Abstract
Two-layer organization of modular arithmetic is considered. Lower layer uses many moduli at hardware precision and simultaneous conversion to/from RNS as described in [2]. Upper layer uses specially selected large moduli allowing for fast reduction and/or reconstruction. Implementation of two different strategies for selecting moduli on the upper layer confirms practicality of proposed approach.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have