Towards unification of accelerated computing and interconnection for extreme-scale computing

Toshihiro Hanawa, Yuetsu Kodama, Taisuke Boku, Hideharu Amano, Hitoshi Murai, Masayuki Umemura, Mitsuhisa Sato

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

Abstract

Heterogeneous clusters using accelerators are widely used for high-performance computing system. In such systems, the inter-node communication among accelerators becomes bottleneck due to the data transfer between the accelerator and the host. To eliminate this overhead, we have been developing a novel communication system realizing direct communication among accelerators over computation nodes under the HA-PACS (Highly Accelerated Parallel Advanced system for Computational Sciences) project. Also we are investigating high-level parallel programming language, and several practical application programs on our concept, as well as studying the enhancement of TCA and developing system software stack in the CREST project.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages463-474
Number of pages12
Volume9040
ISBN (Print)9783319162133
DOIs
Publication statusPublished - 2015
Event11th International Symposium on Applied Reconfigurable Computing, ARC 2015 - Bochum, Germany
Duration: 2015 Apr 132015 Apr 17

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9040
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other11th International Symposium on Applied Reconfigurable Computing, ARC 2015
CountryGermany
CityBochum
Period15/4/1315/4/17

Fingerprint

Accelerator
Unification
Interconnection
Particle accelerators
Extremes
Computing
Computational Science
Parallel programming
Communication
Parallel Programming
Data Transfer
Data transfer
Vertex of a graph
Application programs
Computer programming languages
Software System
Communication Systems
Programming Languages
Communication systems
Eliminate

Keywords

  • Accelerator computing
  • GPU direct
  • Interconnection network
  • PCI express
  • Remote DMA

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Hanawa, T., Kodama, Y., Boku, T., Amano, H., Murai, H., Umemura, M., & Sato, M. (2015). Towards unification of accelerated computing and interconnection for extreme-scale computing. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9040, pp. 463-474). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9040). Springer Verlag. https://doi.org/10.1007/978-3-319-16214-0_43

Towards unification of accelerated computing and interconnection for extreme-scale computing. / Hanawa, Toshihiro; Kodama, Yuetsu; Boku, Taisuke; Amano, Hideharu; Murai, Hitoshi; Umemura, Masayuki; Sato, Mitsuhisa.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9040 Springer Verlag, 2015. p. 463-474 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9040).

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

Hanawa, T, Kodama, Y, Boku, T, Amano, H, Murai, H, Umemura, M & Sato, M 2015, Towards unification of accelerated computing and interconnection for extreme-scale computing. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9040, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9040, Springer Verlag, pp. 463-474, 11th International Symposium on Applied Reconfigurable Computing, ARC 2015, Bochum, Germany, 15/4/13. https://doi.org/10.1007/978-3-319-16214-0_43
Hanawa T, Kodama Y, Boku T, Amano H, Murai H, Umemura M et al. Towards unification of accelerated computing and interconnection for extreme-scale computing. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9040. Springer Verlag. 2015. p. 463-474. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16214-0_43
Hanawa, Toshihiro ; Kodama, Yuetsu ; Boku, Taisuke ; Amano, Hideharu ; Murai, Hitoshi ; Umemura, Masayuki ; Sato, Mitsuhisa. / Towards unification of accelerated computing and interconnection for extreme-scale computing. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9040 Springer Verlag, 2015. pp. 463-474 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{bf1bd27c55034d3cb86d1e4864ccf8b3,
title = "Towards unification of accelerated computing and interconnection for extreme-scale computing",
abstract = "Heterogeneous clusters using accelerators are widely used for high-performance computing system. In such systems, the inter-node communication among accelerators becomes bottleneck due to the data transfer between the accelerator and the host. To eliminate this overhead, we have been developing a novel communication system realizing direct communication among accelerators over computation nodes under the HA-PACS (Highly Accelerated Parallel Advanced system for Computational Sciences) project. Also we are investigating high-level parallel programming language, and several practical application programs on our concept, as well as studying the enhancement of TCA and developing system software stack in the CREST project.",
keywords = "Accelerator computing, GPU direct, Interconnection network, PCI express, Remote DMA",
author = "Toshihiro Hanawa and Yuetsu Kodama and Taisuke Boku and Hideharu Amano and Hitoshi Murai and Masayuki Umemura and Mitsuhisa Sato",
year = "2015",
doi = "10.1007/978-3-319-16214-0_43",
language = "English",
isbn = "9783319162133",
volume = "9040",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "463--474",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Towards unification of accelerated computing and interconnection for extreme-scale computing

AU - Hanawa, Toshihiro

AU - Kodama, Yuetsu

AU - Boku, Taisuke

AU - Amano, Hideharu

AU - Murai, Hitoshi

AU - Umemura, Masayuki

AU - Sato, Mitsuhisa

PY - 2015

Y1 - 2015

N2 - Heterogeneous clusters using accelerators are widely used for high-performance computing system. In such systems, the inter-node communication among accelerators becomes bottleneck due to the data transfer between the accelerator and the host. To eliminate this overhead, we have been developing a novel communication system realizing direct communication among accelerators over computation nodes under the HA-PACS (Highly Accelerated Parallel Advanced system for Computational Sciences) project. Also we are investigating high-level parallel programming language, and several practical application programs on our concept, as well as studying the enhancement of TCA and developing system software stack in the CREST project.

AB - Heterogeneous clusters using accelerators are widely used for high-performance computing system. In such systems, the inter-node communication among accelerators becomes bottleneck due to the data transfer between the accelerator and the host. To eliminate this overhead, we have been developing a novel communication system realizing direct communication among accelerators over computation nodes under the HA-PACS (Highly Accelerated Parallel Advanced system for Computational Sciences) project. Also we are investigating high-level parallel programming language, and several practical application programs on our concept, as well as studying the enhancement of TCA and developing system software stack in the CREST project.

KW - Accelerator computing

KW - GPU direct

KW - Interconnection network

KW - PCI express

KW - Remote DMA

UR - http://www.scopus.com/inward/record.url?scp=84926661910&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84926661910&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-16214-0_43

DO - 10.1007/978-3-319-16214-0_43

M3 - Conference contribution

AN - SCOPUS:84926661910

SN - 9783319162133

VL - 9040

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

SP - 463

EP - 474

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

PB - Springer Verlag

ER -