Trends in Data Locality Abstractions for HPC Systems

Didem Unat 1 Anshu Dubey 2 Torsten Hoefler 3 John Shalf 4 Mark Abraham 5 Mauro Bianco 6 Bradford L. Chamberlain 7 Romain Cledat 8 H. Carter Edwards 9 Hal Finkel 2 Karl Fuerlinger 10 Frank Hannig 11 Emmanuel Jeannot 12 Amir Kamil 13 Jeff Keasler 14 Paul H J Kelly 15 Vitus Leung 9 Hatem Ltaief 16 Naoya Maruyama 17 Chris J. Newburn 18 Miquel Pericàs 19
1 Koç University
2 ANL - Argonne National Laboratory [Lemont]
3 D-INFK - Department of Computer Science [ETH Zürich]
4 LBNL - Lawrence Berkeley National Laboratory [Berkeley]
5 KTH - Royal Institute of Technology [Stockholm]
6 CSCS - Swiss National Supercomputing Centre-SCR department
7 Cray Inc [Switzerland]
8 INTEL
9 SNL - Sandia National Laboratories [Albuquerque]
10 LMU - Institut für Informatik [München/Munich]
11 Department of Computer Science [Erlangen]
12 TADAAM - Topology-Aware System-Scale Data Management for High-Performance Computing
LaBRI - Laboratoire Bordelais de Recherche en Informatique, Inria Bordeaux - Sud-Ouest
13 University of Michigan [Ann Arbor]
14 LLNL - Lawrence Livermore National Laboratory
15 Imperial College London
16 KAUST - King Abdullah University of Science and Technology
17 RIKEN
18 NVIDIA - NVIDIA
19 Chalmers University of Technology [Göteborg]
Abstract : The cost of data movement has always been an important concern in high performance computing (HPC) systems. It has now become the dominant factor in terms of both energy consumption and performance. Support for expression of data locality has been explored in the past, but those efforts have had only modest success in being adopted in HPC applications for various reasons. them However, with the increasing complexity of the memory hierarchy and higher parallelism in emerging HPC systems, locality management has acquired a new urgency. Developers can no longer limit themselves to low-level solutions and ignore the potential for productivity and performance portability obtained by using locality abstractions. Fortunately, the trend emerging in recent literature on the topic alleviates many of the concerns that got in the way of their adoption by application developers. Data locality abstractions are available in the forms of libraries, data structures, languages and runtime systems; a common theme is increasing productivity without sacrificing performance. This paper examines these trends and identifies commonalities that can combine various locality concepts to develop a comprehensive approach to expressing and managing data locality on future large-scale high-performance computing systems.
Keywords : Data locality programming abstractions high-performance computing data layout locality-aware runtimes
Type de document :
Article dans une revue
IEEE Transactions on Parallel and Distributed Systems, Institute of Electrical and Electronics Engineers, 2017, 28 (10), pp.3007 - 3020. 〈10.1109/TPDS.2017.2703149〉
Liste complète des métadonnées

Littérature citée [67 références]  Voir  Masquer  Télécharger
https://hal.inria.fr/hal-01621371
Contributeur : Emmanuel Jeannot <>
Soumis le : mardi 24 octobre 2017 - 11:25:16
Dernière modification le : samedi 20 octobre 2018 - 15:06:01
Document(s) archivé(s) le : jeudi 25 janvier 2018 - 12:37:02

Fichier

TPDS-2016-06-0377.R2.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Collections

UNIV-BORDEAUX | LABRI | INRIA | INRIA2017

Citation

Didem Unat, Anshu Dubey, Torsten Hoefler, John Shalf, Mark Abraham, et al.. Trends in Data Locality Abstractions for HPC Systems. IEEE Transactions on Parallel and Distributed Systems, Institute of Electrical and Electronics Engineers, 2017, 28 (10), pp.3007 - 3020. 〈10.1109/TPDS.2017.2703149〉. 〈hal-01621371〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

310

Téléchargements de fichiers

289

Contact


archive-ouverte@inria.fr