基于LARP BS模型的最大值查找算法

The Maximuim Finding Algorithms Based on LARPBS

具备可重配置流水线总线的线性阵列LARPBS(linear arrays with a reconfigurable pipelined bus systems)是近来出现的一种高效的并行计算模型,与理想的PRAM模型不同,LARPBS是现实可行的。基于LARPBS模型,Y.Pan介绍了2种宽度和精度任意的数据项的最大值查找算法:算法1使用了N^2/2个处理机、O(1)时间,它是目前时间最优的算法;算法2使用了N个处理机、O(loglogN)时间。本文介绍了2种最大值查找算法,时间复杂度同Y.Pan的算法,但所用处理机数减少了一半,这是对Y.Pan算法的重要改进。

Linear arrays with a reconfijgurable pipelined bus systems (LARPBS) is newly proposed as a parallel computational model, which processors Eare connected by a reconfigurable optical bus- In contrast with theoretical PRAM model,LARPBS is practical and (efficient. In this paper,we present two algorithms for finding the maximum among N data elements with unbounded magnitude and precision on LARPBS. The first algorithm uses N2/2 processors and O(1)time. The second ome uses N processors and (XloglpgN)time. The processors occurpied by the proposed algorithms are half of Y. Pan's algorithms in the same time complexity.