摘要:

AbstractWe present a parallel algorithm for distributed memory multiprocessors, which is based on generalized marching (GM), one of the fastest methods in the class of fast Poisson solvers.The GM algorithm is not suited for any but very coarse‐grain parallel processing. The main difficulty with parallelization is that the number of independent processes and the amount of work in each process change exponentially and in inverse proportion of each other.To improve parallelism, the matrices involved in GM are diagonalized performing multiple FFTs. In this way, independent processes extending across all the algorithm are obtained.The parallel GM has been tested on an Ncube/10 and a Symult S2010, running the Express communication system. A performance evaluation has been carried out using a scaled efficiency model and some classical parameter

ISSN:1040-3108    

DOI:10.1002/cpe.4330040702

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

摘要:

AbstractThe spectral transform method is a widely used numerical technique for solving partial differential equations on the sphere in global climate modeling. This paper describes the parallelization and performance of the spectral method for solving the non‐linear shallow water equations on the surface of a sphere using a 128‐node Intel iPSC/860 hypercube. Solving the shallow water equations represents a computational kernel of more complex climate models. This work is part of a research program to develop climate models that are capable of much longer simulations at a significantly finer resolution than current models. Such models are important in understanding the effects of the increasing atmospheric concentrations of greenhouse gases, and the computational requirements are so large that massively parallel multiprocessors will be necessary to run climate model simulations in a reasonable amount of time.The spectral method involves the transformation of data between the physical, Fourier and spectral domains. Each of these domains is two‐dimensional. The spectral method performs Fourier transforms in the longitude direction followed by summation in the latitude direction to evaluate the discrete spectral transform. A simple way of parallelizing the spectral code is to decompose the physical problem domain in just the latitude direction. This allows an optimized sequential FFT algorithm to be used in the longitude direction. However, this approach limits the number of processors that can be brought to bear on the problem. Decomposing the problem over both directions allows the parallelism inherent in the problem to be exploited more effectively‐the grain size is reduced, so that more processors can be used.Results are presented that show that decomposing over both directions does result in a more rapid solution of the problem. The results show that, for a given problem and number of processors, the optimum decomposition has approximately equal numbers of processors in each direction. Load imbalance also has an impact on the performance of the method. The importance of minimizing communication latency and overlapping communication with calculation is stressed. General methods for doing this, that may be applied to many other problems, are di

ISSN:1040-3108    

DOI:10.1002/cpe.4330040703

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

摘要:

AbstractThis paper considers a method for evaluating the concurrent programming language available on a multiprocessor system – in particular, a shared‐memory multiprocessor system. The method is based on the consideration of well‐known algorithms which have been used as a means of demonstrating the expressive power of new concurrent language constructs. The algorithms considered are the Readers and Writers Problem and the Drinking Philosophers Problem. Both algorithms are coded using features that are widely regarded as being state‐of‐the‐art for shared‐memory systems and are then recoded in the language available on an actual multiprocessor system in order to carry out a comparison and evaluation. Several observations and conclusi

ISSN:1040-3108    

DOI:10.1002/cpe.4330040704

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

摘要:

AbstractThree optimization methods derived from natural sciences are considered for allocating data to multicomputer nodes. These are simulated annealing, genetic algorithms and neural networks. A number of design choices and the addition of preprocessing and postprocessing steps lead to versions of the algorithms which differ in solution qualities and execution times. In this paper the performances of these versions are critically evaluated and compared for test cases with different features. The performance criteria are solution quality, execution time, robustness, bias and parallelizability. Experimental results show that the physical algorithms produce better solutions than those of recursive bisection methods and that they have diverse properties. Hence, different algorithms would be suitable for different applications. For example, the annealing and genetic algorithms produce better solutions and do not show a bias towards particular problem structures, but they are slower than the neural network algorithms. Preprocessing graph contraction is one of the additional steps suggested for the physical methods. It produces a significant reduction in execution time, which is necessary for their applicability to large problems.

ISSN:1040-3108    

DOI:10.1002/cpe.4330040705

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

ISSN:1040-3108    

DOI:10.1002/cpe.4330040701

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY