PDAC - A DATA-PARALLEL ALGORITHM FOR THE PERFORMANCE ANALYSIS OF CLOSED QUEUING-NETWORKS

作     者：HANSON, FB MEI, JD TIER, C XU, HH 

作者机构：Laboratory for Advanced Computing University of Illinois at Chicago 851 S. Morgan M/C 249 Chicago IL 60607-7045 USA 

出 版 物：《PARALLEL COMPUTING》 (并行计算)

年 卷 期：1993年第19卷第12期

页      面：1345-1358页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Dept. of Energy, (DE-FG02-93ER25168) UIC Computing center UIC Software Technology Research Center National Science Foundation, NSF, (DMS-88-06099, DMS-89-22988, DMS-91-02343, DMS-93-00136, DMS-93-01107) Argonne National Laboratory, ANL Los Alamos National Laboratory, LANL 

主　　题：DATA PARALLEL ALGORITHMS CLOSED QUEUING NETWORKS PERFORMANCE ANALYSIS 

摘      要：A parallel distribution analysis by chain algorithm (PDAC) is presented for the performance analysis of closed, multiple class queueing networks. The PDAC algorithm uses data parallel computation of the summation indices needed to compute the joint queue length probabilities. The computational cost of the PDAC algorithm is shown to be of polynomial order with a lower degree than the cost of the serial implementation of the DAC algorithm. Examples are presented comparing the PDAC algorithm with the DAC algorithm to illustrate its advantages and limitations.