Trends in Data Locality Abstractions for HPC Systems

Didem Unat; Anshu Dubey; Torsten Hoefler; John Shalf; Mark Abraham; Mauro Bianco; Bradford L. Chamberlain; Romain Cledat; H. Carter Edwards; Hal Finkel; Karl Fuerlinger; Frank Hannig; Emmanuel Jeannot; Amir Kamil; Jeff Keasler; Paul H J Kelly; Vitus Leung; Hatem Ltaief; Naoya Maruyama; Chris J. Newburn; Miquel Pericas

doi:10.1109/TPDS.2017.2703149

Trends in Data Locality Abstractions for HPC Systems

Didem Unat, Anshu Dubey, Torsten Hoefler, John Shalf, Mark Abraham, Mauro Bianco, Bradford L. Chamberlain, Romain Cledat, H. Carter Edwards, Hal Finkel, Karl Fuerlinger, Frank Hannig, Emmanuel Jeannot, Amir Kamil, Jeff Keasler, Paul H J Kelly, Vitus Leung, Hatem Ltaief, Naoya Maruyama, Chris J. NewburnMiquel Pericas

Computer, Electrical and Mathematical Sciences and Engineering

Research output: Contribution to journal › Article › peer-review

71 Scopus citations

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.

Original language	English (US)
Pages (from-to)	3007-3020
Number of pages	14
Journal	IEEE Transactions on Parallel and Distributed Systems
Volume	28
Issue number	10
DOIs	https://doi.org/10.1109/TPDS.2017.2703149
State	Published - May 12 2017

Bibliographical note

KAUST Repository Item: Exported on 2020-10-01
Acknowledgements: Authors would like to thank other PADAL14 and PADAL15 workshop participants: Maciej Besta, Jed Brown, Cy Chan, Sung-Eun Choi, Jack Choquette, Brice Goglin, Jesus Labarta, Leonidas Linardakis, Edward Luke, Satoshi Matsuoka, Peter Messmer, Lawrence Mitchell, Kathryn O'Brien, David Padua, Robert B. Ross, Marie-Christine Sawley, Robert Schreiber, Thomas Schulthess. James Sexton, Suzanne Michelle Shontz, Adrian Tate, Gysi Tobias, Engo Toshio, Mohamed Wahib, Chih-Chieh Yang. This work was partially supported by the German Research Foundation under contract TE163/17-1. This work was partially supported by the Grant 655965 by the European Commission.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/TPDS.2017.2703149

Cite this

Unat, D., Dubey, A., Hoefler, T., Shalf, J., Abraham, M., Bianco, M., Chamberlain, B. L., Cledat, R., Edwards, H. C., Finkel, H., Fuerlinger, K., Hannig, F., Jeannot, E., Kamil, A., Keasler, J., Kelly, P. H. J., Leung, V., Ltaief, H., Maruyama, N., ... Pericas, M. (2017). Trends in Data Locality Abstractions for HPC Systems. IEEE Transactions on Parallel and Distributed Systems, 28(10), 3007-3020. https://doi.org/10.1109/TPDS.2017.2703149

@article{be930aed9c7e4defb9de65fac1e315f1,

title = "Trends in Data Locality Abstractions for HPC Systems",

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.",

author = "Didem Unat and Anshu Dubey and Torsten Hoefler and John Shalf and Mark Abraham and Mauro Bianco and Chamberlain, {Bradford L.} and Romain Cledat and Edwards, {H. Carter} and Hal Finkel and Karl Fuerlinger and Frank Hannig and Emmanuel Jeannot and Amir Kamil and Jeff Keasler and Kelly, {Paul H J} and Vitus Leung and Hatem Ltaief and Naoya Maruyama and Newburn, {Chris J.} and Miquel Pericas",

