Abstract
The concurrency galore is currently defining computing at all levels. At the high end, it is leading to clusters composed of hundreds or thousands of nodes, each integrating 10-100s of computing cores, and fed with instructions by up to two orders of magnitudes more threads. Technology constraints prohibit a reversal of this trend, and the still unsatisfied need for more computing power has led to a pervasive use of accelerators to speed up computations. Both in terms of energy and time, communication is more expensive than computation. Being amplified by the advent of Big Data, we are observing a fundamental transition to communication-centric systems composed of heterogeneous computing units. This talk will review current techniques for data movements in clusters. This is followed by an introduction to the Global GPU Address Spaces (GGAS) project, which we use to explore data movement optimizations in heterogeneous environments. The talk will conclude with a summary of synergies regarding related research projects.
Chapter PDF
References
Dally, B.: Gpu computing: To exascale and beyond (keynote). In: International Conference for High Performance Computing, Networking, Storage and Analysis, SC (2013)
Shalf, J., Dosanjh, S., Morrison, J.: Exascale computing technology challenges. In: Palma, J.M.L.M., Daydé, M., Marques, O., Lopes, J.C. (eds.) VECPAR 2010. LNCS, vol. 6449, pp. 1–25. Springer, Heidelberg (2011)
Yalamanchili, S.: Scaling Resource Compositions in a Flatter World. In: Workshop on Unconventional Cluster Architectures and Applications (UCAA), in conjunction with ICPP 2012, Pittsburgh, US (2012)
Valiant, L.G.: A bridging model for parallel computation. Communications of the ACM 33(8), 103–111 (1990)
Oden, L., Fröning, H.: GGAS: Global GPU Address Spaces for Efficient Communication in Heterogeneous Clusters. In: IEEE International Conference on Cluster Computing 2013, Indianapolis, US, September 23-27 (2013)
Oden, L., Klenk, B., Fröning, H.: On Achieving High Message Rates. In: 14th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid), Chicago, IL, US, May 26-29 (2014)
Aggarwal, V., Sabharwal, Y., Garg, R., Heidelberger, P.: HPCC RandomAccess benchmark for next generation supercomputers. In: IEEE International Symposium on Parallel & Distributed Processing (IPDPS 2009), pp. 1–11. IEEE Computer Society, Washington, DC (2009)
Fröning, H., Nüssle, M., Litz, H., Leber, C., Brüning, U.: On Achieving High Message Rates. In: 13th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid), May 13-16, Delft, The Netherlands (2013)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2014 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Fröning, H. (2014). Data Movement Options in Accelerated Clusters. In: an Mey, D., et al. Euro-Par 2013: Parallel Processing Workshops. Euro-Par 2013. Lecture Notes in Computer Science, vol 8374. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-54420-0_41
Download citation
DOI: https://doi.org/10.1007/978-3-642-54420-0_41
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-54419-4
Online ISBN: 978-3-642-54420-0
eBook Packages: Computer ScienceComputer Science (R0)