Browse Prior Art Database

NEW PERFORMANCE MEASURES FOR DESIGN AND EVALUATION OF REAL-TIME MULTIPROCESSORS

IP.com Disclosure Number: IPCOM000128461D
Original Publication Date: 1983-Nov-01
Included in the Prior Art Database: 2005-Sep-16

Publishing Venue

Software Patent Institute

Related People

Shin, Kang G.: AUTHOR [+4]

Abstract

Conventional performance measures such as throughput, reliability, and availability are not by themselves alone suitable for analyzing and designing real-time multiprocessors. (The real-time system we focus on here consists of a computer controller and controlled processes.). Most of these measures treat the computer as a monolith in terms of the services it delivers, thus failing to accurately describe the nature of diversified real-time application tasks.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 6% of the total text.

Page 1 of 17

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

NEW PERFORMANCE MEASURES FOR DESIGN AND EVALUATION OF REAL-TIME MULTIPROCESSORS

Kang G. Shin and C. M. Krishna

CRL-TR-33-83

NOVEMBER 1983

THE UNIVERSITY OF MICHIGAN COMPUTING RESEARCH LABORATORY1 Room 1079, East Engineering Building
Ann Arbor, Michigan 48109
USA
Tel: (313) 763-8000

ABSTRACT

Conventional performance measures such as throughput, reliability, and availability are not by themselves alone suitable for analyzing and designing real-time multiprocessors. (The real-time system we focus on here consists of a computer controller and controlled processes.). Most of these measures treat the computer as a monolith in terms of the services it delivers, thus failing to accurately describe the nature of diversified real-time application tasks.

As a remedy for this problem, this report presents some new performance measures based on computer(multiprocessor) controller response time that are (i) congruent to the real-time applications, (ii) able to offer an objective comparison of rival computer systems, and (iii) experimentally measurable/determinable. These measures, unlike others, provide the real-time multiprocessor controller with a natural link to controlled processes.

In order to demonstrate their utility and power, these measures are first determined for an example controlled process on the basis of two control performance functionals.

They are then used for an important real-time multiprocessor design application- the number- power tradeoff. Also, some potential applications of these measures are identified. ndex Terms - - Performance measures, real-time multiprocessors, controllers and controlled processes, response time, cost functions, dynamic failure, hard deadlines, number- power tradeoff.

1. INTRODUCTION

Like every other medium of expression, performance measures for computers influence what is expressed almost as much as how it is expressed. Analyses of computer systems therefore depend greatly for their precision and relevance upon the performance measures used.

1 This work was supported in part by National Aeronautics and Space Administration Grant No. NAG 1-196. Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of NASA.

University of Michigan Computing Research Laboratory Page 1 Nov 01, 1983

Page 2 of 17

NEW PERFORMANCE MEASURES FOR DESIGN AND EVALUATION OF REAL-TIME MULTIPROCESSORS

In this report, we consider computer systems used in real-time control of critical processes. A critical process is one whose failure can lead to loss of life, or other catastrophe. Such processes include aircraft, spacecraft, manufacturing processes, and power generation and distribution. They cover a wide spectrum, but have the following in common: in all of them computers are used in control, in all of them the job load of the control computer is precisely characteri...