FPGA based Faster Implementation of MAC Unit in Residual Number System

摘要

In this fast-growing world, where everyone is in hurry, speed has become a critical factor even in the electronics world. It’s not the answer but a fast answer is the need of the time. Artificial intelligence has touched almost all aspects of our lives. But the software implementation of machine learning algorithms has not been able to meet the expectations of solutions in nanoseconds, especially where neural networks with extensive calculations and arithmetic computations are involved. Multiplication and accumulation unit is an integral part of many signal processing and machine learning algorithms. Here we implement an arithmetic module based on distributed arithmetic imbibed with Residual Number System to prove the efficacy of hardware design over software implementation. The purpose is to exploit the parallelism property of FPGAs to accelerate the computations. The target device used is xc6vlx75t3ff484 from Virtex- 6 family in Xilinx. Simulations are tested in MATLAB and ModelSim with associated data to justify its feasibility.