A High-Performance Retargetable Simulator for Parallel Architectures

Abstract

In this thesis, we describe Proteus, a high-performance simulation- based system for the evaluation of parallel algorithms and system software. Proteus is built around a retargetable parallel architecture simulator and a flexible data collection and display component. The simulator uses a combination of simulation and direct execution to achieve high performance, while retaining simulation accuracy. Proteus can be configured to simulate a wide range of shared memory and message passing MIMD architectures and the level of simulation detail can be chosen by the user. Detailed memory, cache and network simulation is supported. Parallel programs can be written using a programming model based on C and a set of runtime system calls for thread and memory management. The system allows nonintrusive monitoring of arbitrary information about an execution, and provides flexible graphical utilities for displaying recorded data. To validate the accuracy of the system, a number of published experiments were reproduced on Proteus. In all cases the results obtained by simulation are very close to those published, a fact that provides support for the reliability of the system. Performance measurements demonstrate that the simulator is one to two orders of magnitude faster than other similar multiprocessor simulators.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1991
Accession Number
ADA237354

Entities

People

  • Chrysanthos N. Dellarocas

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Application Software
  • C Programming Language
  • Classification
  • Computer Architecture
  • Computer Programming
  • Computer Science
  • Computers
  • Data Displays
  • Electrical Engineering
  • Instruction Set Architecture
  • Language
  • Machines
  • Operating Systems
  • Parallel Computing
  • Parallel Processing
  • Programming Languages
  • Simulators

Fields of Study

  • Computer science
  • Engineering

Readers

  • Computational Modeling and Simulation
  • Parallel and Distributed Computing.