Multiple core algorithms for residue number systems

The core function provides an easily implemented and efficient means for performing the traditionally difficult residue operations. However, two problems with core algorithms are the difficulty in choosing core coefficients and the many iterations required for some operations. A method for the design of core functions which are as linear as possible for a small dynamic range is given. This method can then be used in conjunction with a redundant core calculation to avoid the problem of critical cores. Using these results, a methodology to design multiple core algorithms for sign detection and integer division is given. From these two algorithms, the operations of magnitude comparison, overflow detection, and approximate decoding follow directly