Institute of Electrical & Electronic Engineers
Customizable data formats provide an opportunity for exploring trade-offs in accuracy and performance of reconfigurable systems. This paper introduces a novel methodology for mixed-precision comparison, which improves comparison performance by using reduced-precision datapaths while maintaining accuracy by using high-precision datapaths. The authors' methodology adopts reduced-precision data-paths for preliminary comparison and high-precision data-paths when the accuracy for preliminary comparison is insufficient. They develop an analytical model for performance estimation of the proposed mixed-precision methodology. Optimization based on integer linear programming is employed for determining the optimal precision and resource allocation for each of the datapaths.