Computer Architecture Prof. Dr. Uwe Brinkschulte Slide Sets WS 2012/2013 Prof. Dr. Uwe Brinkschulte M.Sc. Benjamin Betting Part 1 Introduction Computer Architecture – Part 1 – page 1 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

M. Sc. Benjamin Betting, room 212 Chair for Embedded Systems betting@es.cs.uni-frankfurt.de Robert-Mayer-Straße 11-15 Secretary: Linda Stapleton, room 211a stapleton@es.cs.uni-frankfurt.de Computer Architecture – Part 1 – page 2 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Lecture Canon Hardware-Architektur & Rechnersysteme SS Computer Architecture WS Rechnertechnologie SS Eingebettete Systeme WS Ausgewählte Themen bei Eingebetteten Systemen - Organic Computing SS/WS Praktikum Grundlagen Hardwaresysteme WS/SS Praktikum Mikrocontroller & Eingebettete Systeme WS Seminar Robuste Systemarchitekturen – Organic Computing SS Computer Architecture – Part 1 – page 3 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Hard- and Software for Embedded Systems Especially: Research Area: Hard- and Software for Embedded Systems Especially: Microcontroller & Microprocessors Embedded Real-time Systems Distributed Embedded Systems Real-time Middleware Organic Computing Self-Organization and Real-time Dependable Embedded Systems Computer Architecture – Part 1 – page 4 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Some Projects Komodo CAR-SoC CARISMA DODOrg MixedCoreSoC OSA+ REMIS Computer Architecture – Part 1 – page 5 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Computer Architecture 3 L + 1 E Lecture: Time and place: Wednesday, 12:15 - 14:45, break at 13:30 - 13:45, SR 11, RM 11-15 Exercise: Time and place: Thursday, 12:00 - 13:00, SR 307, RM 11-15 will be announced in lecture Courses: Bachelor, Diplom, Bioinformatik, L3, (CSC Master auslaufend) ECTS-Credits: 6 Language: German/ English Computer Architecture – Part 1 – page 6 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

People / Office Hours Lecture Exercise M.Sc. Benjamin Betting Institut für Informatik Eingebettete Systeme Robert-Mayer-Str. 11-15 60325 Frankfurt Room 210 Phone 069/798-28234 brinks@es.cs.uni-frankfurt.de Office Hour: on appointment Exercise M.Sc. Benjamin Betting Institut für Informatik Eingebettete Systeme Robert-Mayer-Str. 11-15 60325 Frankfurt Room 212 Phone 069/798-28252 renteln@es.cs.uni-frankfurt.de Office Hour: We. 11:00-12:00 and on appointment Computer Architecture – Part 1 – page 7 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Slides available at: http://www.es.cs.uni-frankfurt.de Computer Architecture – Part 1 – page 8 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Computer Architecture Literature: David A. Patterson, John L. Hennessy: Computer Organization and Design – The Hardware/Software Interface, Morgan Kaufmann Publishers Jurij Silc, Borut Robic, Theo Ungerer: Processor Architecture, Springer Verlag Uwe Brinkschulte, Theo Ungerer: Mikrocontroller & Mikroprcessoren, Springer Verlag (German) Computer Architecture – Part 1 – page 9 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Computer Architecture Content: Fundamentals 01. Introduction 02. Microprocessor Development 03. Fundamentals in Computer Architecture 04. Fundamentals in Computer Technology 05. Fundamentals in Computer Design 06. Fundamentals in Performance Evaluation Computer Architecture – Part 1 – page 10 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Computer Architecture Content (cont.): Current Microprocessor-Architectures 07. Instruction Set Architecture 08. Instruction Level Parallelism - Pipelining 09. Instruction Level Parallelism - Concurrency 10. Thread and Task Level Parallelism Memory-Systems 11. Memory Management 12. Memory Hierarchy and Caches Computer Architecture – Part 1 – page 11 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

