Advertisement

A Strategy to Workload Division for Massively Particle-Particle N-body Simulations on GPUs

  • Daniel Madeira
  • José Ricardo
  • Diego H. Stalder
  • Leonardo Rocha
  • Reinaldo Rosa
  • Otton T. Silveira Filho
  • Esteban Clua
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8584)

Abstract

The new programmable graphical processor units (GPUs) are now often used as high parallel mathematical co-processors, allowing many computational-intensive problems to be executed in less time. It became common and convenient to use single GPUs to implement different kinds of simulations, or to group them in a grid, so the computational power can be highly increased, while the power consumption and physical space increase are significantly lower. The N-body simulation has been successfully ported to the GPUs. This algorithm is typically applied to gravitational simulations, among many other physical problems. In cosmology, one alternative approach seeks to explain the nature of dark matter as a direct result of the non-linear space-time curvature, due to different types of deformation potentials. In order to develop a detailed study of this new approach, our group is developing the COsmic LAgrangian TUrbulence Simulator (COLATUS_ENVIU). The simulator uses the direct particle-particle method to calculate the forces between the particles, so we eliminate the errors included on the hierarchical strategies. It gave robust initial results, but was limited to systems with a small amount of particles. However one limitation on the use of GPUs on N-body simulation to study the details on the mass distribution, is that these systems must have millions (or even billions) of particles, making the use of particle-particle method in one GPU unfeasible due to memory or time computational constraints. In this present work we propose a novel and efficient subdivision method to allow the workload division, allowing a single GPU to be able to solve large simulations, while allowing to watch over a particle during all the simulation.

Keywords

Dark Matter Graphic Process Unit Tile Size Chaotic Advection Multiple GPUs 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Makino, J., Taiji, M.: Scientific Simulations with Special-Purpose Computers–the GRAPE Systems (April 1998)Google Scholar
  2. 2.
    Felix Soehr, S.D.M.W.: Simulation of Galaxy Fomation and Large Scale Structure. Phd thesis, Ludwig-Maximilian Universitat Munchen (2003)Google Scholar
  3. 3.
    Springel, V.: The cosmological simulation code gadget-2. Monthly Notices of the Royal Astronomical Society 364(4), 1105–1134 (2005)CrossRefGoogle Scholar
  4. 4.
    Stalder, D.H., Rosa, R.R., da Silva Junior, J.R., Clua, E., Ruiz, R.S.R., Velho, H.F.C., Ramos, F.M., Araújo, A.D.S., Conrado, V.G.: A new gravitational n-body simulation algorithm for investigation of cosmological chaotic advection. In: The Sixth International School on Field Theory and Gravitation-2012. AIP Conference Proceedings, vol. 1483, pp. 447–452 (2012)Google Scholar
  5. 5.
    Barnes, J., Hut, P.: A hierarchical O(N log N) force-calculation algorithm. Nature 324(6096), 446–449 (1986)CrossRefGoogle Scholar
  6. 6.
    Warren, M., Salmon, J.: Astrophysical N-body simulations using hierarchical tree data structures. In: Proceedings of Supercomputing 1992, pp. 570–576. IEEE Comput. Soc. Press (1992)Google Scholar
  7. 7.
    Ajmera, P., Goradia, R., Chandran, S., Aluru, S.: Fast, Parallel, GPU-based. Direct (2008)Google Scholar
  8. 8.
    Grama, A.Y., Kumar, V., Sameh, A.: Scalable parallel formulations of the barnes-hut method for n -body simulations. In: Proceedings of the 1994 ACM/IEEE Conference on Supercomputing, Supercomputing 1994, p. 439. ACM Press, New York (1994)Google Scholar
  9. 9.
    Hamada, T., Narumi, T., Yokota, R., Yasuoka, K.: 42 TFlops hierarchical N-body simulations on GPUs with applications in both astrophysics and turbulence. In: Proceedings of the, p. 1 (2009)Google Scholar
  10. 10.
    Jeroen, B.: A sparse octree gravitational N-body code that runs entirely on the GPU processor. Journal of Computational Physics, 1–34 (2011)Google Scholar
  11. 11.
    Nakasato, N.: Oct-tree Method on GPU: $ 42 / Gflops Cosmological Simulation. Engineering (2009)Google Scholar
  12. 12.
    Stock, M.J., Gharakhani, A.: Toward efficient GPU-accelerated N -body simulations, 1–13 (2008)Google Scholar
  13. 13.
    Jetley, P., Wesolowski, L., Gioachin, F., Kal, L.V.: Scaling Hierarchical N -body Simulations on GPU Clusters. Science (November 2010)Google Scholar
  14. 14.
    Hairer, E., Lubich, C., Wanner, G.: Geometric numerical integration illustrated by the störmer-verlet method (2003)Google Scholar
  15. 15.
    von Hoerner, S.: Die numerische integration des n-körper-problemes für sternhaufen. i. Z. Astrophys. 50, 184–214 (1960)zbMATHMathSciNetGoogle Scholar
  16. 16.
    Nyland, L., Harris, M., Prins, J.: Fast N-Body Simulation with CUDA. In: Nguyen, H. (ed.) GPU Gems 3. Addison Wesley Professional (August 2007)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Daniel Madeira
    • 1
    • 2
  • José Ricardo
    • 2
  • Diego H. Stalder
    • 3
  • Leonardo Rocha
    • 1
  • Reinaldo Rosa
    • 3
  • Otton T. Silveira Filho
    • 2
  • Esteban Clua
    • 2
  1. 1.Universidade Federal de São João del ReiSão João del ReiBrasil
  2. 2.Universidade Federal FluminenseNiterói, Rio de JaneiroBrasil
  3. 3.Instituto Nacional de Pesquisas EspaciaisSão José dos CamposBrasil

Personalised recommendations