We describe the development of a data-level, massivelyparallel software system for the solution of multicommodity network flow problems. Using a smooth linear-quadratic penalty (LQP) algorithm we transform the multic...
详细信息
We describe the development of a data-level, massivelyparallel software system for the solution of multicommodity network flow problems. Using a smooth linear-quadratic penalty (LQP) algorithm we transform the multicommodity network flow problem into a sequence of independent min-cost network flow subproblems. The solution of these problems is coordinated via a simple, dense, nonlinear master program to obtain a solution that is feasible within some user-specified tolerance to the original multicommodity network flow problem. Particular emphasis is placed on the mapping of both the subproblem and master problem data to the processing elements of a massivelyparallel computer, the Connection Machine CM-2. As a result of this design we can solve large and sparse optimization problems on current SIMD massivelyparallel architectures. Details of the implementation are reported, together with summary computational results with a set of test problems drawn from a Military Airlift Command application.
Until now, buses have been used exclusively to support fast communication and data transfer needs within a parallel machine. Recently, the authors have demonstrated novel uses of buses in a multiprocessor reconfigurab...
详细信息
Until now, buses have been used exclusively to support fast communication and data transfer needs within a parallel machine. Recently, the authors have demonstrated novel uses of buses in a multiprocessor reconfigurable architecture. Specifically, we have argued that buses can be successfully used both as topological descriptors and as powerful computational devices. The purpose of this note is to show that this new paradigm can be used for computing the Hough transform the most widely used technique for detecting lines and curves in binary or grey level images. Specifically, for an N x N digitized image stored in a reconfigurable mesh of size N x N, one pixel per processor, we present an O(n x log(N/n)) time algorithm to compute the Hough transform with n standing for the number of angles in the quantization of the theta-space.
We discuss implementation of additive Schwarz type algorithms on SIMD computers. A recursive, additive algorithm is compared with a two-level scheme. These methods are based on a subdivision of the domain into thousan...
详细信息
ISBN:
(纸本)0898712882
We discuss implementation of additive Schwarz type algorithms on SIMD computers. A recursive, additive algorithm is compared with a two-level scheme. These methods are based on a subdivision of the domain into thousands of micro-patches that can reflect local properties, coupled with a coarser, global discretization where the `macro' behavior is reflected. The two-level method shows very promising flexibility, convergence and performance properties when implemented on a massivelyparallel SIMD computer.
暂无评论