Trends in Data Locality Abstractions for HPC Systems

被引：60

作者：

Unat, Didem ^{[1
]}

Dubey, Anshu ^{[2
]}

Hoefler, Torsten ^{[3
]}

Shalf, John ^{[4
]}

Abraham, Mark ^{[5
]}

Bianco, Mauro ^{[6
]}

Chamberlain, Bradford L. ^{[7
]}

Cledat, Romain ^{[8
]}

Edwards, H. Carter ^{[9
]}

Finkel, Hal ^{[10
]}

Fuerlinger, Karl ^{[11
]}

Hannig, Frank ^{[12
]}

Jeannot, Emmanuel ^{[13
]}

Kamil, Amir ^{[14
,15
]}

Keasler, Jeff ^{[16
]}

Kelly, Paul H. J. ^{[17
]}

Leung, Vitus ^{[9
]}

Ltaief, Hatem ^{[18
]}

Maruyama, Naoya ^{[19
]}

Newburn, Chris J. ^{[20
]}

Pericas, Miquel ^{[21
]}

机构：

[1] Koc Univ, Dept Comp Engn, TR-34450 Istanbul, Turkey

[2] Argonne Natl Lab, 9700 S Cass Ave, Argonne, IL 60439 USA

[3] ETH, CH-8092 Zurich, Switzerland

[4] Lawrence Berkeley Natl Lab, Berkeley, CA 94720 USA

[5] KTH Royal Inst Technol, S-17121 Solna, Sweden

[6] Swiss Natl Supercomp Ctr, CH-6900 Lugano, Switzerland

[7] Cray Inc, Seattle, WA 98164 USA

[8] Intel Cooperat, Santa Clara, CA 95050 USA

[9] Sandia Natl Labs, POB 5800, Albuquerque, NM 87185 USA

[10] Argonne Natl Lab, 9700 S Cass Ave, Argonne, IL 60439 USA

[11] Ludwig Maximilians Univ Munchen, D-80538 Munich, Germany

[12] Univ Erlangen Nurnberg, D-91058 Erlangen, Germany

[13] INRIA Bordeaux Sud Ouest, F-33405 Talence, France

[14] Univ Michigan, Ann Arbor, MI 48109 USA

[15] Lawrence Berkeley Natl Lab, Berkeley, CA 94720 USA

[16] Lawrence Livermore Natl Lab, Livermore, CA 94550 USA

[17] Imperial Coll London, Software Technol, London, England

[18] King Abdullah Univ Sci & Technol, Thuwal 23955, Saudi Arabia

[19] RIKEN, Kobe, Hyogo 6500047, Japan

[20] Nvidia Corp, Santa Clara, CA 95050 USA

[21] Chalmers Univ Technol, S-41296 Gothenburg, Sweden

来源：

IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS | 2017年 / 28卷 / 10期

基金：

英国工程与自然科学研究理事会;

关键词：

Data locality; programming abstractions; high-performance computing; data layout; locality-aware runtimes;

D O I：

10.1109/TPDS.2017.2703149

中图分类号：

TP301 [理论、方法];

学科分类号：

081202 ;

摘要：

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.

引用

页码：3007 / 3020

页数：14

共 50 条

[41] A Data Driven Scheduling Approach for Power Management on HPC Systems
Wallace, Sean
Yang, Xu
Vishwanath, Venkatram
Allcock, William E.
Coghlan, Susan
Papka, Michael E.
Lan, Zhiling
SC '16: PROCEEDINGS OF THE INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS, 2016, : 656 - 666
[42] On the Overhead of Topology Discovery for Locality-aware Scheduling in HPC
Goglin, Brice
2017 25TH EUROMICRO INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED AND NETWORK-BASED PROCESSING (PDP 2017), 2017, : 186 - 190
[43] VLSI SIMULATION AND DATA ABSTRACTIONS
KATZENELSON, J
WEITZ, E
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, 1986, 5 (03) : 371 - 378
[44] Data Jockey: Automatic Data Management for HPC Multi-Tiered Storage Systems
Shin, Woong
Brumgard, Christopher D.
Xie, Bing
Vazhkudai, Sudharshan S.
Ghoshal, Devarshi
Oral, Sarp
Ramakrishnan, Lavanya
2019 IEEE 33RD INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS 2019), 2019, : 511 - 522
[45] A Library for Portable and Composable Data Locality Optimizations for NUMA Systems
Majo Z.
Gross T.R.
2017, Association for Computing Machinery, 2 Penn Plaza, Suite 701, New York, NY 10121-0701, United States (03)
[46] A Library for Portable and Composable Data Locality Optimizations for NUMA Systems
Majo, Zoltan
Gross, Thomas R.
ACM SIGPLAN NOTICES, 2015, 50 (08) : 227 - 238
[47] HPC in the Automotive Industry Trends and Prospects
Guembel, Christoph
IT-INFORMATION TECHNOLOGY, 2013, 55 (03): : 102 - 104
[48] The use of locality information on data intensive parallel file systems
Sugawara Junior, Ricardo Ryoiti
Sato, Liria Matsumoto
2013 IEEE 16TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND ENGINEERING (CSE 2013), 2013, : 167 - 173
[49] Abstractions for dynamic data distribution
Deltz, SJ
Chamberlain, BL
Snyder, L
NINTH INTERNATIONAL WORKSHOP ON HIGH-LEVEL PARALLEL PROGRAMMING MODELS AND SUPPORTIVE ENVIRONMENTS, PROCEEDINGS, 2004, : 42 - 51
[50] Programming with shared data abstractions
Dobson, S
Goodeve, D
SOLVING IRREGULARLY STRUCTURED PROBLEMS IN PARALLEL, 1997, 1253 : 93 - 102

← 1 2 3 4 5 →