A Performance Analysis of Phantom-Cell Adaptive Mesh Refinement on CPUs and GPUs

2020 ◽  
Vol 21 (1) ◽  
Author(s):  
Daniel J. Dunning ◽  
Robert W. Robey ◽  
Jeffery A. Kuehn ◽  
Jeanine Cook
Keyword(s):  
Adaptive Mesh Refinement ◽  
Graphics Processing Unit ◽  
Mesh Refinement ◽  
Adaptive Mesh ◽  
Processing Unit ◽  
Regular Grid ◽  
A Cell ◽  
Grid Applications ◽  
Graphics Processing ◽  
Phantom Cell

Abstract This paper presents an implementation of phantom-cell adaptive mesh refinement (AMR) on a graphics processing unit (GPU) using CLAMR, a cell-based mini-app that runs on a variety of next-generation platforms. Phantom-cell AMR is a hybrid method of cell-based AMR and patch-based AMR that provides a separation of physics and mesh codes. By designing a structure that allows each level of the mesh to be independent, there are minimal development requirements that are needed to convert regular grid applications to AMR. The decoupling of physics and mesh codes through these phantom cells improves composability and creates an easy pathway toward implementing AMR codes on Exascale systems, specifically targeting GPUs. Physics and mesh codes can be accelerated individually, allowing for fewer dependencies and more opportunities for optimization. A complete implementation of phantom-cell AMR on a GPU with opencl is presented for the purpose of showing the simplicity of porting the algorithms to accelerator-based architectures and the performance and optimization improvements that are made as a result.

scholarly journals A Local Adaptive Mesh Refinement for JFO Cavitation Model on Cartesian Meshes

2021 ◽  
Vol 11 (21) ◽  
pp. 9879
Author(s):  
Wanjun Xu ◽  
Kang Li ◽  
Zhengyang Geng ◽  
Mingjie Zhang ◽  
Jiangang Yang
Keyword(s):  
Adaptive Mesh Refinement ◽  
Mesh Refinement ◽  
Computational Cost ◽  
Adaptive Mesh ◽  
Posteriori Error ◽  
Uniform Mesh ◽  
Nonuniform Mesh ◽  
A Cell ◽  
The Difference ◽  
Richardson Extrapolation Method

Nonuniform mesh is beneficial to reduce computational cost and improve the resolution of the interest area. In the paper, a cell-based adaptive mesh refinement (AMR) method was developed for bearing cavitation simulation. The bearing mesh can be optimized by local refinement and coarsening, allowing for a reasonable solution with special purpose. The AMR algorithm was constructed based on a quadtree data structure with a Z-order filling curve managing cells. The hybrids of interpolation schemes on hanging nodes were applied. A cell matching method was used to handle periodic boundary conditions. The difference schemes at the nonuniform mesh for the universal Reynolds equation were derived. Ausas’ cavitation algorithm was integrated into the AMR algorithm. The Richardson extrapolation method was employed as an a posteriori error estimation to guide the areas where they need to be refined. The cases of a journal bearing and a thrust bearing were studied. The results showed that the AMR method provided nearly the same accuracy results compared with the uniform mesh, while the number of mesh was reduced to 50–60% of the number of the uniform mesh. The computational efficiency was effectively improved. The AMR method is suggested to be a potential tool for bearing cavitation simulation.

Dynamic structural analysis of connecting rod through adaptive mesh refinement for different materials

2018 ◽  
Vol 50 (04) ◽  
pp. 561-570
Author(s):  
I. A. QAZI ◽  
A. F. ABBASI ◽  
M. S. JAMALI ◽  
INTIZAR INTIZAR ◽  
A. TUNIO ◽  
...  
Keyword(s):  
Structural Analysis ◽  
Adaptive Mesh Refinement ◽  
Mesh Refinement ◽  
Adaptive Mesh ◽  
Connecting Rod ◽  
Dynamic Structural Analysis
Sign in / Sign up

Export Citation Format

Share Document