note = "KAUST Repository Item: Exported on 2020-10-01 Acknowledgements: Authors would like to thank other PADAL14 and PADAL15 workshop participants: Maciej Besta, Jed Brown, Cy Chan, Sung-Eun Choi, Jack Choquette, Brice Goglin, Jesus Labarta, Leonidas Linardakis, Edward Luke, Satoshi Matsuoka, Peter Messmer, Lawrence Mitchell, Kathryn O'Brien, David Padua, Robert B. Ross, Marie-Christine Sawley, Robert Schreiber, Thomas Schulthess. James Sexton, Suzanne Michelle Shontz, Adrian Tate, Gysi Tobias, Engo Toshio, Mohamed Wahib, Chih-Chieh Yang. This work was partially supported by the German Research Foundation under contract TE163/17-1. This work was partially supported by the Grant 655965 by the European Commission.",

year = "2017",

month = may,

day = "12",

doi = "10.1109/TPDS.2017.2703149",

language = "English (US)",

volume = "28",

pages = "3007--3020",

journal = "IEEE Transactions on Parallel and Distributed Systems",

issn = "1045-9219",

publisher = "IEEE Computer Society",

number = "10",

}

Unat, D, Dubey, A, Hoefler, T, Shalf, J, Abraham, M, Bianco, M, Chamberlain, BL, Cledat, R, Edwards, HC, Finkel, H, Fuerlinger, K, Hannig, F, Jeannot, E, Kamil, A, Keasler, J, Kelly, PHJ, Leung, V, Ltaief, H, Maruyama, N, Newburn, CJ & Pericas, M 2017, 'Trends in Data Locality Abstractions for HPC Systems', IEEE Transactions on Parallel and Distributed Systems, vol. 28, no. 10, pp. 3007-3020. https://doi.org/10.1109/TPDS.2017.2703149

TY - JOUR

T1 - Trends in Data Locality Abstractions for HPC Systems

AU - Unat, Didem

AU - Dubey, Anshu

AU - Hoefler, Torsten

AU - Shalf, John

AU - Abraham, Mark

AU - Bianco, Mauro

AU - Chamberlain, Bradford L.

AU - Cledat, Romain

AU - Edwards, H. Carter

AU - Finkel, Hal

AU - Fuerlinger, Karl

AU - Hannig, Frank

AU - Jeannot, Emmanuel

AU - Kamil, Amir

AU - Keasler, Jeff

AU - Kelly, Paul H J

AU - Leung, Vitus

AU - Ltaief, Hatem

AU - Maruyama, Naoya

AU - Newburn, Chris J.

AU - Pericas, Miquel

N1 - KAUST Repository Item: Exported on 2020-10-01 Acknowledgements: Authors would like to thank other PADAL14 and PADAL15 workshop participants: Maciej Besta, Jed Brown, Cy Chan, Sung-Eun Choi, Jack Choquette, Brice Goglin, Jesus Labarta, Leonidas Linardakis, Edward Luke, Satoshi Matsuoka, Peter Messmer, Lawrence Mitchell, Kathryn O'Brien, David Padua, Robert B. Ross, Marie-Christine Sawley, Robert Schreiber, Thomas Schulthess. James Sexton, Suzanne Michelle Shontz, Adrian Tate, Gysi Tobias, Engo Toshio, Mohamed Wahib, Chih-Chieh Yang. This work was partially supported by the German Research Foundation under contract TE163/17-1. This work was partially supported by the Grant 655965 by the European Commission.

PY - 2017/5/12

Y1 - 2017/5/12

N2 - 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.

AB - 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.

UR - http://hdl.handle.net/10754/625984

UR - http://ieeexplore.ieee.org/document/7924389/

UR - http://www.scopus.com/inward/record.url?scp=85030168276&partnerID=8YFLogxK

U2 - 10.1109/TPDS.2017.2703149

DO - 10.1109/TPDS.2017.2703149

M3 - Article

SN - 1045-9219

VL - 28

SP - 3007

EP - 3020

JO - IEEE Transactions on Parallel and Distributed Systems

JF - IEEE Transactions on Parallel and Distributed Systems

IS - 10

ER -

Trends in Data Locality Abstractions for HPC Systems

Abstract

Bibliographical note

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this