Numerical study of geometric multigrid methods on CPU-GPU heterogeneous computers

DOI10.4208/AAMM.2013.M87MaRDI QIDQ5498620zbMATH OpenOpenAlexFDO

Authors Chunsheng Feng, Shi Shu, Chensong Zhang, Jinchao Xu

Publication date 10 February 2015

Published in Advances in Applied Mathematics and Mechanics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1208.4247

zbMATH Keywords

CPU-GPU heterogeneous computersfast Fourier transformhigh-performance computingmultigrid methodpartial differential equations

Mathematics Subject Classification ID

Composite media; random media in optics and electromagnetic theory (78A48) Finite element, Galerkin and related methods applied to problems in optics and electromagnetic theory (78M10)

Abstract: The geometric multigrid method (GMG) is one of the most efficient solving techniques for discrete algebraic systems arising from elliptic partial differential equations. GMG utilizes a hierarchy of grids or discretizations and reduces the error at a number of frequencies simultaneously. Graphics processing units (GPUs) have recently burst onto the scientific computing scene as a technology that has yielded substantial performance and energy-efficiency improvements. A central challenge in implementing GMG on GPUs, though, is that computational work on coarse levels cannot fully utilize the capacity of a GPU. In this work, we perform numerical studies of GMG on CPU--GPU heterogeneous computers. Furthermore, we compare our implementation with an efficient CPU implementation of GMG and with the most popular fast Poisson solver, Fast Fourier Transform, in the cuFFT library developed by NVIDIA.

Recommendations

• Fast finite difference Poisson solvers on heterogeneous architectures
• Accelerating algebraic multigrid solvers on NVIDIA GPUs
• Exposing fine-grained parallelism in algebraic multigrid methods
• Parallel unsmoothed aggregation algebraic multigrid algorithms on GPUs
• Parallel geometric multigrid

Cited in

(9)

• Numerical solution and convergence analysis of steam injection in heavy oil reservoirs
• A GPU-based discrete element modeling code and its application in die filling
• A GPU-based multi-level algorithm for boundary value problems
• A GPU parallelized spectral method for elliptic equations in rectangular domains
• Exploiting batch processing on streaming architectures to solve 2D elliptic finite element problems: a hybridized discontinuous Galerkin (HDG) case study
• Adaptive relaxation strategy on basic iterative methods for solving linear systems with single and multiple right-hand sides
• Algebraic multigrid block triangular preconditioning for multidimensional three-temperature radiation diffusion equations
• Substructuring Preconditioners with a Simple Coarse Space for 2-D 3-T Radiation Diffusion Equations
• An implementation of tensor product patch smoothers on GPUs

Describes a project that uses

Uses Software

• CUDA
• FFTW
• cuFFT

This page was built for publication: Numerical study of geometric multigrid methods on CPU-GPU heterogeneous computers

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5498620)