ABSTRACT
Parallel programs regularly face performance issues due to inefficient load distribution. Before redistributing load to avoid idle processes, it is essential to identify and measure the imbalance. In this paper, we present our approach to investigate load imbalance by using performance visualization and five imbalance metrics. We analyzed Ondes3D, an earthquake simulator with time-spatial imbalance, used to predict and better understand the impact of disasters. With the Ondes3D computational signature, we obtained further information on load distribution during the application’s iterations to understand how the metrics measures evolve throughout execution. Our approach can be useful to other imbalanced applications, identifying when the load is unevenly processed and how severe is the imbalance during each timestep. As future work, we plan to compare our chosen metrics with others that may present different results.
- Hideo Aochi, Ariane Ducellier, Fabrice Dupros, Mickael Delatre, Thomas Ulrich, Florent De Martin, and Masayuki Yoshimi. 2011. Finite Difference Simulations of Seismic Wave Propagation for the 2007 Mw 6.6 Niigata-ken Chuetsu-Oki Earthquake: Validity of Models and Reliable Input Ground Motion in the Near-Field. Pure and Applied Geophysics 170 (12 2011), 43–64. https://doi.org/10.1007/s00024-011-0429-5Google Scholar
- Hideo Aochi, Ariane Ducellier, Fabrice Dupros, Monique Terrier, and Jérôme Lambert. 2011. Investigation of historical earthquake by seismic wave propagation simulation: Source parameters of the 1887 M6.3 Ligurian, north-western Italy, earthquake. In 8ème colloque AFPS, Vers une maitrise durable du risque sismique. Association Française du Genie Parasismique, Champs-sur-Marne, France, 6 p.https://hal-brgm.archives-ouvertes.fr/hal-00610498Google Scholar
- F. Z. Boito, J. L. Bez, F. Dupros, M. A. R. Dantas, P. O. A. Navaux, and H. Aochi. 2017. High Performance I/O for Seismic Wave Propagation Simulations. In 2017 25th Euromicro Intl. Conf. on Par., Distr. and Network-based Proc. (PDP). IEEE, St. Petersburg, Russia, 31–38. https://doi.org/10.1109/PDP.2017.92Google Scholar
- Luiz DeRose, Bill Homer, and Dean Johnson. 2007. Detecting Application Load Imbalance on High End Massively Parallel Systems. In Proceedings of the 13th Intl. Euro-Par Conf. on Par. Proc.(Euro-Par’07). Springer-Verlag, Berlin, Heidelberg, 150–159. https://doi.org/10.1007/978-3-540-74466-5_17Google ScholarDigital Library
- Fabrice Dupros, Hideo Aochi, Ariane Ducellier, Dimitri Komatitsch, and Jean Roman. 2008. Exploiting Intensive Multithreading for the Efficient Simulation of 3D Seismic Wave Propagation. In ICCSE 2008 : 11th International Conference on Computational Science and Engineering. Computer Society Press, Sao Paulo, Brazil, pp.253–260. https://hal-brgm.archives-ouvertes.fr/hal-01020574Google ScholarDigital Library
- Fabrice Dupros, Florent De Martin, Evelyne Foerster, Dimitri Komatitsch, and Jean Roman. 2010. High-performance finite-element simulations of seismic wave propagation in three-dimensional nonlinear inelastic geological media. Parallel Comput. 36, 5 (2010), 308 – 325. https://doi.org/10.1016/j.parco.2009.12.011 Parallel Matrix Algorithms and Applications.Google ScholarDigital Library
- Kevin A. Huck and Jesus Labarta. 2010. Detailed Load Balance Analysis of Large Scale Parallel Applications. In Proceedings of the 2010 39th International Conference on Parallel Processing(ICPP ’10). IEEE Computer Society, USA, 535–544. https://doi.org/10.1109/ICPP.2010.61Google Scholar
- Jacques Kergommeaux, Benhur Stein, and Pierre-Eric Bernard. 2000. Pajé, an interactive visualization tool for tuning multi-threaded parallel applications. Par. Comput. 26(2000), 1253–1274. https://doi.org/10.1016/S0167-8191(00)00010-7Google ScholarDigital Library
