Title | ||
---|---|---|
A Lightweight Collective Communication Based Parallel Algorithm for the Greedy Point Selection in RBF Mesh Deformation |
Abstract | ||
---|---|---|
The greedy algorithm is one of the most efficient point selection method utilized by radial basis function (RBF) based mesh deformation in computational fluid dynamic (CFD). To improve the computational efficiency of the greedy process, several parallelization methods are developed in the past few years for the data reduction. However, these methods produce additional time cost because of the communication redundancy which can restrict the parallel scale. This paper proposes a new parallel strategy for the greedy selection in which a lightweight collective communication is adopted to replace the point-to-point communication. In addition, this strategy emphasizes that all the parallel processes operate under the same computational complexity as much as possible to reinforce the load balancing. Specifically, a three-dimension undulating fish model is utilized to validate the lightweight collective communication strategy. And finally, the proposed one gets up to 20% time performance improvement on average compared with two classic parallel methods. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/HPCC/SmartCity/DSS.2019.00027 | 2019 IEEE 21st International Conference on High Performance Computing and Communications; IEEE 17th International Conference on Smart City; IEEE 5th International Conference on Data Science and Systems (HPCC/SmartCity/DSS) |
Keywords | Field | DocType |
CFD mesh deformation,RBF interpolation,greedy algorithm,parallelization,collective communication,data reduction | Radial basis function,Computer science,Load balancing (computing),Parallel algorithm,Parallel computing,Greedy algorithm,Redundancy (engineering),Computational fluid dynamics,Computational complexity theory,Performance improvement,Distributed computing | Conference |
ISBN | Citations | PageRank |
978-1-7281-2059-1 | 0 | 0.34 |
References | Authors | |
0 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ran Zhao | 1 | 3 | 1.73 |
Chao Li | 2 | 320 | 46.22 |
Xiaowei Guo | 3 | 8 | 4.60 |
Sijiang Fan | 4 | 3 | 1.06 |
Yi Wang | 5 | 1520 | 135.81 |
Yi Liu | 6 | 2 | 1.06 |
Canqun Yang | 7 | 188 | 29.39 |