Parallelization of RSA Algorithm Based on Compute Unified Device Architecture

In the domain of computer security, how to enhance the speed of RSA algorithm has been the research hot spot. With the recent tremendous increase in Graphics Processing Unit´s computing capability as a co-processor of the CPU, Nvidia´s Compute Unified Device Architecture (CUDA) can greatly benefit single instruction multiple thread styled, computationally expensive programs. This paper presents a novel parallelized implement of RSA algorithm using JCUDA and Hadoop. Firstly the principle of tradition RSA algorithm is studied. Secondly, the parallel RSA algorithm is designed and realized in CUDA framework. Thirdly, with JCUDA, the RSA parallel algorithm implement function is called by each node in Hadoop cluster. Our experimental results demonstrate the speed of RSA algorithm enhanced dramatically compared to the original method on the CPU only.