A Hybrid CPU/GPU Cluster for Encryption and Decryption of Large Amounts of Data
Ewa Niewiadomska‐Szynkiewicz, Michał Marks, Jarosław Jantura, Mikołaj Podbielski · Journal of Telecommunications and Information Technology · 2012
The main advantage of a distributed computing system over standalone computer is an ability to share the workload between cores, processors and computers. In our paper we present a hybrid cluster system – a novel computing architecture with multi-core CPUs working together with many-core GPUs. It integrates two types of CPU, i.e., Intel and AMD processor with advanced graphics processing units, adequately, Nvidia Tesla and AMD FirePro (formerly ATI). Our CPU/GPU cluster is dedicated to perform massive parallel computations which is a common approach in cryptanalysis and cryptography. The efficiency of parallel implementations of selected data encryption and decryption algorithms are presented to illustrate the performance of our system.