Distributed algorithms for global structuring*

摘要

In the search for speed and computing power, many researchers in computer science have turned to networks of computers as a possible solution. 1 , 5 , 8 , 14 , 9 These networks consist of minicomputers connected by links across which communication between processors occurs. In homogeneous networks, the computer at each node is identical to the others, with the possible exception of peripherals. Each processor has its own local memory, does not share memory with any other processor, and communicates with other processors via message passing. In order to fully utilize the speed and power inherent in a network, emphasis must be placed on the development of parallel (as opposed to sequential) algorithms.