Advanced search
1 file | 695.87 KB

Modeling critical sections in Amdahl's law and its implications for multicore design

Stijn Eyerman (UGent) and Lieven Eeckhout (UGent)
Author
Organization
Abstract
This paper presents a fundamental law for parallel performance: it shows that parallel performance is not only limited by sequential code (as suggested by Amdahl’s law) but is also fundamentally limited by synchronization through critical sections. Extending Amdahl’s software model to include critical sections, we derive the surprising result that the impact of critical sections on parallel performance can be modeled as a completely sequential part and a completely parallel part. The sequential part is determined by the probability for entering a critical section and the contention probability (i.e., multiple threads wanting to enter the same critical section). This fundamental result reveals at least three important insights for multicore design. (i) Asymmetric multicore processors deliver less performance benefits relative to symmetric processors than suggested by Amdahl’s law, and in some cases even worse performance. (ii) Amdahl’s law suggests many tiny cores for optimum performance in asymmetric processors, however, we find that fewer but larger small cores can yield substantially better performance. (iii) Executing critical sections on the big core can yield substantial speedups, however, performance is sensitive to the accuracy of the critical section contention predictor.
Keywords
critical sections, synchronization, Analytical performance modeling, Amdahl's law

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 695.87 KB

Citation

Please use this url to cite or link to this publication:

Chicago
Eyerman, Stijn, and Lieven Eeckhout. 2010. “Modeling Critical Sections in Amdahl’s Law and Its Implications for Multicore Design.” In Conference Proceedings Annual International Symposium on Computer Architecture, 362–370. New York, NY, USA: Association for Computing Machinery (ACM).
APA
Eyerman, S., & Eeckhout, L. (2010). Modeling critical sections in Amdahl’s law and its implications for multicore design. Conference Proceedings Annual International Symposium on Computer Architecture (pp. 362–370). Presented at the 37th Annual international symposium on Computer Architecture (ISCA 2010), New York, NY, USA: Association for Computing Machinery (ACM).
Vancouver
1.
Eyerman S, Eeckhout L. Modeling critical sections in Amdahl’s law and its implications for multicore design. Conference Proceedings Annual International Symposium on Computer Architecture. New York, NY, USA: Association for Computing Machinery (ACM); 2010. p. 362–70.
MLA
Eyerman, Stijn, and Lieven Eeckhout. “Modeling Critical Sections in Amdahl’s Law and Its Implications for Multicore Design.” Conference Proceedings Annual International Symposium on Computer Architecture. New York, NY, USA: Association for Computing Machinery (ACM), 2010. 362–370. Print.
@inproceedings{1024596,
  abstract     = {This paper presents a fundamental law for parallel performance: it shows that parallel performance is not only limited by sequential code (as suggested by Amdahl{\textquoteright}s law) but is also fundamentally limited by synchronization through critical sections. Extending Amdahl{\textquoteright}s software model to include critical sections, we derive the surprising result that the impact of critical sections on parallel performance can be modeled as a completely sequential part and a completely parallel part. The sequential part is determined by the probability for entering a critical section and the contention probability (i.e., multiple threads wanting to enter the same critical section).
This fundamental result reveals at least three important insights for multicore design. (i) Asymmetric multicore processors deliver less performance benefits relative to symmetric processors than suggested by Amdahl{\textquoteright}s law, and in some cases even worse performance. (ii) Amdahl{\textquoteright}s law suggests many tiny cores for optimum performance in asymmetric processors, however, we find that fewer but larger small cores can yield substantially better performance. (iii) Executing critical sections on the big core can yield substantial speedups, however, performance is sensitive to the accuracy of the critical section contention predictor.},
  author       = {Eyerman, Stijn and Eeckhout, Lieven},
  booktitle    = {Conference Proceedings Annual International Symposium on Computer Architecture},
  isbn         = {9781450300520},
  issn         = {1063-6897},
  keyword      = {critical sections,synchronization,Analytical performance modeling,Amdahl's law},
  language     = {eng},
  location     = {Saint-Malo, France},
  pages        = {362--370},
  publisher    = {Association for Computing Machinery (ACM)},
  title        = {Modeling critical sections in Amdahl's law and its implications for multicore design},
  url          = {http://dx.doi.org/10.1145/1815961.1816011},
  year         = {2010},
}

Altmetric
View in Altmetric
Web of Science
Times cited: