Fine-Grained Parallel Algorithm for Unstructured Surface Mesh Generation

Surface mesh generation is one time-consuming step in preparing an unstructured mesh model. However, its parallelisation attracts little attention. In this study, a fine-grained parallel surface meshing algorithm is proposed by taking advantage of the parallelism within the meshing process of a single face. Compared with the scheme which meshes the faces individually in parallel, the proposed algorithm behaves better in terms of parallel efficiency and scalability. One integral part of the proposed algorithm is a novel domain decomposition approach, which decomposes the simplified version of an input background mesh rather than the input mesh itself. The simplification procedure is based on a set of welldesigned operations on the dual graphs of the mesh. No undesirable features are formed in subdomain boundaries; hence, no postprocessing steps are required to improve the quality of elements adjacent to subdomain boundaries.