A parallel solution to finding nodal neighbors in generic meshes

Pian Qi,Gang Mei,Nengxiong Xu,Hong Tian

doi:10.1016/j.mex.2020.100954

Pian Qi, Gang Mei + Show 2 more

Open Access

https://doi.org/10.1016/j.mex.2020.100954

Copy DOI

Abstract

In this paper we specifically present a parallel solution to finding the one-ring neighboring nodes and elements for each vertex in generic meshes. The finding of nodal neighbors is computationally straightforward but expensive for large meshes. To improve the efficiency, the parallelism is adopted by utilizing the modern Graphics Processing Unit (GPU). The presented parallel solution is heavily dependent on the parallel sorting, scan, and reduction. Our parallel solution is efficient and easy to implement, but requires the allocation of large device memory.•Our parallel solution can generate the speedups of approximately 55 and 90 over the serial solution when finding the neighboring nodes and elements, respectively.•It is easy to implement due to the reason it does not need to perform the mesh-coloring before finding neighbors•There are no complex data structures, only integer arrays are needed, which makes our parallel solution very effective.

Highlights

Mesh generation plays an important role in geometric modeling, computer graphics, and numerical simulations
The race condition issue appears when two different parallel threads may need to be written in the same memory position [3]
We have presented a parallel solution to finding the neighboring nodes and elements for each vertex in an arbitrary mesh by exploiting the Graphics Processing Unit (GPU)

Summary

Introduction

Mesh generation plays an important role in geometric modeling, computer graphics, and numerical simulations. After generating various types of meshes, typically mesh editing is intentionally performed to modify or improve the generated meshes to meet desired requirements. In mesh editing such as Boolean operations [8] or mesh optimization [9], the local mesh connectivity especially the adjacent/neighboring nodes and elements for each node or element is frequently needed to reduce the computational cost of local search. The simplest method is to loop over all elements in a mesh to identify: (1) which pair of nodes is connected by an edge and (2) which nodes are contained in an element [4,9,10]. This is because that: (1) any pair of nodes connected by an edge is the one-ring neighboring node for each other; and (2) any element is directly the one-ring neighboring element for those nodes it contains

Methods

Discussion

Conclusion