面向众核GPU结构的椭圆曲线加密流化技术  

Streaming Elliptic Curve Cryptography for Many-core GPU

作　　者：甘新标[1] 沈立[1] 王志英[1] 

机构地区：[1]国防科学技术大学计算机学院,湖南长沙410073

出　　处：《四川大学学报（工程科学版）》2011年第2期98-102,共5页Journal of Sichuan University (Engineering Science Edition)

基　　金：国家"973"计划资助项目(2007CB310901);国家自然科学基金资助项目(60803041);国防科技大学优秀研究生创新资助项目(B090603)

摘　　要：针对椭圆曲线加密系统的加密速度不能满足实时性要求的现状,分析了椭圆曲线加密机制(Elliptic Curve Cryptography,ECC)潜在的并行性,研究了GPU众核多线程及共享存储空间和只读Cache对ECC性能提升的影响,提出了以GPU强大计算资源和有效的存储带宽来流化加速ECC的方法。实验结果显示,在支持计算统一设备架构(Computing Unified Device Architecture,CUDA)的GPU上流化实现的ECC原型系统与优化的CPU实现相比可获得高达66×的加速度比。另外,针对ECC的流化并行及优化技术可作为一般方法推广至其它流体系结构。In order to overcome the weakness in real-time ECC（Elliptic Curve Cryptograph）,streaming ECC was proposed,including exploiting the potential parallelism of ECC for streaming and parallelizing it using thousands of threads as well as stream optimization based on shared memory and texture cache or constant cache with best utilization of memory hierarchy of CUDA-enable GPU.Experimental results showed that the prototype of ECC implemented using CUDA（Computing Unified Device Architecture） on NVIDIA＇s GTX280 could achieve as high as 66×speedup than CPU counterpart available.Furthermore,the proposed techniques including streaming for parallelization and optimizations with memory hierarchy could be generalized for other streaming architectures.

关 键 词：GPU 椭圆曲线加密 流化并行 优化 计算统一设备架构 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]