Skip to main content

Deterministic Construction of Regular Geometric Graphs with Short Average Distance and Limited Edge Length

  • Conference paper
  • First Online:
Algorithms and Architectures for Parallel Processing (ICA3PP 2016)

Part of the book series: Lecture Notes in Computer Science ((LNTCS,volume 10048))

  • 1724 Accesses

Abstract

This paper proposes a deterministic method to construct 5-regular geometric graphs with short average distance under the constraint such that the set of vertices is a subset of \(\mathbb {N}\times \mathbb {N}\) and the length of each edge is at most 4. This problem is motivated by the design of efficient floor plan of parallel computers consisting of a number of computing nodes arranged on a two-dimensional array. In such systems, the degree of vertices is determined by the number of ports of the routers and the edge length is limited by a certain value determined by the cycle time. The goodness of the resulting geometric graph is evaluated by the average shortest path length (ASPL) between vertices which reflects the average communication delay between computing nodes. The idea of the proposed method is to arrange the basic component derived from (3, g)-cage in a two-dimensional manner and to connect adjacent components by parallel edges of length 4 each. The result of numerical calculations shows that the average distance in the resulting graph is close to the lower bound so that the gap to the lower bound is less than 0.98 when the number of vertices is 432000.

This work was partially supported by MIC/SCOPE (152103004), JSPS KAKENHI (15K00144, 16H02807) and JST CREST.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 39.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 54.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

References

  1. Ajima, Y., Sumimoto, S., Shimizu, T.: Tofu: a 6D Mesh/Torus interconnect for exascale computers. IEEE Comput. 42(11), 36–40 (2009)

    Article  Google Scholar 

  2. Battista, G.D., Eades, P., Tamassia, R., Tollis, I.G.: Algorithms for drawing graphs: an annotated bibliography. Comput. Geom. Theor. Appl. 4(5), 235–282 (1994)

    Article  MathSciNet  MATH  Google Scholar 

  3. Besta, M., Hoefler, T.: Slim fly: a cost effective low-diameter network topology. In: SC 2014, pp. 348–359 (2014)

    Google Scholar 

  4. Cerf, V.G., Cowan, D.D., Mullin, R.C., Stanton, R.G.: A lower bound on the average shortest path length in regular graphs. Netw. 4(4), 335–342 (1974)

    Article  MathSciNet  MATH  Google Scholar 

  5. Chaix, F., Koibuchi, M., Fujiwara, I.: Suitability of the random topology for HPC applications. In: Proceedings 24th Euromicro Int’l Conference on Parallel, Distributed, and Network-Based Processing (PDP 2016), pp. 301–304 (2016)

    Google Scholar 

  6. Damerell, R.M.: On Moore graphs. Proc. Cambridge Phil. Soc. 74, 227–236 (1973)

    Article  MathSciNet  MATH  Google Scholar 

  7. Erdös, P., Sachs, H.: Regülare graphen gegebener taillenweite mitminimaler knotenzahl. Wiss. Z. Uni. Halle (Math. Nat.) 12, 251–257 (1963)

    Google Scholar 

  8. Fujiwara, I., Koibuchi, M., Matsutani, H., Casanova, H.: Swap-and-randomize: a method for building low-latency HPC interconnects. IEEE Trans. Parallel Distrib. Syst. 26(7), 2051–2060 (2015)

    Article  Google Scholar 

  9. Higham, D.J., Rasajski, M., Przulj, N.: Fitting a geometric graph to a protein-protein interaction network. Bioinform. 24(8), 1093–1099 (2008)

    Article  Google Scholar 

  10. Hoffman, A.J., Singleton, R.R.: On Moore graphs with diameter 2 and 3. IBM J. Res. Develop. 4, 497–504 (1960)

    Article  MathSciNet  MATH  Google Scholar 

  11. Kim, J., Dally, W.J., Scott, S., Abts, D.: Technology-driven, highly-scalable Dragonfly topology. In: Proceedings of the 35th International Symposium on Computer Architecture (ISCA), pp. 77–88 (2008)

    Google Scholar 

  12. Koibuchi, M., Matsutani, H., Amano, H., Hsu, D.F., Casanova, H.: A case for random shortcut topologies for HPC interconnects. In: Proceedings of the 39th International Symposium on Computer Architecture (ISCA), pp. 177–188 (2012)

    Google Scholar 

  13. Mellanox Technologies. IS5024, Mellanox Technologies. http://www.mellanox.com/related-docs/user_manuals/IS5024_User_Manual.pdf

