MGMR

Multi-GPU based MapReduce

Yi Chen, Zhi Qiao, Hai Jiang, Kuan Ching Li, Won Woo Ro

Research output: Chapter in Book/Report/Conference proceedingConference contribution

13 Citations (Scopus)

Abstract

MapReduce is a programming model introduced by Google for large-scale data processing. Several studies have implemented MapReduce model on Graphic Processing Unit (GPU). However, most of them are based on the single GPU and bounded by GPU memory with inefficient atomic operations. This paper intends to develop a standalone MapReduce system, called MGMR, to utilize multiple GPUs, handle large-scale data processing beyond GPU memory limit, and eliminate serial atomic operations. Experimental results have demonstrated MGMR's effectiveness in handling large data set.

Original languageEnglish
Title of host publicationGrid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings
Pages433-442
Number of pages10
DOIs
Publication statusPublished - 2013 Sep 9
Event8th International Conference on Grid and Pervasive Computing, GPC 2013 - Seoul, Korea, Republic of
Duration: 2013 May 92013 May 11

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7861 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other8th International Conference on Grid and Pervasive Computing, GPC 2013
CountryKorea, Republic of
CitySeoul
Period13/5/913/5/11

Fingerprint

MapReduce
Graphics Processing Unit
Data storage equipment
Large Data Sets
Programming Model
Eliminate
Graphics processing unit
Experimental Results
Model

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Chen, Y., Qiao, Z., Jiang, H., Li, K. C., & Ro, W. W. (2013). MGMR: Multi-GPU based MapReduce. In Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings (pp. 433-442). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7861 LNCS). https://doi.org/10.1007/978-3-642-38027-3_46
Chen, Yi ; Qiao, Zhi ; Jiang, Hai ; Li, Kuan Ching ; Ro, Won Woo. / MGMR : Multi-GPU based MapReduce. Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings. 2013. pp. 433-442 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{3239d6b282714f119815935a3d0a87fe,
title = "MGMR: Multi-GPU based MapReduce",
abstract = "MapReduce is a programming model introduced by Google for large-scale data processing. Several studies have implemented MapReduce model on Graphic Processing Unit (GPU). However, most of them are based on the single GPU and bounded by GPU memory with inefficient atomic operations. This paper intends to develop a standalone MapReduce system, called MGMR, to utilize multiple GPUs, handle large-scale data processing beyond GPU memory limit, and eliminate serial atomic operations. Experimental results have demonstrated MGMR's effectiveness in handling large data set.",
author = "Yi Chen and Zhi Qiao and Hai Jiang and Li, {Kuan Ching} and Ro, {Won Woo}",
year = "2013",
month = "9",
day = "9",
doi = "10.1007/978-3-642-38027-3_46",
language = "English",
isbn = "9783642380266",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "433--442",
booktitle = "Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings",

}

Chen, Y, Qiao, Z, Jiang, H, Li, KC & Ro, WW 2013, MGMR: Multi-GPU based MapReduce. in Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7861 LNCS, pp. 433-442, 8th International Conference on Grid and Pervasive Computing, GPC 2013, Seoul, Korea, Republic of, 13/5/9. https://doi.org/10.1007/978-3-642-38027-3_46

MGMR : Multi-GPU based MapReduce. / Chen, Yi; Qiao, Zhi; Jiang, Hai; Li, Kuan Ching; Ro, Won Woo.

Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings. 2013. p. 433-442 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7861 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - MGMR

T2 - Multi-GPU based MapReduce

AU - Chen, Yi

AU - Qiao, Zhi

AU - Jiang, Hai

AU - Li, Kuan Ching

AU - Ro, Won Woo

PY - 2013/9/9

Y1 - 2013/9/9

N2 - MapReduce is a programming model introduced by Google for large-scale data processing. Several studies have implemented MapReduce model on Graphic Processing Unit (GPU). However, most of them are based on the single GPU and bounded by GPU memory with inefficient atomic operations. This paper intends to develop a standalone MapReduce system, called MGMR, to utilize multiple GPUs, handle large-scale data processing beyond GPU memory limit, and eliminate serial atomic operations. Experimental results have demonstrated MGMR's effectiveness in handling large data set.

AB - MapReduce is a programming model introduced by Google for large-scale data processing. Several studies have implemented MapReduce model on Graphic Processing Unit (GPU). However, most of them are based on the single GPU and bounded by GPU memory with inefficient atomic operations. This paper intends to develop a standalone MapReduce system, called MGMR, to utilize multiple GPUs, handle large-scale data processing beyond GPU memory limit, and eliminate serial atomic operations. Experimental results have demonstrated MGMR's effectiveness in handling large data set.

UR - http://www.scopus.com/inward/record.url?scp=84883407352&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84883407352&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-38027-3_46

DO - 10.1007/978-3-642-38027-3_46

M3 - Conference contribution

SN - 9783642380266

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 433

EP - 442

BT - Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings

ER -

Chen Y, Qiao Z, Jiang H, Li KC, Ro WW. MGMR: Multi-GPU based MapReduce. In Grid and Pervasive Computing - 8th International Conference, GPC 2013 and Colocated Workshops, Proceedings. 2013. p. 433-442. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-38027-3_46