Design Space Exploration of the Dragonfly Topology

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10524)


We investigate possible options of creating a Dragonfly topology capable of accommodating a specified number of end-points. We first observe that any Dragonfly topology can be described with two main parameters, imbalance and density, dictating the distribution of routers in groups, and the inter-group connectivity, respectively. We then introduce an algorithm that generates a dragonfly topology by taking the desired number of end-points and these two parameters as input. We calculate a variety of metrics on the generated topologies resulting from a large set of parameter combinations. Based on these metrics, we isolate the subset of topologies that present the best economical and performance trade-off. We conclude by summarizing guidelines for Dragonfly topology design and dimensioning.


Topologies Dragonfly Optical interconnects 


  1. 1.
    Kim, J., Dally, W.J., Scott, S., Abts, D.: Technology-driven, highly-scalable dragonfly topology. In: 2008 International Symposium on Computer Architecture, pp. 77–88, June 2008Google Scholar
  2. 2.
    Kim, J., Dally, W., Abts, D.: Flattened butterfly: a cost-efficient topology for high-radix networks. In: Proceedings of the 34th Annual International Symposium on Computer Architecture, ISCA 2007, New York, NY, USA, pp. 126–137 (2007)Google Scholar
  3. 3.
    Alverson, B., Froese, E., Kaplan, L., Roweth, D.: Cray XC series network (2012),
  4. 4.
    Faanes, G., Bataineh, A., Roweth, D., Court, T., Froese, E., Alverson, B., Johnson, T., Kopnick, J., Higgins, M., Reinhard, J.: Cray cascade: a scalable HPC system based on a dragonfly network. In: Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis, SC 2012, Los Alamitos, CA, USA, pp. 103:1–103:9. IEEE Computer Society Press (2012)Google Scholar
  5. 5.
    Bhatele, A., Jain, N., Livnat, Y., Pascucci, V., Bremer, P.T.: Analyzing network health and congestion in dragonfly-based supercomputers. In: 2016 IEEE International Parallel and Distributed Processing Symposium (IPDPS), pp. 93–102, May 2016Google Scholar
  6. 6.
    Jain, N., Bhatele, A., Ni, X., Wright, N.J., Kale, L.V.: Maximizing throughput on a dragonfly network. In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2014, Piscataway, NJ, USA, pp. 336–347. IEEE Press (2014)Google Scholar
  7. 7.
    Wen, K., Samadi, P., Rumley, S., Chen, C.P., Shen, Y., Bahadroi, M., Bergman, K., Wilke, J.: Flexfly: enabling a reconfigurable dragonfly through silicon photonics. In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2016, Piscataway, NJ, USA, pp. 15:1–15:12. IEEE Press (2016)Google Scholar
  8. 8.
    Camarero, C., Vallejo, E., Beivide, R.: Topological characterization of hamming and dragonfly networks and its implications on routing. ACM Trans. Archit. Code Optim. 11, 39:1–39:25 (2014)Google Scholar
  9. 9.
    Rumley, S., Glick, M., Hammond, S.D., Rodrigues, A., Bergman, K.: Design methodology for optimizing optical interconnection networks in high performance systems. In: Kunkel, J.M., Ludwig, T. (eds.) ISC High Performance 2015. LNCS, vol. 9137, pp. 454–471. Springer, Cham (2015). doi: 10.1007/978-3-319-20119-1_32 CrossRefGoogle Scholar
  10. 10. Accessed 16 Apr 2017
  11. 11.
    Hastings, E., Rincon-Cruz, D., Spehlmann, M., Meyers, S., Bunde, D.P., Leung, V.J.: Comparing global link arrangements for dragonfly networks. In: 2015 IEEE International Conference on Cluster Computing, Chicago, IL, USA, pp. 361–370 (2015)Google Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.Lightwave Research LaboratoryColumbia UniversityNew YorkUSA
  2. 2.Scalable Modeling and AnalysisSandia National LabsLivermoreUSA

Personalised recommendations