In this paper, we describe the decomposition of six algorithms: two partial differential equations (PDE) solvers (successive over-relaxation [SOR] and alternating direction implicit [ADI]), fast Fourier transform (FFT), Monte Carlo simulations, Simplex linear programming, and Sparse solvers. The algorithms were selected not only because of their importance in scientific applications, but also because they represent a variety of computational (structured to irregular) and communication (low to high) requirements. We present the performance results TM 1 of these algorithms on two shared-memory VAX/VMS multiprocessor prototypes: the VAX 6300 series with up to 8 processors and the M31 with up to 22 processors. We demonstrate that by efficient decomposition it is possible to achieve high performance for all algorithms on both prototypes. We describe the efficient decomposition techniques applied to optimize the performance of parallel algorithms. Also, we discuss the performance implications due to different cache designs on two multiprocessors.

In this paper we compare three parallel Astar search algorithms: in the Shared-Lisr algorithm a search space is shared among processors, in the Static Distribution algorithm a search space is dis-and some processors will work on parts of the search space not considered by the sequential algorithm. The resulting load imbalance and extra search will lead to Sub-linear speedups. Dynamic techniques can Potentially Overcome the Problems Of static partitioning. tributed once to all processors, and in the Conrinuous Dillusion algorithm a search space is continuously redistributed. Our results indicate that the Continuous Diffusion algorithm outperforms the other two algorithms on message-passing architertures. Additionally, the Grid-Flow technique developed for Continuous Diffusion is of general importance for nearest-neighbor algorithms that need to share some global information.

In this paper, we describe the decomposition of six algorithms: two Partial Differential Equations (PDE) solvers (Successive Over-Relaxation (SOR) and Alternating Direction Implicit (ADI)), Fast Fourier Transform (FFI'), Monte Carlosimulations, Simplexlinearprogramming, and Sparse solvers. The algorithms were selected not only because of their importance in scientific applications, but also because they represent a variety of computational (structured to irregular) and communication (low to high) requirements. We present the performance results of these algorithms on two shared-memory VAXNMSm multiprocessor prototypes: VAX 6300 series with up to 8 processors and M3 1 with up to 22 processors. We demonstrate that by efficient decomposition it is possible to achieve high performance for all algorithms on both prototypes. We describe the efficient decomposition techniques applied to optimize the performance of parallel algorithms. Also, we discuss the performance implications due to different cache designs on two multip rocessors.