Parallel algorithms and architectures

Lecturer:

Tomáš Oberhuber (gar.)

Tutor:

Tomáš Oberhuber (gar.)

Supervisor:

Department of Mathematics

Synopsis:

This course deals with the parallel data processing. It is important in situations when one processing unit (CPU) is not powerful enough to finish given task in reasonable time. When designing parallel algorithms, good knowledge of the parallel architectures is important. Therefore these architectures are studied as a part of this course too.

Requirements:

Basic algorithms, programing in C/C++.

Syllabus of lectures:

1. Introduction

2. Sequential architectures

3. Parallel architectures

4. Communication networks

5. Communication operations

6. Introduction to CUDA

7. OpenMP / MPI

8. Analysis of parallel algorithms

9. Sorting algorithms

10. Matrix algorithms

11. Graph algorithms, Monte-Carlo methods 12. Combinatorial search

Syllabus of tutorials:

1. Programming in CUDA

2. OpenMP / MPI

3. Analysis of parallel algorithms

4. Sorting algorithms

5. Matrix algorithms

6. Graph algorithms, Monte-Carlo methods

7. Combinatorial search

Study Objective:

Knowledge:

Parallel architectures, basic types of parallel architectures, communication in parallel architectures, programming standards OpenMP, MPI, CUDA/OpenCL, sorting algorithms, matrix algorithms, graph algorithms, Monte-Carlo methods, analysis of the parallel algorithms.

Skills:

The students will learn how to choose appropriate parallel architecture for given problem, how to design proper parallel algorithm, analyze it and derive its efficiency and how to implement the parallel algorithm.

Study materials:

Key references:

Grama A., Karypis G., An Introduction to Parallel Computing: Design and Analysis of Algorithms

Recommended references:

CUDA Programming guide

Media and tools:

Computer lab

Note:

Time-table for winter semester 2011/2012:

Time-table is not available yet

Time-table for summer semester 2011/2012:

Time-table is not available yet

The course is a part of the following study plans: