Abstract
Future exascale supercomputers will be composed of thousands of nodes. In those massive systems, the search for physically close nodes will become essential to deliver an optimal environment to execute parallel applications. Schedulers manage those resources, shared by many users and jobs, searching for partitions in which jobs will run. Significant effort has been devoted to develop allocation strategies that maximize system utilization, while providing partitions that are adequate for the communication demands of applications. In this paper we evaluate a class of strategies based on space-filling curves (SFCs) that search for partitions in which nodes are physically close, compared to other alternatives that relax this requirement (e.g. non-contiguous), or make it even more strict (e.g. contiguous). Several metrics are used to assess the quality of an allocation strategy, some based on system utilization, some others measuring the quality of the resulting partitions. Contiguous allocators suffer from severe degradation in terms of system utilization, while non-contiguous allocators provide inadequate partitions. Somewhere in the middle, SFC allocators offer good system utilization while using quite compact partitions. The final metric to decide which allocator is the best depend on the severity of the slowdown suffered by applications when running in non-optimal partitions.
J.A. Pascual is currently with the APT group in The University of Manchester.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Alber, J., Niedermeier, R.: On multi-dimensional hilbert indexings. Theory Comput. Syst. 33, 195–392 (2000)
Alber, J., Niedermeier, R.: On multidimensional curves with hilbert property. Theory Comput. Syst. 33(4), 295–312 (2000)
Balzuweit, E., Bunde, D.P., Leung, V.J., Finley, A., Lee, A.C.S.: Local search to improve coordinate-based task mapping. Parallel Comput. 51, 67–78 (2016)
Chen, D., Eisley, N.A., Heidelberger, P., Senger, R.M., Sugawara, Y., Kumar, S., Salapura, V., Satterfield, D.L., Steinmacher-Burow, B., Parker, J.J.: The IBM Blue Gene/Q interconnection network and message unit. In: Proceedings of 2011 International Conference for High Performance Computing, Networking, Storage and Analysis, pp. 1–10. ACM, New York (2011)
Cray Inc. http://www.cray.com/assets/pdf/products/xe/idc_948.pdf
Feitelson, D.G., Rudolph, L., Schwiegelshohn, U.: Parallel job scheduling — a status report. In: Feitelson, D.G., Rudolph, L., Schwiegelshohn, U. (eds.) JSSPP 2004. LNCS, vol. 3277, pp. 1–16. Springer, Heidelberg (2005). doi:10.1007/11407522_1
Hilbert, D.: Über die stetige abbildung einer linie auf ein flächenstück. Ann. Math. 38, 459–460 (1891)
IBM Journal of Research, Development staff. Overview of the IBM Blue Gene/P project. IBM Journal of Research and Development 52(1/2), 199–220
Johnson, C.R., Bunde, D.P., Leung, V.J.: A tie-breaking strategy for processor allocation in meshes. In: 39th International Conference on Parallel Processing, ICPP. Workshops 2010, San Diego, California, USA, pp. 331–338. IEEE Computer Society, 13–16 September 2010
Jokanovic, A., Sancho, J., Rodriguez, G., Lucero, A., Minkenberg, C., Labarta, J.: Quiet neighborhoods: key to protect job performance predictability. In: Parallel and Distributed Processing Symposium (IPDPS), pp. 449–459, May 2015
Lawder, J.K., King, P.J.H.: Using space-filling curves for multi-dimensional indexing. In: Lings, B., Jeffery, K. (eds.) BNCOD 2000. LNCS, vol. 1832, pp. 20–35. Springer, Heidelberg (2000). doi:10.1007/3-540-45033-5_3
Lawrence Livermore National Laboratory. Simple linux utility for resource management. https://computing.llnl.gov/linux/slurm/
