Modeling embedded processors and generating fast simulators using the machine description language LISA [Elektronische Ressource] / vorgelegt von Stefan Leo Alexander Pees

Sujets

• Informatik

Informations

• Publié par: rheinisch-westfalischen_technischen_hochschule_-rwth-_aachen
• Publié le: 01 janvier 2002
• Nombre de lectures: 21
• Langue: Deutsch
• Poids de l'ouvrage: 2 Mo

Extrait

Modeling Embedded Processors
and Generating Fast Simulators
Using the Machine Description Language LISA
Von der Fakultat¨ fur¨ Elektrotechnik und Informationstechnik
der Rheinisch–Westfalischen¨ Technischen Hochschule Aachen
zur Erlangung des akademischen Grades
eines Doktors der Ingenieurwissenschaften
genehmigte Dissertation
vorgelegt von
Diplom–Ingenieur Stefan Leo Alexander Pees
aus Aachen
Berichter: Universitatsprofessor¨ Dr. sc. techn. Heinrich Meyr
Universit¨ Dr.–Ing. Bernhard Walke
Tag der mundlichen¨ Prufung:¨ 28. November 2002
Diese Dissertation ist auf den Internetseiten
der Hochschulbibliothek online verfugbar¨ .Vorwort
Die vorliegende Dissertation entstand im Rahmen meiner Tatigk¨ eit als wissenschaftlicher Mit
arbeiter des Lehrstuhls fur¨ Integrierte Systeme der Signalverarbeitung (ISS) an der Rheinisch
Westfalischen¨ Technischen Hochschule Aachen.
Mein besonderer Dank gilt vor allem Herrn Prof. Dr. sc. techn. Heinrich Meyr, der das Referat
¨ ¨fur diese Dissertation ubernommen und meine Arbeit mit konstruktiver Kritik wissenschaftlich
begleitet hat.
¨Herrn Prof. Dr. Ing. Bernhard Walke danke ich fur¨ die Ubernahme des Korreferats und fur¨ das
meiner Arbeit entgegengebrachte Interesse.
Herrn Dr. Ing. Vojin Zivojnovic, Herrn Dipl. Ing. Andreas Ropers, Herrn Dr. Ing. Andreas Hoff
mann und Herrn Dipl. Ing. Achim Nohl bin ich fur¨ die besonders intensive und gute Zusammenar-
beit sowie fur¨ die konstruktiven Diskussionen und hilfreichen Anregungen zu besonderem Dank
verpflichtet.
Andere Projekte der DSP Tools Gruppe haben diese Arbeit beeinflusst. Mein Dank gilt Herrn Dr.
Ing. Thorsten Grotk¨ er und Herrn Dr. Ing. Markus Willems fur¨ ihre kritischen und weiterfuhrenden¨
Anregungen.
Frau Wanda Gass und Herrn Subbu Venkat von der Firma Texas Instruments danke ich fur¨ die Un
terstutzung¨ meiner Forschungsarbeit und fur¨ die Einblicke in Entwurf und Modellierung moderner
Signalprozessoren.
Bei meinen Diplomanden und studentischen Mitarbeitern bedanke ich mich fur¨ ihr großes Engage
ment. Allen Mitarbeitern des Lehrstuhls danke ich fur¨ die gute und freundliche Arbeitsatmosphare.¨
Mein abschließender und spezieller Dank gilt meiner Familie, vor allem meiner Frau Karin fur¨
ihre liebevolle Hilfe und Ermutigung in allen Phasen meiner Arbeit sowie meinen Eltern Eva und
Hans Pees fur¨ ihr Vertrauen und ihre uneingeschrankte¨ Unterstutzung.¨
Aachen, im August 2003 Stefan PeesContents
1 Introduction 1
2 Processor Modeling 5
2.1 Accuracy Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Application Domains of Hardware/software Models . . . . . . . . . . . . . . . . . 9
2.3 Retargetable Software Development Tools . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Requirements of Hardware/Software Co Design . . . . . . . . . . . . . . . . . . . 16
2.5 Limitations and Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.6 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 LISA – Machine Description Language and Generic Model 25
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Language Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3 Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 Local Declarations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.6 Behavioral Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.7 Timing Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.8 Instruction Set Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4 Retargetable Processor Simulation 58
4.1 Compiled Processor . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2 Implementation of the Generic Machine Model . . . . . . . . . . . . . . . . . . . 63
4.3 Simulator Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.4 Generating the Processor Specific Model . . . . . . . . . . . . . . . . . . . . . . . 71
5 Case Studies 80
5.1 Processor Models and Modeling Efficiency . . . . . . . . . . . . . . . . . . . . . 80
