Abstract
The paper analyzes parallelization of computations for the problem of magnetohydrodynamics. Parallelization is performed by cutting the computation graph into subgraphs. The algorithm allows for the architecture of the computing medium — the number and performance of processors, the topology of the communication subsystem, and the data transfer rate of the communication channels. Application of the algorithms for particular computing systems is described.
Similar content being viewed by others
References
V. V. Voevodin, Mathematical Principles of Parallel Computation [in Russian], Izd. MGU, Moscow (1991).
N. N. Popova, “Parallel computation in models of magnetohydrodynamics”, in: System Programming and Optimization Topics [in Russian], Izd. MGU, Moscow (1987), pp. 77–87.
N. M. Ershov and A. M. Popov, “Computer analysis of parallel computations for models of computational physics”, in: Mathematical Modeling and Solution of Inverse Problems of Mathematical Physics [in Russian], Izd. MGU, Moscow (1993).
Additional information
Translated from Programmno-apparatnye Sredstva i Matematicheskoe Obespechenie Vychislitel'nykh Sistem, pp. 20–23, 1994.
Rights and permissions
About this article
Cite this article
Ershov, N.M., Popova, N.N. Computer analysis of parallelization for computational models of continua. Comput Math Model 6, 204–207 (1995). https://doi.org/10.1007/BF01138126
Issue Date:
DOI: https://doi.org/10.1007/BF01138126