A GPU-based multi-level algorithm for boundary value problems

被引：2

作者：

Becerra-Sagredo, Julian-Tercero ^{[1
]}

Malaga, Carlos ^{[2
]}

Mandujano, Francisco ^{[2
]}

机构：

[1] Univ Autonoma Metropolitana, Dept Matemat, Mexico City, DF, Mexico

[2] Univ Nacl Autonoma Mexico, Fac Ciencias, Mexico City, DF, Mexico

来源：

JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS | 2020年 / 368卷

关键词：

Parallel multilevel; GPU; CUDA; MULTIGRID METHODS; PERFORMANCE; SOLVER; FLOW;

D O I：

10.1016/j.cam.2019.112481

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

A novel geometric single-grid multi-level (GSGML) algorithm is presented for the numerical solution of boundary value problems (BVPs). The algorithm is specifically designed for parallel scalability with high occupancy of streaming processors inside many-core architectures like the Graphics Processing Unit (GPU). The algorithm consists of iterative, superposed operations on a single grid, and is composed of two full-grid routines: a restriction and a coarsened interpolation-relaxation. The restriction is a local average of scalar or vector fields, and the interpolation-relaxation is a simultaneous application of coarsened finite-difference operators and interpolations in complementary subsets of the grid. The routines are scheduled in a saw-like refining cycle. Convergence to machine precision is achieved repeating the full cycle using residuals and superposition. The algorithm is attractive for its computational simplicity and provides a competitive GPU-based fast solver for elliptic partial differential equations like the Poisson-Helmholtz equation. Applications shown in this work include the deformation of two-dimensional grids, the computation of three-dimensional streamlines for a singular trifoil-knot vortex and the calculation of three-dimensional electric potentials in heterogeneous dielectric media. (C) 2019 Elsevier B.V. All rights reserved.

引用

页数：11

共 33 条

[11] Using GPUs to improve multigrid solver performance on a cluster [J].

Goeddeke, Dominik ;

Strzodka, Robert ;

Mohd-Yusof, Jamaludin ;

McCormick, Patrick ;

Wobker, Hilmar ;

Becker, Christian ;

Turek, Stefan .

INTERNATIONAL JOURNAL OF COMPUTATIONAL SCIENCE AND ENGINEERING, 2008, 4 (01) :36-55

[12]

Goodnight N., 2003, P 2003 ACM SIGGR EUR

[13]

Grady L, 2008, LECT NOTES COMPUT SC, V5303, P252, DOI 10.1007/978-3-540-88688-4_19

[14] Numerical analysis and implementational aspects of a new multilevel grid deformation method [J].

Grajewski, Matthias ;

Koester, Michael ;

Turek, Stefan .

APPLIED NUMERICAL MATHEMATICS, 2010, 60 (08) :767-781

[15]

Grama A., 2003, Introduction to parallel computing, V2nd

[16]

Grossauer H, 2008, LECT NOTES COMPUT SC, V5008, P141

[17]

Haase G, 2010, LECT NOTES COMPUT SC, V5938, P38, DOI 10.1007/978-3-642-11842-5_5

[18]

Hackbusch W., 1985, MULTIGRID METHODS AP

[19] General theory of domain decomposition: Beyond Schwarz methods [J].

Herrera, I ;

Yates, R .

NUMERICAL METHODS FOR PARTIAL DIFFERENTIAL EQUATIONS, 2001, 17 (05) :495-517

[20] Multigrid methods for space fractional partial differential equations [J].

Jiang, Yingjun ;

Xu, Xuejun .

JOURNAL OF COMPUTATIONAL PHYSICS, 2015, 302 :374-392

← 1 2 3 4 →