Options
Mapping dynamic programming algorithms on graphics processing units
Citation Link: https://doi.org/10.15480/882.1184
Other Titles
Abbildung von Algorithmen der dynamischen Programmierung auf Graphikprozessoren
Publikationstyp
Doctoral Thesis
Date Issued
2014
Sprache
English
Author(s)
Advisor
Title Granting Institution
Technische Universität Hamburg
Place of Title Granting Institution
Hamburg
Examination Date
2014-07-03
Institut
TORE-DOI
Alignment is the fundamental operation used to compare biological sequences. It also serves to identify regions of similarity that are eventually consequences of structural, functional, or evolutionary relationships. Today, the processing of sequences from large DNA or protein databases is a big challenge. Graphics Processing Units (GPUs) are based on a highly parallel, many-core streaming architecture and can be used to tackle the processing of large biological data. In the thesis, progressive alignment methods and their parallel implemenation by modern GPUs are studied. It turns out that wavefront and matrix-matrix product techniques can cope best with the data dependencies and so are highly appropriate for the implementation on a GPU. The performance of these methods is analyzed and the method with the highest speed-up is used to realize the alignment stage in the well-known software package ClustalW. Similar studies are made for the hidden Markov model. General principles and guidelines for GPU programming of matrix-based algorithms are discussed.
Subjects
graphics processing units
hidden Markov model
dynamic programming
Loading...
Name
Muhammad_Kashif_Hanif.pdf
Size
1012.71 KB
Format
Adobe PDF