基于SIMD结构的矩形行列式并行算法研究  

Research of parallel SIMD-based algorithm for matrix determinant

作　　者：王艾昕[1] 

机构地区：[1]辽宁科技学院工程实践中心,辽宁本溪117004

出　　处：《计算机工程与应用》2012年第25期48-51,56,共5页Computer Engineering and Applications

摘　　要：在运用行列式Schur余子式算法的理论基础上,提出了对SIMD结构的并行机,可适用于对行列式按行分块并行处理算法,把一个n阶行列式的求值过程分解成相对独立的若干个二阶行列式的求值过程,而且它们的求值过程是相对独立的,具有并行性,从而设计出n阶行列式求值的并行算法。给出了该算法的实现步骤,分析了算法的加速比;对算法进行了模拟实验,结果说明了其性能。On the theoretical basis of applying determinant＇ s Schur algorithm, this paper puts forward the applica- ble parallel algorithm which divides pieces by the line on the SIMD-based computer. It disassembles one n-factorial determinant into some 2-factorial determinants which are independent relatively. The evaluating process of these 2-factorial determinants has parallelism. In addition, the paper offers the steps of algorithm and analyzes its accelerating ratio. Then it designs a series of experiments to explain the capability of the algorithm.

关 键 词：单指令多数据流 并行计算机 并行算法 加速比 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]