Solving Grid Equations in Heterogeneous Systems with Limited Memory Size

The paper considers the problem of solving systems of grid equations arising from the discretization of mathematical physics problems in heterogeneous computing systems with limited memory. An improved model for decomposing the computational domain is proposed, taking into account the technical characteristics of both CPU and GPU. Parallel algorithms based on the pipeline computation method for the modified alternating-triangular iterative method (MATM) are developed. The results of numerical experiments demonstrating the speedup and efficiency of the proposed approach on heterogeneous systems are presented.

Keywords

iteration methods; parallel programming; heterogeneous systems; limited memory.

References

1. Sukhinov A.I., Chistyakov A.E., Litvinov V.N., Nikitina A.V., Belova Yu.V., Filina A.A. Computational Aspects of Mathematical Modeling of Hydrobiological Processes. Numerical Methods and Programming, 2020, vol. 21, no. 4, pp. 452-469. DOT: 10.26089/NumMet.v21r436
2. Konovalov A.N. The Steepest Descent Method with an Adaptive Alternating-Triangular Preconditioner. Differential Equations, 2004, vol. 40, no. 7, pp. 1018-1028. DOI: 10.1023/B:DIEQ.0000047032.23099.e3
3. Klimonov I.A., Korneev V.D., Sveshnikov V.M. Parallelization Technologies for Solving Three-Dimensional Boundary Value Problems Using the CPU+GPU Hybrid Computing Environment. Numerical Methods and Programming, 2016, vol. 17, no. 1, pp. 65-71. DOI: 10.26089/NumMet.v17r107
4. Peng Ding, Zhe Liu. Accelerating Phase-Field Modeling of Solidification with a Parallel Adaptive Computational Domain Approach. International Communications in Heat and Mass Transfer, 2020, vol. 111, Article ID. 104452. DOI: 10.1016/j.icheatmasstransfer.2019.104452
5. Yifen Ke, Changfeng Ma. Adaptive Parameter Based Matrix Splitting Iiteration Method for the Large and Sparse Linear Systems. Computers and Mathematics with Applications, 2022. vol. 122, no. 1, pp. 19-27. DOI: 10.1016/j.camwa.2022.07.010
6. Liang Zheng, Taras Gerya, Matthew Knepley et al. GPU Implementation of Multigrid Solver for Stokes Equation with Strongly Variable Viscosity. Lecture Notes in Computer Science, 2013, pp. 321-333. DOI: 10.1007/978-3-642-16405-7_21
7. Gracheva N.N., Litvinov V.N., Nikitina A.V et al. Computational Aspects of Solving Grid Equations on Graphics Accelerators. Numerical Methods and Programming, 2021, no. 5, pp. 120-135. DOI: 10.18522/2311-3103-2021-5-93-105
8. Litvinov V.N., Rudenko N.B., Gracheva N.N. Development of a Parallel-Pipeline Computational Process Model for Solving Grid Equations. Advanced Engineering Research (Rostov-on-Don), 2023, vol. 23, no. 3, pp. 329-339. DOI: 10.23947/2687-1653-2023-23-3-329-339
9. Krasnopolsky B.E., Medvedev A.A., Chulyunin A.Y. On Application of GPUs for Modelling of Hydrodynamic Characteristics of Screw Marine Propellers in OpenFOAM Package. Proceedings of the Institute for System Programming of RAS, 2014, vol. 26, no. 5, pp. 155-172. DOI: 10.15514/ISPRAS-2014-26(5)-8
10. Chetverushkin B.N., Yakobovskiy M.V., Kornilina M.A., Semenova A.V. Numerical Algorithms for HPC Systems and Fault Tolerance. Communications in Computer and Information Science, 2019, vol. 1063, no. 1, pp. 34-44.