BDS-MAJ: A BDD-based logic synthesis tool exploiting majority logic decomposition

Despite the impressive advance of logic synthesis during the past decades, a general methodology capable of efficiently synthesizing both control and datapath logic is still missing. Indeed, while synthesis techniques for random control logic (AND/OR-intensive) are well established, no dominant method for automated synthesis of datapath logic (XOR/MAJ-intensive) has yet emerged. Recently, Binary Decision Diagrams (BDDs) have been adopted to create an optimization system, named BDS, that supports integrated synthesis of both AND/OR- and XOR-intensive functions through functional logic decomposition on the BDD structure. However, it does not support direct decomposition and manipulation of majority logic which, instead, is widely used in datapath circuits. In this paper, we present the first BDD-based majority logic decomposition method and a logic decomposition system, BDS-MAJ, that enables efficient logic synthesis for both random control and datapath circuits. Experimental results show that logic synthesis based on BDSMAJ produces CMOS circuits having on average 28.8% and 26.4% less area and, at the same time, 12.8% and 20.9% smaller delay with respect to academic ABC and BDS synthesis tools. Compared to commercial Synopsys Design Compiler synthesis tool, BDS-MAJ reduces on average the circuit area by 6.0% and decreases the delay by 7.8%.