A multiple-GPU based parallel independent coefficient reanalysis method and applications for vehicle design. (July 2015)

• Record Type: Journal Article
• Authors: He, Guanqiang; Wang, Hu; Li, Enying; Huang, Guanxin; Li, Guangyao
• Is Part Of: Advances in engineering software. Volume 85(2015)
• Page Start: 108
• Page End: 124
• Publication Date: 2015-07
• Subjects: Parallel -- GPU -- MPI -- Reanalysis -- Vehicle design; Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2015.03.006
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: An independent coefficient reanalysis method is reconstructed for multi-GPU platform by using MPI and CUDA. The data partition for multiple GPUs is successfully implemented to achieve good load balance. The suggested non-blocking communication strategy achieves higher speedups compared with blocking one. The bottleneck of GPU memory can be solved. Abstract: The main limits of reanalysis method using CUDA (Compute Unified Device Architecture) for large-scale engineering optimization problems are low efficiency on single GPU and memory bottleneck of GPU. To breakthrough these bottlenecks, an efficient parallel independent coefficient (IC) reanalysis method is developed based on multiple GPUs platform. The IC reanalysis procedure is reconstructed to accommodate the use of multiple GPUs. The matrices and vectors are successfully partitioned and prepared for each GPU to achieve good load balance as well as little communication between GPUs. This study also proposes an effective technique to overlap the computation and communication by using non-blocking communication strategy. GPUs would continue their succeeding tasks while communication is still carried out simultaneously. Furthermore, the CSR format is used in each GPU for saving the memory. Finally, large-scale vehicle design problems are implemented by the developed solver. According to the test results, the multi-GPU based IC reanalysis method has potential capability for handling the real large scale problem and reducing the design cycle.