Advertisement

Extending Distributed Shared Memory for the Cell Broadband Engine to a Channel Model

  • Kenneth Skovhede
  • Morten N. Larsen
  • Brian Vinter
Part of the Lecture Notes in Computer Science book series (LNCS, volume 7133)

Abstract

As the performance gains from traditional processors decline, alternative processor designs are becoming available. One such processor is the CELL-BE processor, which theoretically can deliver a sustained performance close to 205 GFLOPS per processor[10]. Unfortunately, the high performance comes at the price of a quite complex programming model. In this paper we present an easy-to-use, CSP-like, communication method, which enables transfers of shared memory objects. The channel based communication method can significantly reduce the complexity of massively parallel programs. By implementing a few scientific computational cores we show that performance and scalability of the system is acceptable for most problems.

Keywords

CSP CELL-BE DSMCBE channel communication 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Backus, J.: Can programming be liberated from the von Neumann style?: A functional style and its algebra of programs. Commun. ACM 21(8), 613–641 (1978)MathSciNetCrossRefzbMATHGoogle Scholar
  2. 2.
    Beltran, V., Carrera, D., Torres, J., Ayguade, E.: CellMT: A Cooperative Multithreading Library for the Cell/B.E. In: The Proceedings of the 16th Annual IEEE International Conference on High Performance Computing, HiPC 2009 (December 2009)Google Scholar
  3. 3.
    Bjørndalen, J.M., Vinter, B., Anshus, O.: PyCSP - Communicating Sequential Processes for PythonGoogle Scholar
  4. 4.
    Brown, N., Welch, P.: An Introduction to the Kent C++CSP Library. SlidesGoogle Scholar
  5. 5.
    Chen, T.: Cell Broadband Engine Architecture and its first implementation - A Performance View (2005), http://www.ibm.com/developerworks/power/library/pa-cellperf/ (accessed July 26, 2010)
  6. 6.
    Hoare, C.A.R.: Communicating sequential processes. Communications of the ACM 21(8), 666–677 (1978)CrossRefzbMATHGoogle Scholar
  7. 7.
    Hoare, C.A.R.: Communicating Sequential Processes. Prentice-Hall (1985)Google Scholar
  8. 8.
    IBM: IBM doubles down on cell blade (2007), http://www-03.ibm.com/press/us/en/pressrelease/22258.wss (accessed July 26, 2010)
  9. 9.
    Jowkar, M.: Exploring the Potential of the Cell Processor for High Performance Computing (2007), http://www.diku.dk/~rehr/cell/docs/mohammad_jowkar_thesis.pdf (accessed July 26, 2010)
  10. 10.
    Langou, J., Langou, J., Luszczek, P., Kurzak, J., Buttari, A., Dongarra, J.: Exploiting the Performance of 32 bit Floating Point Arithmetic in Obtaining 64 bit Accuracy (2006), http://www.netlib.org/lapack/lawnspdf/lawn175.pdf (accessed March 29, 2010)
  11. 11.
    Larsen, M.N., Skovhede, K., Vinter, B.: Distributed Shared Memory for the Cell Broadband Engine (DSMCBE). In: International Symposium on Parallel and Distributed Computing, pp. 121–124 (2009)Google Scholar
  12. 12.
    Mcewan, A.A., Schneider, S., Ifill, W., Welch, P., Brown, N.: C++CSP2: A Many-to-Many Threading Model for Multicore Architectures (2007)Google Scholar
  13. 13.
    Rehr, M.: Application Porting and Tuning on the Cell-BE Processor (2008), http://dk.migrid.org/public/doc/published_papers/nqueens.pdf (accessed July 26, 2010)
  14. 14.
    Schaller, N.C., Hilderink, G.H., Welch, P.H.: Using Java for Parallel Computing - JCSP versus CTJ. In: Communicating Process Architectures 2000, pp. 205–226 (2000)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • Kenneth Skovhede
    • 1
  • Morten N. Larsen
    • 1
  • Brian Vinter
    • 1
  1. 1.Department of Computer ScienceUniversity of CopenhagenCopenhagenDenmark

Personalised recommendations