simple decomposition exists.