„The types of architecture are established not by architects but by society, according to the needs of the different institutions. Society sets the goals and assigns to the architect the job of finding the means of achieving them.“ (Encyclopaedia Britannica) Computer Architecture – Part 1 – page 12 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Architecture (construction) vs. Computer architecture Architecture in general follows one or more different goals (functionalities). These functionalities can be specified by types Architecture (construction) Example: architecture type 1 (private goals): {home building, factory, ...} architecture type 2 (public goals): {sports arena, hospital, railway station, airport, ...} . . . Computer Architecture – Part 1 – page 13 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Architecture (construction) vs. Computer architecture computer architecture type 1: {universal computer} computer architecture type 2: {special computer} computer architecture type special computer: {signal processing computer, simulation computer, graphics computer, …} summary: The increasing importance of embedded and networked systems in future will create further types in computer architecture concerning their functionality. Computer Architecture – Part 1 – page 14 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Architecture (construction) vs. Computer architecture Note: The term “architecture” is also frequently used in other context as e.g.: system architecture software architecture chip architecture Computer Architecture – Part 1 – page 15 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Nonfunctional constraints Beside the most important functional concept of the architecture, several nonfunctional constraints has to be considered for the design. Some of these nonfunctional constraints are: speed performance security safety scalability power awareness etc. The design process of computer architectures is controlled and directed by these nonfunctional constraints. Computer Architecture – Part 1 – page 16 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Parallel Computing Superscalar VLIW Instruction-level Parallelism Multi-threaded Instruction-level Parallelism SMP Multi-core FPGA Microgrid Parallel embedded system (MPSoC) Multi-Processing Cluster of workstations Grid computing Public resource computing High Performance Computing Computer Architecture – Part 1 – page 17 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Topics The Information Technology (IT) is based more and more on completely networked systems, so called ubiquitous computing systems with adaptive, flexible and biologically inspired cooperative system behavior a comfortable user interface © C. Müller-Schloer 2003 Systems with these features are called „Organic“ which refers to the biological inspiration. Computer Architecture – Part 1 – page 18 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Computing Trends Number Size many computers per person 1 computer many people 1 computer per person © C. Müller-Schloer 2003 Computer Architecture – Part 1 – page 19 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Scenarios Many computers per person means First of all: Many computers per person means There is great potential for parallel processing! But: Highly heterogeneous devices No stable networks, very dynamic Spontaneous entry and exit of devices Therefore: Very different from classical parallel computing!  Consider the following scenarios for innovative application of parallel computing: many computers per person © Hartmut Schmeck Computer Architecture – Part 1 – page 20 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Scenarios Smart house Intelligent control of house functions Knows what you need when you get home Smart car Adapts to different drivers, road conditions Gives advice on currently best routes Communicates with other cars on special events Integrates your personal devices into its network Smart factory Intelligent control of production (federations of robots,…) Integrates supply chain management Reacts to unexpected disturbances Maintains predetermined quality levels many computers per person © Hartmut Schmeck Computer Architecture – Part 1 – page 21 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Scenarios Smart office Takes care of your time-table, projects, events,… Provides all the office resources you need Takes care of travel arrangements … Smart shop / smart warehouse Monitors current supply levels and takes appropriate actions Knows your shopping preferences Observes your shopping habits Smart clothes / Wearable devices Check your “personal parameters” Adapt their properties / behaviour to current personal data Give proactive advice on fitness / wellness actions Enabling technology for e-health / e-care many computers per person © Hartmut Schmeck Computer Architecture – Part 1 – page 22 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Emergence Emergent creation of a snow flake Emergence in perception Computer Architecture – Part 1 – page 23 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Emergence Local actions/behaviour of the members of a self-organizing system may lead to observable, emergent global patterns, structure, or behaviour. This global behaviour is of a different kind than the behaviour of its components (in particular, not a linear combination of the individual actions). The removal of (single) components does not lead to a failure of the global functions of the system. The global behaviour is completely new compared to that of the existing components, i.e. the emergent behaviour seems to be unpredictable and not deducible from the individual components of the system, and it cannot be reduced on these. (cf: Emergence, a Journal of Complexity Issues in Organisation and Management) Computer Architecture – Part 1 – page 24 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Vision Organic computer systems are biological or life inspired tems consist of autonomous system parts (autonomous agents) and behave selforganizing. self organisation means: self-configuring self optimizing self healing They are called in general self-X features © Hartmut Schmeck Computer Architecture – Part 1 – page 25 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

Organic Computing It is not the question, whether adaptive and self-organising systems will emerge, but how they will be designed © Hartmut Schmeck Computer Architecture – Part 1 – page 26 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting

High performance computing Beside modern trends as e.g. embedded systems, ubiquitous computing, organic computing, pervasive systems etc. exists still classical parallel computing for high performance application Parallel High performance computing (vector machines) Computer cluster Grid computing systems Public resource computing (Internet) Multi processor systems (SMP) (Multi and many core processors) High performance computing Computer Architecture – Part 1 – page 27 of 27 – Prof. Dr. Uwe Brinkschulte, M.Sc. Benjamin Betting