  14. Mellanox Technologies. http://www.mellanox.com/page/cables

  15. Miller, M., Širáň, J.: Moore graphs and beyond: a survey of the degree/diameter problem. Electron. J. Comb., No. DS14 (2005)

    Google Scholar 

  16. Nakano, K., Takafuji, D., Fujita, S., Matsutani, H., Fujiwara, I., Koibuchi, M.: Randomly optimized grid graph for low-latency interconnection networks. In: Proceedings International Conference on Parallel Processing (ICPP) (2016, to appear)

    Google Scholar 

  17. Pinheiro, M.A., Kybic, J.: Path descriptors for geometric graph matching and registration. In: Campilho, A., Kamel, M. (eds.) ICIAR 2014. LNCS, vol. 8814, pp. 3–11. Springer, Heidelberg (2014). doi:10.1007/978-3-319-11758-4_1

    Google Scholar 

  18. Hemmert, K.S., Vetter, J.S., Bergman, K., Das, C., Emami, A., Janssen, C., Panda, D.K., Stunkel, C., Underwood, K., Yalamanchili, S.: Report on institute for advanced architectures and algorithms. In: Proceedings Interconnection Networks Workshop (2008)

    Google Scholar 

  19. Tomkins, J.: Interconnects: a buyers point of view. In: Proceedings ACS Workshop (2007)

    Google Scholar 

  20. Towles, B., Grossman, J.P., Greskamp, B., Shaw, D.E.: Unifying on-chip and inter-node switching within the Anton 2 network. In: Proceedings of the 41st International Symposium on Computer Architecture (ISCA), pp. 1–12 (2014)

    Google Scholar 

  21. SimGrid: Versatile simulation of distributed systems. http://simgrid.gforge.inria.fr/

  22. Yang, H., Tripathi, J., Jerger, N.E., Gibson, D.: Dodec: random-link, low-radix on-chip networks. In: MICRO, pp. 496–508 (2014)

    Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Satoshi Fujita .

Editor information

Editors and Affiliations

A Application of the Result Graph to Off- and On-chip Interconnection Networks

A Application of the Result Graph to Off- and On-chip Interconnection Networks

Recent interconnection networks potentially have tight constraints on link length especially when conventional electric media is used. Short-cable design is still preferred in various interconnection networks, such as supercomputers [1, 20]. This constraint efficiently reduces power consumption of links, thus giving a significant impact on the design of interconnection networks. For example, when 40 Gbps passive electric cables whose length can be up to 7 m are used, the Mellanox IS5024 36-port InfiniBand switch saves the power consumption to 111.54 W, while active optical cables impose the same switch for 200.4 W [13, 14]. Since a supercomputer has a large number of cables (remind that the K computer has 200,000 cables reaching 1,000 km), the network design under cable geometric constraints becomes important for low-power interconnection network design.

The link length limitation is also a severe issue in on-chip interconnection network designs, especially with an advanced CMOS process technology. In general, a wire delay increases with the square of the wire length. Inserting repeater buffers on the wire can mitigate the wire delay, while it increases the energy consumption on the wire. Thus the longest wire length between on-chip routers should be carefully tuned under given design constraints, such as operating frequency, dimensions of cores, wire delay, and wire energy. However, it is most likely to be automatically determined by underlying regular topologies (e.g., 2-D mesh and torus). As an alternative to such regular topologies, a low-latency topology optimized for a given longest wire length parameter would become an attractive design option when the design constraints severely affect the cost and the performance.

The above fact motivates us to explore the use of short links in the off- and on-chip network topology design. This is just the application of the result graph in this study to such interconnection networks.

Rights and permissions

Reprints and permissions

Copyright information

© 2016 Springer International Publishing AG

About this paper

Cite this paper

Fujita, S., Nakano, K., Koibuchi, M., Fujiwara, I. (2016). Deterministic Construction of Regular Geometric Graphs with Short Average Distance and Limited Edge Length. In: Carretero, J., Garcia-Blas, J., Ko, R., Mueller, P., Nakano, K. (eds) Algorithms and Architectures for Parallel Processing. ICA3PP 2016. Lecture Notes in Computer Science(), vol 10048. Springer, Cham. https://doi.org/10.1007/978-3-319-49583-5_23

Download citation

  • DOI: https://doi.org/10.1007/978-3-319-49583-5_23

  • Published:

  • Publisher Name: Springer, Cham

  • Print ISBN: 978-3-319-49582-8

  • Online ISBN: 978-3-319-49583-5

  • eBook Packages: Computer ScienceComputer Science (R0)

Publish with us

Policies and ethics