Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 112
PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED, GPU AND CLOUD COMPUTING FOR ENGINEERING
Edited by: P. Iványi and B.H.V. Topping
Paper 24

Memory saving GPU implementation of contact search for the discrete element method

R. Pacevic1,2, A. Kaceniauskas2, R. Kacianauskas2 and R. Barauskas1

1Department of Applied Informatics, Kaunas University of Technology, Lithuania
2Department of Graphical Systems, Vilnius Gediminas Technical University, Lithuania

Full Bibliographic Reference for this paper
R. Pacevic, A. Kaceniauskas, R. Kacianauskas, R. Barauskas, "Memory saving GPU implementation of contact search for the discrete element method", in P. Iványi, B.H.V. Topping, (Editors), "Proceedings of the Sixth International Conference on Parallel, Distributed, GPU and Cloud Computing for Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 24, 2019. doi:10.4203/ccp.112.24
Keywords: GPU, discrete element method, contact search, OpenCL, hopper flow.

Summary
The paper presents a memory saving GPU implementation of contact search for the discrete element method (DEM) simulations. The developed GPU implementation is particularly suitable for DEM simulations of large number of discrete particles, moving in large computational domains with empty regions, because the size of data structures for contact search does not depend on the number of uniform grid cells. The implemented hash table of fixed size has the added benefit of allowing the grid to be unbounded in size. The performance of the developed OpenCL code is evaluated solving applications of particle fall under gravity. Performance achieved by using the developed implementation of contact search is compared with that attained by using the standard uniform grid method. The performance measured on NVIDIA® Tesla™P100 GPU is compared with that attained by using the same OpenCL code on Intel®Xeon™ E5-2630 CPU with 20 cores. Sufficiently high speedup values are observed for different numbers of particles in spite of intensive usage of advanced vector extensions by OpenCL on CPU. Performed analysis reveals that the developed GPU implementation of contact search significantly reduce the memory required for discrete element method simulations of hopper flow.

purchase the full-text of this paper (price £22)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description