Leung, V., Arkin, E., Bender, M., Bunde, D., Johnston, J., Lal, A., Mitchell, J., Phillips, C., Seiden, S.: Processor allocation on cplant: achieving general processor locality using one-dimensional allocation strategies. pp. 296–304 (2002)
Li, K., Malawski, M., Nabrzyski, J.: Topology-aware scheduling on blue waters with proactive queue scanning and migration-based job placement. In: 20th Workshop on Job Scheduling Strategies for Parallel Processing (JSSPP) (2016)
Liu, W., Lo, V., Windisch, K., Nitzberg, B.: Non-contiguous processor allocation algorithms for distributed memory multicomputers. In: Proceedings of ACM/IEEE Conference on Supercomputing, pp. 227–236. IEEE Computer (1994)
Lo, V., Windisch, K., Liu, W., Nitzberg, B.: Noncontiguous processor allocation algorithms for mesh-connected multicomputers. IEEE Trans. Parallel Distrib. Syst. 8, 712–726 (1997)
Morton. A computer oriented geodetic data base and a new technique in file sequencing. Technical report, Ottawa, Ontario, Canada (1966)
Navaridas, J., Pascual, J.A., Miguel-Alonso, J.: Effects of job and task placement on the performance of parallel scientific applications. In: Proceedings 17th Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, pp. 55–61. IEEE Computer Society, February 2009
Pascual, J.A., Miguel-Alonso, J., Lozano, J.A.: Strategies to map parallel applications onto meshes. In: de Leon, F., et al. (eds.) Distributed Computing and Artificial Intelligence. AISC, vol. 79, pp. 197–204. Springer, Heidelberg (2010). doi:10.1007/978-3-642-14883-5_26
Pascual, J.A., Miguel-Alonso, J., Lozano, J.A.: Optimization-based mapping framework for parallel applications. J. Parallel Distrib. Comput. 71(10), 1377–1387 (2011)
Pascual, J.A., Miguel-Alonso, J., Lozano, J.A.: Application-aware metrics for partition selection in cube-shaped topologies. Parallel Comput. 40(5–6), 129–139 (2014)
Pascual, J.A., Miguel-Alonso, J., Lozano, J.A.: A fast implementation of the first fit contiguous partitioning strategy for cubic topologies. Concurrency Comput. Pract. Experience 26(17), 2792–2810 (2014)
Pascual, J.A., Miguel-Alonso, J., Lozano, J.A.: Locality-aware policies to improve job scheduling on 3D tori. J. Supercomputing 71(3), 966–994 (2015)
Walker, P., Bunde, D.P., Leung, V.J.: Faster high-quality processor allocation. In: Proceedings of the 11th LCI International Conference on High-Performance Cluster Computing (2010)
Weisstein, E.W.: Hilbert Curve. http://mathworld.wolfram.com/HilbertCurve.html
Windisch, K., Lo, V., Bose, B.: Contiguous and non-contiguous processor allocation algorithms for k-ary n-cubes. IEEE Trans. Parallel Distrib. Syst. 8, 712–726 (1995)
Acknowledgments
This work has been partially supported by the Research Groups 2013–2018 (IT-609-13) program (Basque Government), TIN2013-41272P (Ministry of Science and Technology). Jose A. Lozano is also supported by BERC program 2014–2017 (Basque government) and Severo Ochoa Program SEV-2013-0323 (Spanish Ministry of Economy and Competitiveness). Jose Miguel-Alonso is member of the HiPEAC European Network.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this paper
Cite this paper
Pascual, J.A., Lozano, J.A., Miguel-Alonso, J. (2017). Analyzing the Performance of Allocation Strategies Based on Space-Filling Curves. In: Desai, N., Cirne, W. (eds) Job Scheduling Strategies for Parallel Processing. JSSPP JSSPP 2015 2016. Lecture Notes in Computer Science(), vol 10353. Springer, Cham. https://doi.org/10.1007/978-3-319-61756-5_13
Download citation
DOI: https://doi.org/10.1007/978-3-319-61756-5_13
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-61755-8
Online ISBN: 978-3-319-61756-5
eBook Packages: Computer ScienceComputer Science (R0)