PERFORMANCE EVALUATION OF DIFFERENT LINEAR EQUATION SOLVERS FOR SOLVING NONLINEAR FE PROBLEMS ON MULTICORE ARCHITECTURES

Michal Bošanský,Bořek Patzák

doi:10.14311/app.2018.15.0006

Abstract

<p>The aim of this paper is to evaluate the performance of existing parallel linear equation solvers to solving large-scale, nonlinear finite element analysis problems on systems with distributed memory. The parallel approach allows us to take an advantage of the distributed memory enabling forming large system matrices and of multiple processing units to achieve significant speedups. Our study is based on comparison of parallel direct solver and parallel iterative solver implemented in SuperLU DIST library from Portable, Extensible Toolkit for Scientific Computation (PETSc). Both considered solvers are designed for distributed system memory model and are based on a Massage Passing Interface (MPI).</p><p>The efficiency of individual solvers is evaluated on a selected benchmark problems, with different solution strategies by comparing computation times and obtained speedups.</p>

Highlights

Computational advancements have started a new trend in computational science and engineering that were powered by development and increases in the power and pervasiveness of computer and communication
Traditional finite element solvers run in serial mode and are limited by available resources represented by a single machine
The serial computers permit us to run simulation codes only sequentially and often have limited available resources represented by the single processing unit and available system memory

Summary

Introduction

Computational advancements have started a new trend in computational science and engineering that were powered by development and increases in the power and pervasiveness of computer and communication. Capitalization on those advances by developing techniques for modern hardware enable the solution of large and complex problems. The serial computers permit us to run simulation codes only sequentially and often have limited available resources represented by the single processing unit and available system memory. Parallelization can significantly reduce the solution time by more efficient use of modern hardware and enable to solve large problems by utilizing distributed memory resources. The explicit communications between processing units are needed to establish data exchange

Objectives

Results

Conclusion