A linear program driven scheduling and allocation method followed by an interconnect optimization algorithm

Abstract

A new method for high level synthesis is reported whose basic feature is the tight interaction and coupling of the scheduling and allocation phases providing a global direction to synthesis. A linear program based allocation is proposed which uses multifunction ALU cost estimation, and iteratively drives a tree search for scheduling. A major contribution of this paper is a new interconnect optimization algorithm which is based on several interconnect transformations for multiplexer input collapsing and merging. Several other important synthesis aspects are included, e.g. register and interconnect bindings, operation chaining and operation multicycling. The method has been implemented in C on a Sun 3/60,