Channel interface: A primitive model for memory efficient communication

Takeshi Nanri, Takeshi Soga, Yuichiro Ajima, Yoshiyuki Morie, Hiroaki Honda, Taizo Kobayashi, Toshiya Takami, Shinji Sumimoto

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

3 Citations (Scopus)

Abstract

Though the size of the system is getting larger towards exa-scale computation, the amount of available memory on computing nodes is expected to remain the same or to decrease. Therefore, memory efficiency is becoming an important issue for achieving scalability. This paper pointed out the problem of memory-inefficiency in the de-facto standard parallel programming model, Message Passing Interface (MPI). To solve this problem, the channel interface was introduced in the paper. This interface enables the programmers to appropriately allocate and de-allocate channels so that the program consumes just-enough amount of memory for communication. In addition to that, by limiting the message transfer supported by a channel as simple as possible, the memory consumption and the overhead for handling messages with this interface can be minimal. This paper showed a sample implementation of this interface. Then, the memory efficiency of the implementation is examined by the models of the memory consumption and the performance.

Original languageEnglish
Title of host publicationProceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015
EditorsJohan Lilius, Masoud Daneshtalab, Mats Brorsson, Ville Leppanen, Marco Aldinucci
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages177-181
Number of pages5
ISBN (Electronic)9781479984909
DOIs
Publication statusPublished - Jan 1 2015
Event23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015 - Turku, Finland
Duration: Mar 4 2015Mar 6 2015

Publication series

NameProceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015

Other

Other23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015
CountryFinland
CityTurku
Period3/4/153/6/15

Fingerprint

Interfaces (computer)
Data storage equipment
Communication
Parallel programming
Message passing
Scalability

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Computer Networks and Communications

Cite this

Nanri, T., Soga, T., Ajima, Y., Morie, Y., Honda, H., Kobayashi, T., ... Sumimoto, S. (2015). Channel interface: A primitive model for memory efficient communication. In J. Lilius, M. Daneshtalab, M. Brorsson, V. Leppanen, & M. Aldinucci (Eds.), Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015 (pp. 177-181). [7092717] (Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/PDP.2015.83

Channel interface : A primitive model for memory efficient communication. / Nanri, Takeshi; Soga, Takeshi; Ajima, Yuichiro; Morie, Yoshiyuki; Honda, Hiroaki; Kobayashi, Taizo; Takami, Toshiya; Sumimoto, Shinji.

Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015. ed. / Johan Lilius; Masoud Daneshtalab; Mats Brorsson; Ville Leppanen; Marco Aldinucci. Institute of Electrical and Electronics Engineers Inc., 2015. p. 177-181 7092717 (Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015).

