Optimization of gate-level area in high throughput Multiple Constant Multiplications

Abstract

This paper addresses the problem of optimizing gate-level area in a pipelined Multiple Constant Multiplications (MCM) operation and introduces a high-level synthesis algorithm, called HCUB-DC+ILP. In the HCUB-DC+ILP algorithm, initially, a solution with the fewest number of operations under a minimum delay constraint is found by the Hcub-DC algorithm. Then, the area around this local minimum point is explored exactly using a 0–1 Integer Linear Programming (ILP) technique that considers the gate-level implementation of the pipelined MCM operation. The experimental results at both high-level and gate-level clearly show the efficiency of HCUB-DC+ILP over previously proposed prominent MCM algorithms.