A massively parallel and memory-efficient FEM toolbox with a hybrid total FETI solver with accelerator support. (July 2019)

• Record Type: Journal Article
• Authors: Riha, Lubomir; Merta, Michal; Vavrik, Radim; Brzobohaty, Tomas; Markopoulos, Alexandros; Meca, Ondrej; Vysocky, Ondrej; Kozubek, Tomas; Vondrak, Vit
• Is Part Of: International journal of high performance computing applications. Volume 33:Number 4(2019)
• Page Start: 660
• Page End: 677
• Publication Date: 2019-07
• Subjects: Finite element method -- hybrid total finite element tearing and interconnecting -- graphics processing unit -- Xeon Phi -- Knights Corner -- Knights Landing -- massively parallel -- sparse linear solver; High performance computing -- Periodicals; Supercomputers -- Periodicals; 004.1105
• DOI: 10.1177/1094342018798452
• Languages: English
• ISSNs: 1094-3420

Abstract

In this article, we present the ExaScale PaRallel finite element tearing and interconnecting SOlver (ESPRESO) finite element method (FEM) library, which includes an FEM toolbox with interfaces to professional and open-source simulation tools, and a massively parallel hybrid total finite element tearing and interconnecting (HTFETI) solver which can fully utilize the Oak Ridge Leadership Computing Facility Titan supercomputer and achieve superlinear scaling. This article presents several new techniques for finite element tearing and interconnecting (FETI) solvers designed for efficient utilization of supercomputers with a focus on (i) performance—we present a fivefold reduction of solver runtime for the Laplace equation by redesigning the FETI solver and offloading the key workload to the accelerator. We compare Intel Xeon Phi 7120p and Tesla K80 and P100 accelerators to Intel Xeon E5-2680v3 and Xeon Phi 7210 central processing units; and (ii) memory efficiency—we present two techniques which increase the efficiency of the HTFETI solver 1.8 times and push the limits of the largest possible problem ESPRESO that can solve from 124 to 223 billion unknowns for problems with unstructured meshes. Finally, we show that by dynamically tuning hardware parameters, we can reduce energy consumption by up to 33%.