Course
PPRA - Parallel Programming and Computerarchitektur
PDF Course Catalog Deutsche Version: PPRA
Version: 3 | Last Change: 29.04.2022 08:41 | Draft: 2 | Status: Entwurf
| Long name | Parallel Programming and Computerarchitektur |
|---|---|
| Approving CModule | PPRA_BaTIN |
| Responsible |
Prof. Dr. Lothar Thieling
Professor Fakultät IME |
| Level | Bachelor |
| Semester in the year | summer semester |
| Duration | Semester |
| Hours in self-study | 60 |
| ECTS | 5 |
| Professors |
Lehrbeauftragte(r) / Thieling
|
| Requirements | basic skills in procedural programming basic skills in programming multiple tasks structure and mode of operation of a simple computer basics in digital systems (Automata, Hardware Description Language) |
| Language | German |
| Separate final exam | Yes |
Literature
Barlas, Gerassimos: Multicore and GPU Programming: An Integrated Approach
Tanenbaum, Goodman: Computerarchitektur, Pearson Studium (Prentice Hall)
Pacheco: Parallel Programming with MPI
Eijkhout: Introduction to High Performance Computing
Tanenbaum, Goodman: Computerarchitektur, Pearson Studium (Prentice Hall)
Pacheco: Parallel Programming with MPI
Eijkhout: Introduction to High Performance Computing
Final exam
Details
The students should demonstrate the following competencies in a written exam:The students should demonstrate the following skills in a written exam: 1.) Confident handling of basic terms, mechanisms and concepts. 2.) Parallel programming using common design tools (e.g. MPI and CUDA). 3.) Development of problem solutions that are predestined for the use of parallel computer systems.
Minimum standard
At least 50% of the total number of pointsExam Type
The students should demonstrate the following competencies in a written exam:The students should demonstrate the following skills in a written exam: 1.) Confident handling of basic terms, mechanisms and concepts. 2.) Parallel programming using common design tools (e.g. MPI and CUDA). 3.) Development of problem solutions that are predestined for the use of parallel computer systems.
Learning goals
Knowledge
basics of parallel programming
introduction
approach/basic idea
Data dependencies and synchronization
Parallel computer architectures
classification
MMID
SIMD
design of parallel programs
development process
decomposition pattern
completely parallel
task parallelism (incl. task pool)
divide and conquer
pipeline (or general task graph)
data parallel (geometric data)
recursive data
design of parallel programs
design pattern for parallel programming
master slave (master worker)
fork and join
single program multiple data (SPMD)
multiple program multiple data (MPMD)
map reduce
loop parallelism
mapping of program structure patterns to decomposition patterns
design of parallel programs
performance Metrics
speedup
amdahl's law
efficiency
scalability
loss of performance
load balancing
performance measurement
classification of standard libraries with regard to the preceding design options and their use based on design patterns
MPI (distributed memory)
CUDA (GPU programming)
computer architectures (according to Von-Neumann)
conceptual components to increase performance regarding ...
storage
processing units
GPU (see above)
communication
protection
implementation of the above concepts in concrete computer architectures
IA32e (AMD64)
ARM
alternative architectures in addition to von-neuman
connection of FPGAs to von veumann architectures
veural networks implemented in FPGAs
introduction
approach/basic idea
Data dependencies and synchronization
Parallel computer architectures
classification
MMID
SIMD
design of parallel programs
development process
decomposition pattern
completely parallel
task parallelism (incl. task pool)
divide and conquer
pipeline (or general task graph)
data parallel (geometric data)
recursive data
design of parallel programs
design pattern for parallel programming
master slave (master worker)
fork and join
single program multiple data (SPMD)
multiple program multiple data (MPMD)
map reduce
loop parallelism
mapping of program structure patterns to decomposition patterns
design of parallel programs
performance Metrics
speedup
amdahl's law
efficiency
scalability
loss of performance
load balancing
performance measurement
classification of standard libraries with regard to the preceding design options and their use based on design patterns
MPI (distributed memory)
CUDA (GPU programming)
computer architectures (according to Von-Neumann)
conceptual components to increase performance regarding ...
storage
processing units
GPU (see above)
communication
protection
implementation of the above concepts in concrete computer architectures
IA32e (AMD64)
ARM
alternative architectures in addition to von-neuman
connection of FPGAs to von veumann architectures
veural networks implemented in FPGAs
Skills
The students are able to
- discuss the structure, organization and operating principle of computer systems,
- analyze the connection between hardware concepts and the effects on the software, to be able to create efficient programs,
- to understand and apply the basic principles of design from the understanding of the interactions of technology, computer concepts and applications,
- evaluate and compare computer concepts.
The students are able to
- describe architectural features of parallel computers,
- evaluate parallel computers, programming paradigms and design patterns and select them for a specific application,
- to program parallel computers
specifying system behavior (derived from text documents)
- discuss the structure, organization and operating principle of computer systems,
- analyze the connection between hardware concepts and the effects on the software, to be able to create efficient programs,
- to understand and apply the basic principles of design from the understanding of the interactions of technology, computer concepts and applications,
- evaluate and compare computer concepts.
The students are able to
- describe architectural features of parallel computers,
- evaluate parallel computers, programming paradigms and design patterns and select them for a specific application,
- to program parallel computers
specifying system behavior (derived from text documents)
Expenditure classroom teaching
| Type | Attendance (h/Wk.) |
|---|---|
| Lecture | 2 |
| Exercises (whole course) | 1 |
| Exercises (shared course) | 1 |
| Tutorial (voluntary) | 0 |
Special literature
keine/none
Special requirements
none
Accompanying material
lecture slides (electronic)
set of exercises (electronic)
tool chains (compile, link, debug, simulate)
set of example program codes
set of exercises (electronic)
tool chains (compile, link, debug, simulate)
set of example program codes
Separate exam
none
Learning goals
Skills
refer to "Vorlesung/Übung->Lernziele->Fertigkeiten"
targeted use of the software development environment
manage complex tasks as a small team
Development of more complex solutions to problems in the field of compute/data intensive algorithm, signal processing or artificial intelligence or graphic animation that are specific for the use of parallel computers.
targeted use of the software development environment
manage complex tasks as a small team
Development of more complex solutions to problems in the field of compute/data intensive algorithm, signal processing or artificial intelligence or graphic animation that are specific for the use of parallel computers.
Expenditure classroom teaching
| Type | Attendance (h/Wk.) |
|---|---|
| Practical training | 1 |
| Tutorial (voluntary) | 0 |
Special literature
keine/none
Special requirements
none
Accompanying material
problem and task description (electronic)
tool chaine (compile, link, debug, simulate)
set of example-codes
self-study tutorials for the tool chain
tool chaine (compile, link, debug, simulate)
set of example-codes
self-study tutorials for the tool chain
Separate exam
none
Bei Fehlern, bitte Mitteilung an die
Webredaktion der Fakultät IME
Webredaktion der Fakultät IME
© 2022 Technische Hochschule Köln