众核计算平台的高吞吐率密码算法加速  

Accelerating high throughput cryptography algorithm on many-core computing platforms

作　　者：符鹤[1] 李春江[2] 王昊 谢永芳[1] FU He;LI Chun-jiang;WANG Hao;XIE Yong-fang(School of Information Science and Engineering,Central South University,Changsha 410083;College of Computer,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083 [2]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2018年第4期580-586,共7页Computer Engineering & Science

基　　金：国家自然科学基金(61321003;61170046)

摘　　要：众核处理器适应于加速高吞吐率的计算密集型应用,而密码算法需要进行大量的数学计算,特别需要使用高吞吐率的计算平台。提出了一种面向众核平台的粗粒度并行加速框架,该框架不考虑算法内部的运算过程,将数据以计算函数为单位分配到众核协处理器上执行。使用MIC众核协处理器,采用三级并行结构及任务分配机制,提升了高吞吐率密码算法处理的并行性。针对多种密码算法应用的实验结果表明,该框架可充分利用众核平台实现粗粒度并行的高吞吐率加解密处理。Many-core processors are suitable for accelerating high-throughput computing-intensive applications,while cryptography algorithms require large amounts of mathematical calculations and hence require high-throughput computing platforms.This paper proposes a coarse-grained parallel acceleration framework for many-core computing platforms.The framework does not take into account the internal computing process of the algorithm,and allocates the data to the many-core coprocessor for execution in units of computational functions.Based on the MIC many-core accelerator,the framework adopts the three-level parallel structure and task allocation mechanism,and develops the parallelism of the high-throughput cryptography algorithm.Experimental results for a variety of algorithms show that the framework can make full use of many-core computing platforms to achieve high-throughput encryption and decryption processing coarse-grained parallelism.

关 键 词：众核平台 密码算法 高吞吐率 粗粒度并行 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]