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

Nanri, T, Soga, T, Ajima, Y, Morie, Y, Honda, H, Kobayashi, T, Takami, T & Sumimoto, S 2015, Channel interface: A primitive model for memory efficient communication. in J Lilius, M Daneshtalab, M Brorsson, V Leppanen & M Aldinucci (eds), Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015., 7092717, Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015, Institute of Electrical and Electronics Engineers Inc., pp. 177-181, 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015, Turku, Finland, 3/4/15. https://doi.org/10.1109/PDP.2015.83
Nanri T, Soga T, Ajima Y, Morie Y, Honda H, Kobayashi T et al. Channel interface: A primitive model for memory efficient communication. In Lilius J, Daneshtalab M, Brorsson M, Leppanen V, Aldinucci M, editors, Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015. Institute of Electrical and Electronics Engineers Inc. 2015. p. 177-181. 7092717. (Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015). https://doi.org/10.1109/PDP.2015.83
Nanri, Takeshi ; Soga, Takeshi ; Ajima, Yuichiro ; Morie, Yoshiyuki ; Honda, Hiroaki ; Kobayashi, Taizo ; Takami, Toshiya ; Sumimoto, Shinji. / Channel interface : A primitive model for memory efficient communication. Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015. editor / Johan Lilius ; Masoud Daneshtalab ; Mats Brorsson ; Ville Leppanen ; Marco Aldinucci. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 177-181 (Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015).
@inproceedings{7013ace8c3fd41d7aee87d47f4b9e18d,
title = "Channel interface: A primitive model for memory efficient communication",
abstract = "Though the size of the system is getting larger towards exa-scale computation, the amount of available memory on computing nodes is expected to remain the same or to decrease. Therefore, memory efficiency is becoming an important issue for achieving scalability. This paper pointed out the problem of memory-inefficiency in the de-facto standard parallel programming model, Message Passing Interface (MPI). To solve this problem, the channel interface was introduced in the paper. This interface enables the programmers to appropriately allocate and de-allocate channels so that the program consumes just-enough amount of memory for communication. In addition to that, by limiting the message transfer supported by a channel as simple as possible, the memory consumption and the overhead for handling messages with this interface can be minimal. This paper showed a sample implementation of this interface. Then, the memory efficiency of the implementation is examined by the models of the memory consumption and the performance.",
author = "Takeshi Nanri and Takeshi Soga and Yuichiro Ajima and Yoshiyuki Morie and Hiroaki Honda and Taizo Kobayashi and Toshiya Takami and Shinji Sumimoto",
year = "2015",
month = "1",
day = "1",
doi = "10.1109/PDP.2015.83",
language = "English",
series = "Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "177--181",
editor = "Johan Lilius and Masoud Daneshtalab and Mats Brorsson and Ville Leppanen and Marco Aldinucci",
booktitle = "Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015",
address = "United States",

}

TY - GEN

T1 - Channel interface

T2 - A primitive model for memory efficient communication

AU - Nanri, Takeshi

AU - Soga, Takeshi

AU - Ajima, Yuichiro

AU - Morie, Yoshiyuki

AU - Honda, Hiroaki

AU - Kobayashi, Taizo

AU - Takami, Toshiya

AU - Sumimoto, Shinji

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Though the size of the system is getting larger towards exa-scale computation, the amount of available memory on computing nodes is expected to remain the same or to decrease. Therefore, memory efficiency is becoming an important issue for achieving scalability. This paper pointed out the problem of memory-inefficiency in the de-facto standard parallel programming model, Message Passing Interface (MPI). To solve this problem, the channel interface was introduced in the paper. This interface enables the programmers to appropriately allocate and de-allocate channels so that the program consumes just-enough amount of memory for communication. In addition to that, by limiting the message transfer supported by a channel as simple as possible, the memory consumption and the overhead for handling messages with this interface can be minimal. This paper showed a sample implementation of this interface. Then, the memory efficiency of the implementation is examined by the models of the memory consumption and the performance.

AB - Though the size of the system is getting larger towards exa-scale computation, the amount of available memory on computing nodes is expected to remain the same or to decrease. Therefore, memory efficiency is becoming an important issue for achieving scalability. This paper pointed out the problem of memory-inefficiency in the de-facto standard parallel programming model, Message Passing Interface (MPI). To solve this problem, the channel interface was introduced in the paper. This interface enables the programmers to appropriately allocate and de-allocate channels so that the program consumes just-enough amount of memory for communication. In addition to that, by limiting the message transfer supported by a channel as simple as possible, the memory consumption and the overhead for handling messages with this interface can be minimal. This paper showed a sample implementation of this interface. Then, the memory efficiency of the implementation is examined by the models of the memory consumption and the performance.

UR - http://www.scopus.com/inward/record.url?scp=84962881149&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84962881149&partnerID=8YFLogxK

U2 - 10.1109/PDP.2015.83

DO - 10.1109/PDP.2015.83

M3 - Conference contribution

AN - SCOPUS:84962881149

T3 - Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015

SP - 177

EP - 181

BT - Proceedings - 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2015

A2 - Lilius, Johan

A2 - Daneshtalab, Masoud

A2 - Brorsson, Mats

A2 - Leppanen, Ville

A2 - Aldinucci, Marco

PB - Institute of Electrical and Electronics Engineers Inc.

ER -