Performance Evaluation of Parallel Large-Scale Lattice Boltzmann Applications on Three Supercomputing Architectures

ABSTRACT

Computationally intensive programs with moderate communication requirements such as CFD codes suffer from the standard slow interconnects of commodity "off the shelf" (COTS) hardware. We will introduce different large-scale applications of the Lattice Boltzmann Method (LBM) in fluid dynamics, material science, and chemical engineering and present results of the parallel performance on different architectures. It will be shown that a high speed communication network in combination with an efficient CPU is mandatory in order to achieve the required performance. An estimation of the necessary CPU count to meet the performance of 1 TFlop/s will be given as well as a prediction as to which architecture is the most suitable for LBM. Finally, ratios of costs to application performance for tailored HPC systems and COTS architectures will be presented.

REFERENCES

• 1: [1] Technical Report, RRZE, 2004. Available at [e-mail].
• 2: [2] J. Wilke, T. Pohl, and M. Kowarschik and U. Rüde. Cache Performance Optimizations for Parallel Lattice Boltzmann Codes in 2D. In Lecture Notes in Computer Science, volume 2790, pages 441-450. Springer, 2003.
• 3: [3] T. Pohl, M. Kowarschik, J. Wilke, K. Iglberger, and U. Rüde. Optimization and Profiling of the Cache Performance of Parallel Lattice Boltzmann Codes. Parallel Processing Letters, 13(4):549-560, 2003.
• 4: [4] Horst D. Simon, C. William McCurdy, and William T. C. Kramer et al. Creating Science-Driven Computer Architecture: A New Path to Scientific Leadership, 2003. http://www.nersc.gov/news/HECRTF-V4- 2003.pdf.
• 5: [5] Y. H. Qian, D. d'Humières, and P. Lallemand. Lattice BGK Models for Navier-Stokes Equation. Europhys. Lett., 17(6):479-484, 1992.
• 6: [6] Dieter A. Wolf-Gladrow. Lattice-Gas Cellular Automata and Lattice Boltzmann Models. Springer, 2000.
• 7: [7] A. J. C. Ladd. Numerical Simulations of particulate Suspensions via a discrete Boltzmann Equation Part 1. Theoretical foundation. Journal of Fluid Mechanics, pages 271-285, 1994.
• 8: [8] Shuling Hou, James D. Sterling, Shiyi Chen, and Gary Doolen. A Lattice Boltzmann Subgrid Model for High Reynolds Number Flow. Fields Institute Communications, 6:151-166, 1996.
• 9: [9] Christian Wimmer. Bewertung von Lattice-Boltzmann und Finite-Volumen Verfahren anhand von praxisrelevanten Aufgabenstellungen aus dem Bereich der Automobilaerodynamik. Master's thesis, Lehrstuhl für Strömungsmechanik, Universität Erlangen-Nürnberg, 2001.
• 10: [10] Peter Lammers, Jovan Jovanovic, and Franz Durst. Numerical experiment on wall turbulence. Phys. Fluids, 2003. submitted.
• 11: [11] John Kim, Parviz Moin, and Robert Moser. Turbulence statistics in fully developed channel flow at low Reynolds number. J. Fluid Mech., 177:133-166, 1987.
• 12: [12] Robert Moser, John Kim, and Nagi Mansour. Direct numerical simulation of turbulent channel flow up to Re¿ = 590. Phys. Fluids, 11, 1999.
• 13: [13] Peter Lammers, Kamen Beronov, Gunther Brenner, and Franz Durst. Direct simulation with the lattice Boltzmann code BEST of developed turbulence in channel flows. In S. Wagner, W. Hanke, A. Bode, and F. Durst, editors, High Performance Computing in Science and Engineering, Munich 2002, pages 43-58. Springer, 2003.
• 14: [14] Peter Lammers, Kamen Beronov, Thomas Zeiser, and Franz Durst. Testing of closure assumption for fully developed turbulent channel flow with the aid of a lattice Boltzmann simulation. In High Performance Computing in Science and Engineering, Munich 2004, pages 77-92. Springer, 2004.
• 15: [15] C. Körner and R. F. Singer. Processing of Metal Foams - Challenges and Opportunities. Advanced Engineering Materials, 2(4):159-65, 2000.
• 16: [16] William Lorensen and Harvey Cline. Marching Cubes: A High Resolution 3D Surface Reconstruction Algorithm. In Computer Graphics Vol. 21, No. 4, pages 163-169, August 1987.
• 17: Stanley Osher , James A. Sethian, Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations, Journal of Computational Physics, v.79 n.1, p.12-49, Nov. 1988 [doi>10.1016/0021-9991(88)90002-2]
• 18: A Superscalar RISC Processor with 160 FPRs for Large Scale Scientific Processing, Proceedings of the 1999 IEEE International Conference on Computer Design, p.279, October 10-13, 1999