The Community for Technology Leaders
2013 IEEE 24th International Conference on Application-Specific Systems, Architectures and Processors (2013)
Washington, DC, USA USA
June 5, 2013 to June 7, 2013
ISSN: 2160-0511
ISBN: 978-1-4799-0494-5
pp: 321-328
Li Tang , Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, Indiana 46556, USA
X. Sharon Hu , Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, Indiana 46556, USA
Danny Z. Chen , Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, Indiana 46556, USA
Michael Niemier , Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, Indiana 46556, USA
Richard F. Barrett , Center for Computing Research, Sandia National Laboratories, Albuquerque, New Mexico 87123, USA
Simon D. Hammond , Center for Computing Research, Sandia National Laboratories, Albuquerque, New Mexico 87123, USA
Genie Hsieh , Center for Computing Research, Sandia National Laboratories, Albuquerque, New Mexico 87123, USA
ABSTRACT
The Finite Element Method (FEM) is a numerical technique widely used in finding approximate solutions for many scientific and engineering problems. The Data Assembly (DA) stage in FEM can take up to 50% of the total FEM execution time. Accelerating DA with Graphics Processing Units (GPUs) presents challenges due to DA's mixed compute-intensive and memory-intensive workloads. This paper uses a representative finite element mini-application to explore DA acceleration on CPU+GPU platforms. Implementations based on different thread, kernel and task design approaches are developed and compared. Their performance and energy consumption are measured on four CPU+GPU and two CPU only platforms. The results show that (i) the performance and energy for different implementations on the same platform can vary significantly but the performance and energy trends are the same, and (ii) there exist performance and energy tradeoffs across some platforms if the best implementation is chosen for each of the platforms.
INDEX TERMS
Graphics processing units, Instruction sets, Kernel, Finite element analysis, Acceleration, Computer architecture, Data communication
CITATION

L. Tang et al., "GPU acceleration of Data Assembly in Finite Element Methods and its energy implications," 2013 IEEE 24th International Conference on Application-Specific Systems, Architectures and Processors(ASAP), Washington, DC, USA USA, 2013, pp. 321-328.
doi:10.1109/ASAP.2013.6567597
81 ms
(Ver 3.3 (11022016))