并行计算在矩阵运算中的应用

The Application of Parallel Computing in Matrix Operation

随着时代的发展,现在计算机的CPU中能够集成多个内核,使计算机在处理数据时能够多个核心、多个线程对数据进行处理,这也意味着需要新的适应多核的并行计算程序来匹配软件与硬件,最大化发挥硬件的性能。笔者以矩阵的乘法计算为研究对象,用API、Open MP、MPI和PPL四种多核并行算法,比较分析出了适合科学计算研究的并行算法。

With the development of the times,now the computer CPU can integrate multiple cores,and enable the computer to a plurality of core and multiple threads for data processing in data processing.This also means the need for new multi-core parallel computing programs to match software and hardware,and maximize the performance of the hardware.The author takes matrix multiplication calculation as the object of study,and uses API,OpenMP,MPI and PPL four multi-core parallel algorithms to compare and analyze parallel algorithms which are suitable for scientific computing research.