Layer-by-layer ordering in parallel finite element composition on shared-memory multiprocessors

11th International Conference on "Mesh methods for boundary-value problems and applications"

Novikov, A.K.^1 ; Piminova, N.K.^1 ; Kopysov, S.P.^1 ; Sagdeeva, Y.A.^1

Institute of Mechanics, Ural Branch of the Russian Academy of Sciences, 34 T. Baramzinoy, Izhevsk

426067, Russia^1

关键词: Element compositions;    Layer by layer;    Parallel assembling;    Parallel process;    Partitioning algorithms;    Potential conflict;    Shared memory multiprocessor;    Unstructured meshes;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/158/1/012071/pdf
DOI  :  10.1088/1757-899X/158/1/012071

来源: IOP

【 摘 要 】

In this paper, we present new partitioning algorithms for unstructured meshes that prevent conflicts during parallel assembling of FEM matrices and vectors in shared memory. These algorithms use a ratio which we introduce to determine if any two mesh cells are adjacent. This adjacency ratio defines mesh layers, which are combined into domains and assigned to different parallel processes/threads. The proposed partitioning algorithms are compared with the existing algorithms on quasi-structured and unstructured meshes by the number of potential conflicts and by the load imbalance.