Abstract
The implementation of residue number system (RNS) adders based on binary adders is described. These adders use two cycles of addition and support any class of modulus. A technique for choosing the correct sum in a two-cycle residue addition is presented and proved correct. Three VLSI layout approaches for residue adders are described and performance figures for area and speed are given. The two approaches using one binary adder offer savings of about 30% in area and significant improvement in speed/area product over the approach using two binary adders.

This publication has 5 references indexed in Scilit: