Advertisement

The Square-Root Rule for Scaling Memory Channels in Servers

  • Dezső SimaEmail author
Chapter
Part of the Topics in Intelligent Engineering and Informatics book series (TIEI, volume 14)

Abstract

Since core counts are rising faster than memory transfer rates, subsequent generations of multicore servers will inherently suffer from an ever-widening bandwidth gap if their system architectures are not suitably enhanced. Leading vendors typically handle this well-known issue by supporting the fastest memory rates that are feasible and raising the memory bandwidth by providing additional memory channels in their succeeding server generations. Although several papers address the memory bandwidth gap and ways to ameliorate it, no paper discusses it quantitatively. Our paper focuses on this point by deriving a scaling rule for the memory bandwidth of multicore servers with respect to the core count for preserving the per-core memory-bandwidth over technology generations in a high-end server family. The number of memory channels implemented in Intel’s, AMD’s and IBM’s high-end server lines show a good correlation with results obtained from the square-root rule.

Keywords

Memory bandwidth gap Memory wall Memory bandwidth scaling 

References

  1. 1.
    H. Esmaeilzadeh et al., Dark silicon and the end of multicore scaling, in Proceedings of the 38th International Symposium on Computer Architecture (ISCA ’11) (2011)Google Scholar
  2. 2.
    M. Ferdman et al., Clearing the clouds: a study of emerging scale-out workloads on modern hardware, in Proceedings ASPLOS’17 (2012)Google Scholar
  3. 3.
    C. Gianos, Intel Xeon processor E5-2600 v3 product family architectural overview, in Intel HPC Developer Conference at SuperComputing 2014, Presentation (2014) http://ihpcc2014.com/pdf/IntelR%20XeonR%20Processor%20E5-2600%20v3%20Overview%20for%20SC14.pdf
  4. 4.
    D.J. Gelas, The Intel Xeon E5 v4 review: testing broadwell-EP with demanding server workloads, AnandTech (2016). https://www.anandtech.com/show/10158/the-intel-xeon-e5-v4-review
  5. 5.
    R. Hadidi et al., Performance implications of NoCs on 3D-stacked memories: insights from the hybrid memory cube, in Proceedings of the 2018 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) (2018)Google Scholar
  6. 6.
    A. Hartstein et al., On the nature of cache miss behavior: is it √2? J. Instr. Lev. Parallelism 10, 1–22 (2008)Google Scholar
  7. 7.
    J. Issa, S. Figueira, HADOOP and memcached performance and power characterization and analysis. J. Cloud Comput. Adv. Syst. Appl. 1(1), 1–10 (2012)CrossRefGoogle Scholar
  8. 8.
    P. Jacob et al., Mitigating memory wall effects in high-clock-rate and multicore CMOS 3-D processor memory stacks. Proc. IEEE 97(1), 108–122 (2009)CrossRefGoogle Scholar
  9. 9.
    S. Kanev et al., Profiling a warehouse-scale computer, in Proceedings ISCA’15 (2015)Google Scholar
  10. 10.
    K.T. Malladi et al., Towards energy-proportional datacenter memory with mobile DRAM, in Proceedings ISCA’12 (2012)Google Scholar
  11. 11.
    O. Perks et al., WMTools—assessing parallel application memory utilization at scale, in Computer Performance Engineering, LNCS 6977 (2011)Google Scholar
  12. 12.
    M.K. Qureshi, Memory Scaling is Dead, Long Live Memory Scaling, University of Texas at Austin (2014). http://hps.ece.utexas.edu/yale75/qureshi_slides.pdf
  13. 13.
    B. Rogers et al., Scaling the bandwidth wall: challenges in and avenues for CMP scaling, in Proceedings of the ISCA (2009)Google Scholar
  14. 14.
    P. Rosenfeld et al., Peering over the memory wall: design space and performance analysis of the hybrid memory cube. University of Maryland, Systems and Computer Architecture Group Technical Report UMD-SCA-2012-10-01 (2012)Google Scholar
  15. 15.
    D. Sima, Decisive Aspects in the Evolution of Microprocessors. Proc. IEEE 92(12), 1896–1926 (2004)CrossRefGoogle Scholar
  16. 16.
    G. Sun et al., Moguls: a model to explore the memory hierarchy for bandwidth improvements, in Proceedings of the 38th International Symposium on Computer Architecture (ISCA ’11) (2011)Google Scholar
  17. 17.
    A. Wulf, S.A. McKee., Hitting the memory wall: implications of the obvious. Technical Report (University of Virginia, 1994)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.Óbuda UniversityBudapestHungary

Personalised recommendations