1. Theorie der Parallelität (parallele Computermodelle, parallele Spezifikationsformen und -sprachen, Performanzmodelle und -berechnung)
2. Klassifikation paralleler und skalierbarer Rechnerarchitekturen (Multiprozessoren und Multicomputer, Vektorrechner, Datenflussmaschinen, VLSI-Rechenfelder)
3. Programmierung paralleler Rechner (Sprachen und Modelle, Entwurfsmethoden und Compiler, Optimierung)
4. Massive Parallelität: Vom Algorithmus zur Schaltung

Theoretische und praktische Übungen mit rechnergestützten Werkzeugen

Title:

Parallel Systems

Credits: 2,5

Contents

Today's PCs consist of multi-core processors and graphics cards that again comprise hundreds to thousands of simple processors. As a result of this, a very high degree of parallel data processing becomes possible, which was subjected to supercomputers a couple of years ago. The efficient exploitation of parallel processing requires not only multiple processors but also parallelism inherent in the problem to be solved. In this lecture, properties of different parallel computer architectures and corresponding quality metrics are examined. Further, models and parallel programming languages are introduced. In addition to programming general parallel computers, design methods (CAD) are presented that systematically transform an algorithmic problem description into a massive parallel processor array (VLSI), which can optimally execute the given problem in parallel. Such highly parallel circuits play an essential role at the bit level and circuit level (arithmetics) as well as in the case of signal processing and image processing (e.g., filter).

In detail, the following topics are covered:

1. Theory of parallelism (parallel models of computation, parallel specification and parallel languages, performance models)
2. Classification of parallel and scalable computer architectures (multi-processors and multi-computers, vector computers, data-flow machines, VLSI processor arrays)
3. Programmable System-on-Chip (SoC) and multi-core architetcures (graphics processors, Cell, etc.)
4. Programming of parallel computers (languages and models, design methods and compiler, optimization)
5. Massive parallelism: From algorithm to circuit

Theoretic tasks and practical exercises with computer-aided design tools.

UE: Erweiterte Übungen zu Parallele Systeme

Dozentinnen/Dozenten: Marcel Brand, M. Sc., Dipl.-Inf. Michael Witterauf

UE: Übung zu Parallele Systeme

Dozent/in: PD Dr.-Ing. Frank Hannig