Estimation of Numerical Reproducibility on CPU and GPU

• Autorzy: Fabienne Jézéquel , Jean-Luc Lamotte , Issam Saïd
• Języki publikacji: EN

Abstrakty

EN

Differences in simulation results may be observed from one architecture to another or even inside the same architecture. Such reproducibility failures are often due to different rounding errors generated by different orders in the sequence of arithmetic operations. Reproducibility problems are particularly noticeable on new computing architectures such as multicore processors or GPUs (Graphics Processing Units). DSA (Discrete Stochastic Arithmetic) enables one to estimate rounding error propagation in simulation programs. In this paper, it is shown that DSA can be used to estimate which digits in simulation results may be different from one environment to another because of rounding errors. A particular implementation of DSA, which enables numerical validation in hybrid CPU-GPU environments, is described. The estimation of numerical reproducibility using DSA is illustrated by a wave propagation code which can be affected by reproducibility problems when executed on different architectures.(original abstract)

Słowa kluczowe

EN

Information science

PL

Informatyka

• Czasopismo: Annals of Computer Science and Information Systems
• Rocznik: 2015
• Tom: 5
• Strony: 675--680

Twórcy

autor

Fabienne Jézéquel

• Sorbonne Universités, Paris

autor

Jean-Luc Lamotte

• Sorbonne Universités, Paris

autor

Issam Saïd

• Sorbonne Universités, Paris

Bibliografia

• O. Villa, D. Chavarría-Miranda, V. Gurumoorthi, A. Márquez, and S. Krishnamoorthy, "Effects of floating-point non-associativity on numerical computations on massively multithreaded systems," in Cray User Group Meeting (CUG 2009), Atlanta, Georgia, USA, May 2009, pp. 1-11.
• Y. He and C. Ding, "Using accurate arithmetics to improve numerical reproducibility and stability in parallel applications," The Journal of Supercomputing, vol. 18, no. 3, pp. 259-277, 2001.
• M. Cleveland, T. Brunner, N. Gentile, and J. Keasler, "Obtaining identical results with double precision global accuracy on different numbers of processors in parallel particle Monte Carlo simulations," Journal of Computational Physics, vol. 251, pp. 223-236, 2013.
• M. Taufer, O. Padron, P. Saponaro, and S. Patel, "Improving numerical reproducibility and stability in large-scale numerical simulations on GPUs," in IEEE International Symposium on Parallel Distributed Processing (IPDPS), Atlanta, Georgia, USA, 2010, pp. 1-9.
• R. W. Robey, J. M. Robey, and R. Aulwes, "In search of numerical consistency in parallel programming," Parallel Computing, vol. 37, no. 4-5, pp. 217-229, Apr. 2011.
• J. Demmel and H. D. Nguyen, "Fast reproducible floating-point summation," in 21st IEEE Symposium on Computer Arithmetic (ARITH 21), Austin, Texas, USA, Apr. 2013, pp. 163-172.
• "Parallel reproducible summation," in IEEE Transactions on Computers, Special Section on Computer Arithmetic, to appear.
• S. Collange, D. Defour, S. Graillat, and R. Iakymchuk, "A reproducible accurate summation algorithm for High-Performance Computing," in SIAM Workshop on Exascale Applied Mathematics Challenges and Opportunities (EX14) held as part of the 2014 SIAM Annual Meeting, Chicago, Illinois, USA, Jul. 2014.
• "Reproducible and accurate matrix multiplication for HighPerformance Computing," in The 16th GAMM-IMACS International Symposium on Scientific Computing, Computer Arithmetic and Validated Numerics (SCAN'14), Würzburg, Germany, Sep. 2014.
• R. Iakymchuk, S. Collange, D. Defour, and S. Graillat, "Reproducible triangular solvers for high-performance computing," in 12th International Conference on Information Technology: New Generations (ITNG), Las Vegas, Nevada, USA, Apr. 2015. [Online]. Available: http://hal.archives-ouvertes.fr/hal-01116588v2
