The High-Speed Implementation Of Information Security Algorithms Based On GPU

In recent years, the need of information security has became increasingly urgent with the rapidly growing popularity of high-speed network. However, the information technology security keeps at its low level, the software implementation of information security algorithms became the bottleneck of network performance. At the same time, Graphics Processing Unit also has been greatly developed, changed from the initial graphics card, which limited to graphics rendering, into present programmable parallel computing platform. Different from serial computing model of CPU, GPU is a highly parallel stream processor. So, it will be a very valuable and helpful research to solve the speed restraint of information security and improve implementing efficiency by programming on GPU to explore fully its high calculation capability.Six typical encryption algorithms and their implementation processes will be firstly introduced in this paper, followed by GPU programming model that pay much importance on unified shader architecture which proposed by DirectX 10, and then the CUDA(Compute Unified Device Architecture)programming model and memory model, to transfer encryption algorithm into the GPU.The key point of this paper focus on a new method—GPU-based encryption in parallel way.In the end of the paper, five GPU encryption algorithms' results will be given. From experiments, you can see the implementation on GPU is much faster than that on CPU, and much more potential in performance.