Vectorized implementation of the FEM numerical integration algorithm on a modern CPU / Filip Krużel // W: Proceedings of the 33rd International ECMS Conference on Modelling and Simulation : ECMS 2019, ... więcej

Vectorized implementation of the FEM numerical integration algorithm on a modern CPU / Filip Krużel // W: Proceedings of the 33rd International ECMS Conference on Modelling and Simulation : ECMS 2019, June 11th - June 14th, 2019, Caserta, Italy [Dokument elektroniczny] / ed. by Mauro Iacono [et al.] ; org. by ECMS - European Council for Modelling and Simulation. – [S.l.] : European Council for Modelling and Simulation, 2019. – (Communications of the ECMS, ISSN 2522-2422 ; Vol. 33, Iss. 1). – S. 414-420 : tab. – Tryb dostępu: http://www.scs-europe.net/dlib/2019/ecms2019acceptedpapers/0414_dis_ecms2019_0049.pdf. – Oznaczenie ref./art.: 0414. – Bibliogr. s. 420, Abstr. – doi: 10.7148/2019-0414. – ISBN 978-3-937436-65-4. – ISBN 978-3-937436-66-1 CD

The main aim of this study is to answer the question: how to effectively implement the creation of the finite element stiffness matrix in parallel simulations of Finite Element Method using the full advantages ... więcej

The main aim of this study is to answer the question: how to effectively implement the creation of the finite element stiffness matrix in parallel simulations of Finite Element Method using the full advantages of modern multiprocessors such as parallelization combined with vectorization. In this work, an efficient method for implementation of a Finite Element Method numerical integration algorithm on a modern Intel Haswell CPU architecture was developed. This algorithm was chosen, due to its non-trivial structure and the fact that its optimization is often omitted in research in favour of accelerating the other phases of FEM. Tests included two types of tasks to solve, with the use of two types of approximation and two types of finite elements. During this study, several methods for the implementation of the chosen algorithm was investigated, including Intel Cilk Plus, Intel Intrinsics and other computing techniques. Results were compared with an older Sandy Bridge architecture, showing a significant impact of vectorization and large cache on the performance of the modern CPUs. Our research gives suggestions for choosing the optimal design of algorithms and effectively using all of the features of the modern CPUs.