5.2 Model Compilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.3 Application Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.4 Simulation Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.5 Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6 Conclusions 89
6.1 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
iii CONTENTS
A Acknowledgement 98
B LISA Grammar 99
C Simulator API 107
D Simulator Hooks 108
E LISA description of the DLX processor 109
F LISA of the Texas Instruments C62x 121Chapter 1
Introduction
The digitalization of information and entertainment systems today covers the whole area of text,
graphics, speech, audio and video processing. A variety of digital signal processing applications
like compression, decompression, encryption and different kinds of quality improvements have
evolved from these systems. The commercialization of these systems for the consumer market
requires cost effective solutions in silicon. Due to the drop in prices, enormously increased effi
ciency and advanced miniaturization, such consumer products become more and more available,
incorporating application specific architectures that are referred to as embedded systems. The
probability that one of us has used a product that incorporates embedded systems within the last
hour or minute is extremely high. Today they can be found in many products of our every day
life. Mobile phones, personal organizers and DVD players are the most obvious ones. But they are
also increasingly installed invisible to the user in vehicles, home appliances, cameras, computer
peripherals and medical equipment.
At the same time, a universal trend of convergence can be observed. New consumer products
increasingly unify functions of different application areas. The third generation mobile telecom
munication standard (3G) will provide a single platform for different services that support the
transmission of voice, audio, video, text, graphics and other data. This platform demands new
versatile architectures that can execute a set of applications instead of just a single application.
Programmable components will be crucial element that enables the growth and success of these
products.
Furthermore, programmability is an important factor in the system design process. The pro
grammability helps to raise the designer’s productivity and the flexibility of software allows late
design changes and provides a high degree of reusability, thus shortening the design cycles. For this
reason, programmable architectures like off the shelf digital signal processors (DSPs), microcon
trollers and application specific instruction set processors (ASIPs) are increasingly employed into
embedded systems and a growing amount of system functions is implemented in software rather
than in hardware [68]. Today, one can look at embedded systems as software based solutions
that are complemented with dedicated hardware components in order to accelerate computation
intensive operations, to save power or to provide interfaces for input and output.
Designers of today’s embedded systems are facing a rapidly growing system complexity. Driven by
the advances in semiconductor technology, the amount of system functionality that can be realized
12 CHAPTER 1. INTRODUCTION
on a single chip is growing enormously. The increasing integration density of integrated circuits
has led to a shift from dedicated hardware devices that are spread over several chips towards single
chip solutions. In these systems on chips(SoC), designers are confronted with the challenge of
integrating heterogeneous hardware and software components.
Product innovation constantly shortens product life cycles to the point that companies which fail
to enter the market timely with a new must reduce the prices (and margins) in order to be
competitive. Thus, time to market is crucial for the profitability of products and market share.
Due to the complexity and time to market constraints, the designer’s productivity has become a
vital factor for successful products. But designers are facing a dilemma – on the one hand the sys
tem complexity and heterogeneity is exploding and on the other hand the design time must shrink
continuously. The only way out of this dilemma is the use of an appropriate design methodology
and efficient design automation tools.
A particular important part in the design methodology of electronic systems is design verification.
According to statements from industry, two thirds of the man power in design teams is spent on
verification. Furth