In this contribution we will describe in detail a Density Functional Theory method based on a Daubechies wavelets basis set, named BigDFT. We will see that, thanks to wavelet properties, this code shows high systematic convergence properties, very good performances and an excellent efficiency for parallel calculations. BigDFT code operation are also well-suited for GPU acceleration. We will discuss how the problematic of fruitfully benefit of this new technology can be match with the needs of robustness and flexibility of a complex code like BigDFT. This work may be of interest not only for expert in electronic structure calculations, but may also provide feedback to the wider community of high performance scientific computing.
Dans cet article nous allons décrire en détail BigDFT, une mise en oeuvre de la Théorie de la Fonctionnelle de la Densité basée sur les ondelettes de Daubechies. Nous verrons que, grâce aux propriétés des ondelettes, ce code présente une bonne convergence systématique, de très bonnes performances et un excellent passage à l'échelle lors de calculs distribués. Les opérations constituantes de BigDFT sont également bien adaptée aux accélérateurs de type GPU. Nous analyserons comment bénéficier efficacement de cette nouvelle technologie tout en respectant les contraintes de robustesse et de flexibilité d'un programme de la complexité de BigDFT. Ce travail peut intéresser non seulement les experts en calcul des structures électroniques, mais également constituer un retour d'expérience pour la communauté plus large du calcul scientifique à hautes performances.
Mots-clés : Informatique, Théorie de la Fonctionnelle de la Densité, Ondelettes de Daubechies, BigDFT
Luigi Genovese 1; Brice Videau 2; Matthieu Ospici 3, 4, 2; Thierry Deutsch 2; Stefan Goedecker 5; Jean-François Méhaut 3
@article{CRMECA_2011__339_2-3_149_0,
author = {Luigi Genovese and Brice Videau and Matthieu Ospici and Thierry Deutsch and Stefan Goedecker and Jean-Fran\c{c}ois M\'ehaut},
title = {Daubechies wavelets for high performance electronic structure calculations: {The} {BigDFT} project},
journal = {Comptes Rendus. M\'ecanique},
pages = {149--164},
publisher = {Elsevier},
volume = {339},
number = {2-3},
year = {2011},
doi = {10.1016/j.crme.2010.12.003},
language = {en},
}
TY - JOUR AU - Luigi Genovese AU - Brice Videau AU - Matthieu Ospici AU - Thierry Deutsch AU - Stefan Goedecker AU - Jean-François Méhaut TI - Daubechies wavelets for high performance electronic structure calculations: The BigDFT project JO - Comptes Rendus. Mécanique PY - 2011 SP - 149 EP - 164 VL - 339 IS - 2-3 PB - Elsevier DO - 10.1016/j.crme.2010.12.003 LA - en ID - CRMECA_2011__339_2-3_149_0 ER -
%0 Journal Article %A Luigi Genovese %A Brice Videau %A Matthieu Ospici %A Thierry Deutsch %A Stefan Goedecker %A Jean-François Méhaut %T Daubechies wavelets for high performance electronic structure calculations: The BigDFT project %J Comptes Rendus. Mécanique %D 2011 %P 149-164 %V 339 %N 2-3 %I Elsevier %R 10.1016/j.crme.2010.12.003 %G en %F CRMECA_2011__339_2-3_149_0
Luigi Genovese; Brice Videau; Matthieu Ospici; Thierry Deutsch; Stefan Goedecker; Jean-François Méhaut. Daubechies wavelets for high performance electronic structure calculations: The BigDFT project. Comptes Rendus. Mécanique, High Performance Computing, Volume 339 (2011) no. 2-3, pp. 149-164. doi : 10.1016/j.crme.2010.12.003. https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.1016/j.crme.2010.12.003/
[1] Phys. Rev., 140 (1965), p. A1133
[2] ABINIT: first-principles approach to material and nanosystem properties, Comput. Phys. Comm., Volume 180 (2009), pp. 2582-2615
[3] et al. J. Chem. Phys., 129 (2008), p. 014109 http://inac.cea.fr/sp2m/L_Sim/BigDFT
[4] Ten Lectures on Wavelets, SIAM, Philadelphia, 1992
[5] Phys. Rev. B, 54 (1996), p. 1703
[6] Phys. Rev. B, 58 (1998), p. 3641
[7] Theor. Chem. Acc., 114 (2005), p. 145
[8] Wavelets and Their Application for the Solution of Partial Differential Equations, Presses Polytechniques Universitaires Romandes, Lausanne, Switzerland, 1998 (ISBN: 2-88074-398-2)
[9] SIAM J. Numer. Anal., 6 (1992), p. 1716
[10] J. Comput. Phys., 217 (2006), pp. 312-339
[11] Constr. Approx., 5 (1989), p. 49
[12] Efficient solution of Poisson's equation with free boundary conditions, J. Chem. Phys., Volume 125 (2006), p. 074105
[13] Efficient and accurate three-dimensional Poisson solver for surface problems, J. Chem. Phys., Volume 127 (2007), p. 054704
[14] Performance Optimization of Numerically Intensive Codes, SIAM Publishing Company, Philadelphia, USA, 2001 (ISBN: 0-89871-484-2)
[15] CPMD Version 3.8: developed by J. Hutter, A. Alavi, T. Deutsch, M. Bernasconi, S. Goedecker, D. Marx, M. Tuckerman, M. Parrinello, Max-Planck-Institut für Festkörperforschung and IBM Zürich Research Laboratory, 1995–1999.
[16] et al. J. Comput. Phys., 221 (2007), p. 779
[17] et al. Comput. Phys. Comm., 177 (2007), p. 298
[18] et al. Parallel Comput., 33 (2007), p. 10685
[19] www.amd.com/stream (ATI stream technology)
[20] http://www.nvidia.com/object/cuda_home.html (NVidia CUDA Programming Guide, version 3.1, see)
[21] The OpenCL Standard, 2009 http://www.khronos.org/opencl/
[22] www.apc.com/ (APC–Metered Rack PDU)
[23] SPIRAL: code generation for DSP transforms, Proc. IEEE (special issue on Program Generation, Optimization, and Adaptation), Volume 93 (2005) no. 2, pp. 232-275
[24] The design and implementation of FFTW3, Proc. IEEE (special issue on Program Generation, Optimization, and Platform Adaptation), Volume 93 (2005) no. 2, pp. 216-231 (Invited paper)
Cited by Sources:
Comments - Policy