Theoretical Nabladot Analysis of Amdahl’s Law for Agent-Based Simulations

  • Claudio Cioffi-Revilla
Conference paper

DOI: 10.1007/978-3-319-14325-5_38

Part of the Lecture Notes in Computer Science book series (LNCS, volume 8805)
Cite this paper as:
Cioffi-Revilla C. (2014) Theoretical Nabladot Analysis of Amdahl’s Law for Agent-Based Simulations. In: Lopes L. et al. (eds) Euro-Par 2014: Parallel Processing Workshops. Euro-Par 2014. Lecture Notes in Computer Science, vol 8805. Springer, Cham

Abstract

Amdahl’s Law states that execution speedup S is nonlinearly proportional to the percentage of parallelizable code P and the number N of processors. Additional terms must be added to Amdahl’s Law when applied to agent-based simulations, depending on how synchronization is implemented. Since P is continuous but N is discrete, traditional multivariate operators based on nabla or del ∇ are applicable only for P, not for N, regardless of synchronization architecture (linear, logarithmic, constant, among other). Moreover, relatively low values of N (bound by Miller’s number 7 ±2) are common in some cases. Here I apply a novel and exact operator, called “nabladot” and denoted by the symbol “nabladot”, that is defined for hybrid function such as Amdahl’s Law. The main results show how exact solutions using nabladot differ from traditional approximations, particularly in the logarithmic case that is characteristic of hierarchical synchronization. Improvements in precision are inversely proportional to P and N, converging to 0.8 as N → 2.

Keywords

Amdahl’s Law nabladot multivariate vector analysis distributed systems concrete mathematics hybrid functions 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Claudio Cioffi-Revilla
    • 1
  1. 1.Center for Social Complexity and Department of Computational Social ScienceGeorge Mason UniversityFairfaxUSA

Personalised recommendations