The k-ary n-direct s-indirect family of topologies for large-scale interconnection networks

The final publication is available at Springer via http://dx.doi.org/10.1007/s11227-016-1640-zIn large-scale supercomputers, the interconnection network plays a key role in system performance. Network topology highly defines the performance and cost of the interconnection network. Direct topologies are sometimes used due to its reduced hardware cost, but the number of network dimensions is limited by the physical 3D space, which leads to an increase of the communication latency and a reduction of network throughput for large machines. Indirect topologies can provide better performance for large machines, but at higher hardware cost. In this paper, we propose a new family of hybrid topologies, the k-ary n-direct s-indirect, that combines the best features from both direct and indirect topologies to efficiently connect an extremely high number of processing nodes. The proposed network is an n-dimensional topology where the k nodes of each dimension are connected through a small indirect topology of s stages. This combination results in a family of topologies that provides high performance, with latency and throughput figures of merit close to indirect topologies, but at a lower hardware cost. In particular, it doubles the throughput obtained per cost unit compared with indirect topologies in most of the cases. Moreover, their fault-tolerance degree is similar to the one achieved by direct topologies built with switches with the same number of ports.This work was supported by the Spanish Ministerio de Economa y Competitividad (MINECO) and by FEDER funds under Grant TIN2012-38341-C04-01 and by Programa de Ayudas de Investigacion y Desarrollo (PAID) from Universitat Politecnica de Valencia.Peñaranda Cebrián, R.; Gómez Requena, C.; Gómez Requena, ME.; López Rodríguez, PJ.; Duato Marín, JF. (2016). The k-ary n-direct s-indirect family of topologies for large-scale interconnection networks. Journal of Supercomputing. 72(3):1035-1062. https://doi.org/10.1007/s11227-016-1640-z10351062723Connect-IB. http://www.mellanox.com/related-docs/prod_adapter_cards/PB_Connect-IB.pdf . Accessed 3 Feb 2016Mellanox store. http://www.mellanoxstore.com . Accessed 3 Feb 2016Mellanox technology. http://www.mellanox.com . Accessed 3 Feb 2016Myricom. http://www.myri.com . Accessed 3 Feb 2016Quadrics homepage. http://www.quadrics.com . Accessed 22 Sept 2008TOP500 supercomputer site. http://www.top500.org . Accessed 3 Feb 2016Balkan A, Qu G, Vishkin U (2009) Mesh-of-trees and alternative interconnection networks for single-chip parallelism. IEEE Trans Very Large Scale Integr(VLSI) Syst 17(10):1419–1432. doi: 10.1109/TVLSI.2008.2003999Bermudez Garzon D, Gomez ME, Lopez P, Duato J, Gomez C (2014) FT-RUFT: a performance and fault-tolerant efficient indirect topology. In: 22nd Euromicro international conference on parallel, distributed and network-based processing (PDP). IEEE, pp 405–409Bhandarkar SM, Arabnia HR (1995) The Hough transform on a reconfigurable multi-ring network. J Parallel Distrib Comput 24(1):107–114Boku T, Nakazawa K, Nakamura H, Sone T, Mishima T, Itakura K (1996) Adaptive routing technique on hypercrossbar network and its evaluation. Syst Comput Jpn 27(4):55–64Dally W, Towles B (2004) Principles and practices of interconnection networks. Morgan Kaufmann, San FranciscoDas R, Eachempati S, Mishra A, Narayanan V, Das C (2009) Design and evaluation of a hierarchical on-chip interconnect for next-generation CMPs. In: IEEE 15th international symposium on high performance computer architecture (HPCA’09), pp 175–186. doi: 10.1109/HPCA.2009.4798252Mahdaly AI, Mouftah HT, Hanna NN (1990) Topological properties of WK-recursive networks. In: Proceedings of IEEE workshop on future trends of distributed computing systems, pp 374–380. doi: 10.1109/FTDCS.1990.138349Duato J (1996) A necessary and sufficient condition for deadlock-free routing in cut-through and store-and-forward networks. IEEE Trans Parallel Distrib Syst 7:841–854. doi: 10.1109/71.532115Duato J, Yalamanchili S, Lionel N (2002) Interconnection networks: an engineering approach. Morgan Kaufmann Publishers Inc., USAFlich J, Malumbres M, López P, Duato J (2000) Improving routing performance in Myrinet networks. In: International on parallel and distributed processing symposium, p 27. doi: 10.1109/IPDPS.2000.845961García M, Beivide R, Camarero C, Valero M, Rodríguez G, Minkenberg C (2015) On-the-fly adaptive routing for dragonfly interconnection networks. J Supercomput 71(3):1116–1142Gómez C, Gilabert F, Gómez M, López P, Duato J (2007) Deterministic versus adaptive routing in fat-trees. In: IEEE international on parallel and distributed processing symposium (IPDPS’07), pp 1–8. doi: 10.1109/IPDPS.2007.370482Gómez C, Gilabert F, Gómez M, López P, Duato J (2008) RUFT: simplifying the fat-tree topology. In: 14th IEEE international conference on parallel and distributed systems (ICPADS’08), pp 153–160. doi: 10.1109/ICPADS.2008.44Guo C, Lu G, Li D, Wu H, Zhang X, Shi Y, Tian C, Zhang Y, Lu S (2009) BCube: a high performance, server-centric network architecture for modular data centers. In: SIGCOMM ’09: proceedings of the ACM SIGCOMM 2009 conference on data communication. ACM, New York, pp 63–74. doi: 10.1145/1592568.1592577 . http://www.bibsonomy.org/bibtex/23a5da89fbf099e3c70f4559ab38082c5/chesteve . Accessed 22 Sept 2008Gupta A, Dally W (2006) Topology optimization of interconnection networks. Comput Arch Lett 5(1):10–13. doi: 10.1109/L-CA.2006.8Kim J, Dally W, Abts D (2007) Flattened butterfly: a cost-efficient topology for high-radix networks. In: Proceedings of the 34th annual international symposium on computer architecture (ISCA’07). ACM, New York, pp 126–137. doi: 10.1145/1250662.1250679Kim J, Dally W, Scott S, Abts D (2008) Technology-driven, highly-scalable dragonfly topology. In: Proceedings of the 35th annual international symposium on computer architecture (ISCA’08). IEEE Computer Society, Washington, DC, pp 77–88. doi: 10.1109/ISCA.2008.19Leighton F (1992) Introduction to parallel algorithms and architectures: arrays, trees, hypercubes v. 1. M. Kaufmann Publishers, San FranciscoLeiserson CE (1985) Fat-trees: universal networks for hardware-efficient supercomputing. IEEE Trans Comput 34(10):892–901Matsutani H, Koibuchi M, Amano H (2007) Performance, cost, and energy evaluation of fat H-tree: a cost-efficient tree-based on-chip network. In: IEEE international on parallel and distributed processing symposium (IPDPS’07), pp 1–10. doi: 10.1109/IPDPS.2007.370271Rahmati D, Kiasari A, Hessabi S, Sarbazi-Azad H (2006) A performance and power analysis of wk-recursive and mesh networks for network-on-chips. In: International conference on computer design (ICCD’06), pp 142–147. doi: 10.1109/ICCD.2006.4380807Towles B, Dally WJ (2002) Worst-case traffic for oblivious routing functions. In: Proceedings of the fourteenth annual ACM symposium on parallel algorithms and architectures (SPAA’02). ACM, New York, pp 1–8. doi: 10.1145/564870.564872Yang Y, Funahashi A, Jouraku A, Nishi H, Amano H, Sueyoshi T (2001) Recursive diagonal torus: an interconnection network for massively parallel computers. IEEE Trans Parallel Distrib Syst 12(7):701–715. doi: 10.1109/71.94074