A Period Graph Throughput Estimator for Multiprocessor Systems

Abstract

A critical challenge in synthesis techniques for iterative applications is the efficient analysis of performance in the presence of communication resource contention. To address this challenge, we introduce the concept of the period graph. The period graph is constructed from the output of a simulation of the system, with idle states included in the graph, and its maximum cycle mean is used to estimate overall system throughput. As an example of the utility of the period graph, we demonstrate its use in a joint power/performance optimization solution that uses either a nested genetic algorithm, or a simulated annealing algorithm. We analyze the fidelity of this estimator, and quantify the speedup and optimization accuracy obtained compared to simulation.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 2006
Accession Number
ADA448068

Entities

People

  • Neal K. Bambha
  • Shuvra S. Bhattacharyya

Organizations

  • University of Maryland

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Annealing
  • Computations
  • Computers
  • Energy Consumption
  • Estimators
  • Genetic Algorithms
  • Multiprocessors
  • Optimization
  • Reliability
  • Scheduling (Production)
  • Simulations
  • Simulators
  • Specifications
  • Steady State
  • Throughput
  • Universities

Fields of Study

  • Computer science

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Parallel and Distributed Computing.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms
  • Biotechnology