论文部分内容阅读
随着微处理器的出现,由多至2~(14)到2~(16)个微处理器组成的大型并行处理系统已经有实现的可能。一种多处理机结构——SIMD(单指令流多数据流)并行处理方式——特别适合于发挥某些任务中固有并行性的作用,SIMD系统目前正用来做诸如矩阵计算和图象处理等科学计算。将来,其范围有可能扩大到商用计算。商用计算要求对许多不同的数组执行相同的程序,例如进行银行来往帐目的利息计算。
With the advent of microprocessors, large parallel processing systems consisting of up to 2-14 (14) to 2-16 microprocessors have been implemented. A multiprocessor architecture, SIMD (Single Instruction Flow Multiple Data Stream) parallel processing, is particularly well suited for exploiting the inherent parallelism in certain tasks that SIMD systems are currently using for applications such as matrix computation and image processing Such as scientific computing. In the future, its scope is likely to expand to commercial computing. Commercial calculations require the same procedure to be performed on many different arrays, such as interest calculations for bank accounts.