• S. Collange, D. Defour, S. Graillat, and R. Iakymchuk, "Fullspeed deterministic bit-accurate parallel floating-point summation on multi- and many-core architectures," http://hal.archives-ouvertes.fr/hal- 00949355v3/PDF/superaccumulator.pdf, Feb. 2015. [Online]. Available: http://hal.archives-ouvertes.fr/hal-00949355v3
• J. Vignes, "A stochastic arithmetic for reliable scientific computation," Mathematics and Computers in Simulation, vol. 35, pp. 233-261, 1993.
• "Discrete Stochastic Arithmetic for validating results of numerical software," Numerical Algorithms, vol. 37, no. 1-4, pp. 377-390, Dec. 2004.
• F. Jézéquel and J.-M. Chesneaux, "CADNA: a library for estimating round-off error propagation," Computer Physics Communications, vol. 178, no. 12, pp. 933-955, 2008.
• F. Jézéquel, J.-M. Chesneaux, and J.-L. Lamotte, "A new version of the CADNA library for estimating round-off error propagation in Fortran programs," Computer Physics Communications, vol. 181, no. 11, pp. 1927-1928, 2010.
• J.-L. Lamotte, J.-M. Chesneaux, and F. Jézéquel, "CADNA_C: A version of CADNA for use with C or C++ programs," Computer Physics Communications, vol. 181, no. 11, pp. 1925-1926, 2010.
• S. Montan and C. Denis, "Numerical verification of industrial numerical codes," in ESAIM: Proc., vol. 35, Mar. 2012. doi: 10.1051/proc/201235006 pp. 107-113.
• F. Jézéquel and J.-L. Lamotte, "Numerical validation of Slater integrals computation on GPU," in The 14th GAMM-IMACS International Symposium on Scientific Computing, Computer Arithmetic and Validated Numerics (SCAN'10), Lyon, France, Sep. 2010, pp. 78-79.
• H. Calandra, R. Dolbeau, P. Fortin, J.-L. Lamotte, and I. Said, "Forward seismic modeling on AMD Accelerated Processing Unit," in Rice Oil & Gas HPC Workshop, Houston, Texas, USA, Mar. 2013.
• B. Gustafsson, H.-O. Kreiss, and J. Oliger, Time Dependent Problems and Difference Methods, 2nd ed. Wiley, 2013.
• IEEE Computer Society, IEEE Standard for Floating-Point Arithmetic. IEEE Standard 754-2008, Aug. 2008. ISBN 978-0-7381-5752-8
• J.-M. Chesneaux and J. Vignes, "Sur la robustesse de la méthode CESTAC," Comptes Rendus de l'Académie des Sciences - Series I - Mathematics, vol. 307, pp. 855-860, 1988.
• J.-M. Chesneaux, "L'arithmétique stochastique et le logiciel CADNA," Habilitation à diriger des recherches, Université Pierre et Marie Curie, Paris, France, Nov. 1995.
• J.-M. Chesneaux and J. Vignes, "Les fondements de l'arithmétique stochastique," Comptes Rendus de l'Académie des Sciences - Series I - Mathematics, vol. 315, pp. 1435-1440, 1992.
• F. Jézéquel, F. Rico, J.-M. Chesneaux, and M. Charikhi, "Reliable computation of a multiple integral involved in the neutron star theory," Math. Comput. Simulation, vol. 71, no. 1, pp. 44-61, 2006.
• N. Scott, F. Jézéquel, C. Denis, and J.-M. Chesneaux, "Numerical 'health check' for scientific codes: the CADNA approach," Computer Physics Communications, vol. 176, no. 8, pp. 507-521, Apr. 2007.
• F. Jézéquel, J.-L. Lamotte, and O. Chubach, "Parallelization of Discrete Stochastic Arithmetic on multicore architectures," in 10th International Conference on Information Technology: New Generations (ITNG), Las Vegas, Nevada (USA), Apr. 2013.
• J. Brajard, P. Li, F. Jézéquel, H.-S. Benavidès, and S. Thiria, "Numerical Validation of Data Assimilation Codes Generated by the YAO Software," in SIAM Annual Meeting, San Diego, California (USA), Jul. 2013.
• W. Li, S. Simon, and S. Kiess, "On the numerical sensitivity of computer simulations on hybrid and parallel computing systems," in International Conference on High Performance Computing and Simulation (HPCS), Istanbul, Turkey, Jul. 2011, pp. 510-516.
• P. Eberhart, J. Brajard, P. Fortin, and F. Jézéquel, "Towards high performance stochastic arithmetic," in The 16th GAMM-IMACS International Symposium on Scientific Computing, Computer Arithmetic and Validated Numerics (SCAN'14), Würzburg, Germany, Sep. 2014.

Identyfikatory

DOI

10.15439/2015F29