Erweiterte Suche

The Münster Skeleton Library ‚Muesli‘

A Comprehensive Overview

Titel: The Münster Skeleton Library ‚Muesli‘
Untertitel: A Comprehensive Overview
Verfasser: Ciechanowicz, Philipp GND
Poldner, Michael GND
Kuchen, Herbert
Organisation: European Research Center for Information Systems (ERCIS)
FB 04: Wirtschaftswissenschaftliche Fakultät
Dokumenttyp: Arbeitspapier
Medientyp: Text
Erscheinungsdatum: Januar 2009
Publikation in MIAMI: 29.08.2012
Datum der letzten Änderung: 25.11.2015
Reihe Working Papers / ERCIS - European Research Center for Information Systems ; 7
Fachgebiete: Wirtschaft; Datenverarbeitung; Informatik
Sprache: Englisch
Format: PDF-Dokument
URN: urn:nbn:de:hbz:6-39359473618
Permalink: http://nbn-resolving.de/urn:nbn:de:hbz:6-39359473618
Onlinezugriff:
Inhalt:
1 Introduction ..... 5
2 Concepts ..... 8
2.1 Polymorphic Types ..... 8
2.2 Higher-Order Functions ..... 9
2.3 Partial Applications ..... 9
2.4 Serialization ..... 11
3 Distributed Data Structures ..... 12
3.1 Concepts ..... 12
3.1.1 Skeletons ..... 12
3.1.2 Local vs. Global View ..... 14
3.1.3 Multi-Core Processing ..... 15
3.2 Distributed Array ..... 16
3.2.1 Constructors ..... 16
3.2.2 Skeletons ..... 17
3.2.2.1 Computation Skeletons ..... 17
3.2.2.2 Communication Skeletons ..... 19
3.2.2.3 Combined Skeletons ..... 20
3.2.3 Auxiliary Functions ..... 22
3.3 Distributed Matrix ..... 23
3.3.1 Constructors ..... 24
3.3.2 Skeletons ..... 25
3.3.2.1 Computation Skeletons ..... 25
3.3.2.2 Communication Skeletons ..... 27
3.3.2.3 Combined Skeletons ..... 29
3.3.3 Auxiliary Functions ..... 29
3.4 Distributed Sparse Matrix ..... 31
3.4.1 Concepts ..... 32
3.4.1.1 Submatrices ..... 32
3.4.1.2 Compression Scheme ..... 33
3.4.1.3 Distribution Scheme ..... 33
3.4.2 Constructors ..... 33
3.4.3 Skeletons ..... 34
3.4.3.1 Computation Skeletons ..... 34
3.4.3.2 Communication Skeletons ..... 37
3.4.3.3 Combined Skeletons ..... 38
3.4.4 Auxiliary Functions ..... 38
3.4.5 Results ..... 40
4 Task Parallel Skeletons ..... 45
4.1 Atomic Buildung Blocks ..... 46
4.1.1 Initial ..... 46
4.1.2 Final ..... 47
4.1.3 Atomic ..... 47
4.1.4 Filter ..... 47
4.2 Skeletons ..... 48
4.2.1 Branch & Bound ..... 48
4.2.2 Divide & Conquer ..... 51
4.2.3 Pipe ..... 55
4.2.4 Farm ..... 56
5 Selected Implementation Aspects ..... 58
5.1 Serialization ..... 58
5.2 DistributedSparseMatrix<E,S,D> ..... 62
5.3 Submatrix<E> ..... 63
5.4 Distribution ..... 64
5.5 Enhanced Skeletons ..... 65
6 Case Studies ..... 68
6.1 Combining Task and Data Parallelism ..... 68
6.2 Mergesort ..... 71
References ..... 79