Research On An Implementation of Fault-Tolerant Processor Interconnected Networks

点击翻译按钮获取中文摘要

In this rsearch, we developed algorithms to realize an arbitrary BPC permutation in chrdal ring networks and hypercube connected networks, and evalated the number of required steps to realize some typical BPC permutations in 3 kinds f networks : chordal ring networks, hypercube cnnected netwoks, and omega networks.A developmet of an algorithm to realize an arbitrary BPC permutatin is of value for the estimation of routing steps required in a network of an arbitrary size, that is for the estimation of the upper bound of required steps in a generalized Chordal ring network. On the other hand, the numbers of required steps in both hypercube cnnection and omega networks are almost real the minimum values. Some BPC routing algorithms for cube or mesh connected networks has been reported, and each link in these networks was assumed to be bidirectional simultaneously. The results presented in Chordal ring and hypercube connected networks in this research are for unidirectional communications simultaneously. For omega or delta networks, first, how differnt are the comlexities required when using a route decision algorithm and an improved algorithm having taken some inherent properties into consideration is discussed and solved by obtaining rea data. Next, hOW many passes are required to realize BPC permutations such as Bit-Reversal, Matrix-Transpose, Perfect-Shuffe, and Bit-Compement permutations n delta networks when when faults are present and when not present.