Different Approaches to Parallelization of Sparse Matrix Assembly Operation

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21110%2F16%3A00235664" target="_blank" >RIV/68407700:21110/16:00235664 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Different Approaches to Parallelization of Sparse Matrix Assembly Operation

Popis výsledku v původním jazyce

The aim of this paper is to evaluate efficiency of different approaches to parallelization of stiffness matrix assembly operations, that can be found in any finite element software. OpenMP [1, 2] and POSIX Threads (Pthreads) programming models are two considered in this paper. The OpenMP model consist of an Application Program Interface (API) for multi-platform shared-memory parallel programming in C/C++. POSIX is an acronym for Portable Operating System Interface and Pthreads standards [3] defined as a set of C/C++ language [4] programming types and procedure calls for shared-memory parallel programming. The paper shows that parallelization can efficiently exploit the power of modern available hardware, significantly reducing the needed computation time. Different approaches of each programming model are used for parallelization of stiffness matrix assembly compared and their efficiency is evaluated in this paper. The different parallelization strategies were implemented in OOFEM [5] which is a free finite element code with object oriented architecture for solving mechanical, transport and fluid mechanics problems that operates on various platforms. The finite element method leads to set of algebraic equations which components are assembled from contributions of individual elements. In this paper we focus on assembly of sparse matrix contributions, such as stiffness as mass matrices. Domain decomposition paradigm, where the whole domain is decomposed into sub domain, which contributions are evaluated and assembled by individual threads is hard. The paper discuss the differences between approaches based on OpenMP and Pthreads, data scoping specification for correct parallel programming and memory allocation synchronization and scheduling.

Název v anglickém jazyce

Different Approaches to Parallelization of Sparse Matrix Assembly Operation

Popis výsledku anglicky

The aim of this paper is to evaluate efficiency of different approaches to parallelization of stiffness matrix assembly operations, that can be found in any finite element software. OpenMP [1, 2] and POSIX Threads (Pthreads) programming models are two considered in this paper. The OpenMP model consist of an Application Program Interface (API) for multi-platform shared-memory parallel programming in C/C++. POSIX is an acronym for Portable Operating System Interface and Pthreads standards [3] defined as a set of C/C++ language [4] programming types and procedure calls for shared-memory parallel programming. The paper shows that parallelization can efficiently exploit the power of modern available hardware, significantly reducing the needed computation time. Different approaches of each programming model are used for parallelization of stiffness matrix assembly compared and their efficiency is evaluated in this paper. The different parallelization strategies were implemented in OOFEM [5] which is a free finite element code with object oriented architecture for solving mechanical, transport and fluid mechanics problems that operates on various platforms. The finite element method leads to set of algebraic equations which components are assembled from contributions of individual elements. In this paper we focus on assembly of sparse matrix contributions, such as stiffness as mass matrices. Domain decomposition paradigm, where the whole domain is decomposed into sub domain, which contributions are evaluated and assembled by individual threads is hard. The paper discuss the differences between approaches based on OpenMP and Pthreads, data scoping specification for correct parallel programming and memory allocation synchronization and scheduling.

Druh

D - Stať ve sborníku

CEP obor

JM - Inženýrské stavitelství

OECD FORD obor

—

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2016

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Modern Methods of Experimental and Computational Investigations in Area of Construction

ISBN

978-3-03835-603-5

ISSN

1660-9336

e-ISSN

—

Počet stran výsledku

8

Strana od-do

91-98

Název nakladatele

Trans Tech Publications Inc.

Místo vydání

Pfaffikon

Místo konání akce

Praha

Datum konání akce

17. 9. 2015

Typ akce podle státní příslušnosti

EUR - Evropská akce

Kód UT WoS článku

—