- Yann Klinger, Chen Ji, Zheng-Kang Shen, and W. Bakun. 2010. Introduction to the Special Issue on the 2008 Wenchuan, China, Earthquake. Bull. of The Seismological Soc. of America 100 (10 2010), 2353–2356. https://doi.org/10.1785/0120100172Google Scholar
- Andreas Knüpfer, Holger Brunst, Jens Doleschal, Matthias Jurenz, Matthias Lieber, Holger Mickler, Matthias S Müller, and Wolfgang E Nagel. 2008. The vampir performance analysis tool-set. In Tools for High Perf. Comp.Springer, Germany, 139–155. https://doi.org/10.1007/978-3-540-68564-7_9Google Scholar
- Andreas Knüpfer, Christian Feld, Dieter Mey, Scott Biersdorff, Kai Diethelm, Dominic Eschweiler, Markus Geimer, Michael Gerndt, Daniel Lorenz, Allen Malony, Wolfgang Nagel, Yury Oleynik, Peter Philippen, Pavel Saviankou, Dirk Schmidl, Sameer Shende, Ronny Tschüter, Michael Wagner, Bert Wesarg, and Felix Wolf. 2012. Score-P: A Joint Performance Measurement Run-Time Infrastructure for Periscope, Scalasca, TAU, and Vampir. Springer, Berlin, Heidelberg, 79–91. https://doi.org/10.1007/978-3-642-31476-6_7Google Scholar
- Olga Pearce, Todd Gamblin, Bronis de Supinski, Martin Schulz, and Nancy Amato. 2012. Quantifying the Effectiveness of Load Balance Algorithms. In Proceedings of the 26th ACM Intl. Conf. on Supercomputing(ICS ’12). ACM, New York, NY, USA, 185–194. https://doi.org/10.1145/2304576.2304601Google ScholarDigital Library
- Vincent Pillet, Vincent Pillet, Jesús Labarta, Toni Cortes, Toni Cortes, Sergi Girona, Sergi Girona, and Departament D’arquitectura De Computadors. 1995. Paraver: A Tool to Visualize and Analyze Parallel Code. Technical Report. In WoTUG-18.Google Scholar
- Flavio Rodrigues. 2016. Study of load distribution measures for high-performance applications. Master’s thesis. UFRGS. http://hdl.handle.net/10183/149593Google Scholar
- Rafael Tesser, Lucas Schnorr, Arnaud Legrand, Fabrice Dupros, and Philippe O A Navaux. 2017. Using Simulation to Evaluate and Tune the Performance of Dynamic Load Balancing of an Over-decomposed Geophysics Application. In Euro-Par 2017: 23rd Intl. European Conf. on Par. and Distr. Comput.Springer, Santiago de Compostela, Spain, 15. https://hal.inria.fr/hal-01567792Google Scholar
- Rafael Tesser, Lucas Schnorr, Arnaud Legrand, Franz Heinrich, Fabrice Dupros, and Philippe Navaux. 2018. Performance Modeling of a Geophysics Application to Accelerate the Tuning of Over-decomposition Parameters through Simulation. Concurr. Comput. Pract. Exp. 31, 11 (2018), e5012. https://doi.org/10.1002/cpe.5012Google ScholarCross Ref
- Rafael Tesser, Lucas Schnorr, Philippe Navaux, and Arnaud Legrand. 2016. Using SimGrid to Evaluate the Impact of AMPI Load Balancing In a Geophysics HPC Application. https://doi.org/10.13140/RG.2.1.2417.2407Google Scholar
- R. K. Tesser, L. L. Pilla, F. Dupros, P. O. A. Navaux, J. Méhaut, and C. Mendes. 2014. Improving the Performance of Seismic Wave Simulations with Dynamic Load Balancing. In 2014 22nd Euromicro Intl. Conf. on Par., Distr. and Network-based Proc. (PDP). IEEE, Torino, Italy, 196–203. https://doi.org/10.1109/PDP.2014.37Google Scholar
Index Terms
- Using Visualization of Performance Data to Investigate Load Imbalance of a Geophysics Parallel Application
Recommendations
Quantifying the effectiveness of load balance algorithms
ICS '12: Proceedings of the 26th ACM international conference on SupercomputingLoad balance is critical for performance in large parallel applications. An imbalance on today's fastest supercomputers can force hundreds of thousands of cores to idle, and on future exascale machines this cost will increase by over a factor of a ...
Comments