456-0917/01 – Parallel Algorithms (PA)
Gurantor department | Department of Computer Science | Credits | 10 |
Subject guarantor | RNDr. Ondřej Jakl, CSc. | Subject version guarantor | RNDr. Ondřej Jakl, CSc. |
Study level | postgraduate | Requirement | Choice-compulsory |
Year | | Semester | winter + summer |
| | Study language | Czech |
Year of introduction | 1997/1998 | Year of cancellation | 2009/2010 |
Intended for the faculties | FEI | Intended for study types | Doctoral |
Subject aims expressed by acquired skills and competences
A general review of design, realization and evaluation of parallel algorithms, with specifical extensions in the directions matching the student's Ph.D. study.
Overview in the field of parallel systems and their programming.
Design of parallel algorithms, their analysis, efficient implementation and optimization.
Application of those methods and techniques in a specific area.
Teaching methods
Summary
The course provides the students with necessary fundamentals in the area
of parallel programming and algorithms, to be able to make effective use of modern
multiprocessor hardware (including supercomputers) for solution of demanding task
from various application fields. It focuses on the most general parallel architecture, distributed-memory systems, where the interaction of parallel processes is based on message passing, but techniques specific for symmetric multiprocessors (with shared memory) are covered, too. Practical training is provided on the most powerful platforms available at VŠB-TUO, including the Termit cluster having 32 processors. Individual approach,
allowing application of the gained knowledge in the student's specific field of study,
is supposed.
Compulsory literature:
Recommended literature:
Way of continuous check of knowledge in the course of semester
E-learning
Other requirements
Prerequisities
Subject has no prerequisities.
Co-requisities
Subject has no co-requisities.
Subject syllabus:
Lectures:
Introduction. Motivation and historical remarks. Basic concepts.
Parallel computer systems. Flynn's classification. Interconnection subsystems.
Loosely coupled multiprocessors. Computer networks and massively parallel systems. The Termit cluster and IBM SP.
Parallel programming models. The message passing model.
Design of a parallel algorithm: decomposition, communication analysis, agglomeration, mapping to processors.
Load-balancing techniques.
Evaluation of parallel algorithms. Speedup, efficiency, cost. Amdahl's law.
Scalability of parallel applications. Isoefficiency function.
Parallel Virtual Machine (PVM). Application programming interface.
PVM (continued). Overview of library routines: process control, information retrieval, message passing,
etc.
PVM (continued). Collective communication. Debugging issues. Visualization of a parallel execution.
Selected algorithms in a chosen application area and their parallelization.
Introduction to the Message Passing Interface (MPI). Comparison with PVM, advanced features.
Programming of symmetric multiprocessors. Threads and OpenMP.
Current developments in high performance computing.
Conditions for subject completion
Occurrence in study plans
Occurrence in special blocks
Assessment of instruction
Předmět neobsahuje žádné hodnocení.