Parallel Programming on Amoeba Using Efficient Distributed Shared Memory

H.E. Bal, M.F. Kaashoek, A.S. Tanenbaum

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

Programming using message passing is considered difficult and therefore many researchers have proposed a model based on distributed shared memory (DSM). Currently, there is a trend to weaken the consistency of distributed shared memory. This allows for a good performance but complicates programming of applications. We have designed a new model for distributed shared memory that is efficient and consistent. Our approach is based on efficient low-level communication protocols and on a language that allows us to use implementation techniques that existing DSM systems cannot use effectively. We have implemented our DSM system on top of the Amoeba distributed operating system using its fast remote procedure call and group communication mechanisms. Using this implementation we measured the performance of the basic costs for operations on DSM and of several parallel applications.
Original languageEnglish
Title of host publicationScientific Computing on Supercomputers III
EditorsJ.T. Devreese, P.E. van Camp
Place of PublicationBoston, MA
PublisherSpringer
Pages89-105
ISBN (Electronic)978-1-4899-2581-7
ISBN (Print)978-1-4899-2583-1
DOIs
Publication statusPublished - 1992
EventInternational Workshop on the Use of Supercomputers in Theoretical Science - Antwerp, Belgium
Duration: 24 Jan 199125 Jan 1991

Workshop

WorkshopInternational Workshop on the Use of Supercomputers in Theoretical Science
CountryBelgium
CityAntwerp
Period24/01/9125/01/91

Fingerprint

Parallel programming
Data storage equipment
Computer systems
Computer operating systems
Message passing
Computer programming
Network protocols
Communication
Costs

Cite this

Bal, H. E., Kaashoek, M. F., & Tanenbaum, A. S. (1992). Parallel Programming on Amoeba Using Efficient Distributed Shared Memory. In J. T. Devreese, & P. E. van Camp (Eds.), Scientific Computing on Supercomputers III (pp. 89-105). Boston, MA: Springer. https://doi.org/10.1007/978-1-4899-2581-7_5
Bal, H.E. ; Kaashoek, M.F. ; Tanenbaum, A.S. / Parallel Programming on Amoeba Using Efficient Distributed Shared Memory. Scientific Computing on Supercomputers III. editor / J.T. Devreese ; P.E. van Camp. Boston, MA : Springer, 1992. pp. 89-105
@inproceedings{f4dcf4a912834f94a33675a00b282d45,
title = "Parallel Programming on Amoeba Using Efficient Distributed Shared Memory",
abstract = "Programming using message passing is considered difficult and therefore many researchers have proposed a model based on distributed shared memory (DSM). Currently, there is a trend to weaken the consistency of distributed shared memory. This allows for a good performance but complicates programming of applications. We have designed a new model for distributed shared memory that is efficient and consistent. Our approach is based on efficient low-level communication protocols and on a language that allows us to use implementation techniques that existing DSM systems cannot use effectively. We have implemented our DSM system on top of the Amoeba distributed operating system using its fast remote procedure call and group communication mechanisms. Using this implementation we measured the performance of the basic costs for operations on DSM and of several parallel applications.",
author = "H.E. Bal and M.F. Kaashoek and A.S. Tanenbaum",
year = "1992",
doi = "10.1007/978-1-4899-2581-7_5",
language = "English",
isbn = "978-1-4899-2583-1",
pages = "89--105",
editor = "J.T. Devreese and {van Camp}, P.E.",
booktitle = "Scientific Computing on Supercomputers III",
publisher = "Springer",

}

Bal, HE, Kaashoek, MF & Tanenbaum, AS 1992, Parallel Programming on Amoeba Using Efficient Distributed Shared Memory. in JT Devreese & PE van Camp (eds), Scientific Computing on Supercomputers III. Springer, Boston, MA, pp. 89-105, International Workshop on the Use of Supercomputers in Theoretical Science, Antwerp, Belgium, 24/01/91. https://doi.org/10.1007/978-1-4899-2581-7_5

Parallel Programming on Amoeba Using Efficient Distributed Shared Memory. / Bal, H.E.; Kaashoek, M.F.; Tanenbaum, A.S.

Scientific Computing on Supercomputers III. ed. / J.T. Devreese; P.E. van Camp. Boston, MA : Springer, 1992. p. 89-105.

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Parallel Programming on Amoeba Using Efficient Distributed Shared Memory

AU - Bal, H.E.

AU - Kaashoek, M.F.

AU - Tanenbaum, A.S.

PY - 1992

Y1 - 1992

N2 - Programming using message passing is considered difficult and therefore many researchers have proposed a model based on distributed shared memory (DSM). Currently, there is a trend to weaken the consistency of distributed shared memory. This allows for a good performance but complicates programming of applications. We have designed a new model for distributed shared memory that is efficient and consistent. Our approach is based on efficient low-level communication protocols and on a language that allows us to use implementation techniques that existing DSM systems cannot use effectively. We have implemented our DSM system on top of the Amoeba distributed operating system using its fast remote procedure call and group communication mechanisms. Using this implementation we measured the performance of the basic costs for operations on DSM and of several parallel applications.

AB - Programming using message passing is considered difficult and therefore many researchers have proposed a model based on distributed shared memory (DSM). Currently, there is a trend to weaken the consistency of distributed shared memory. This allows for a good performance but complicates programming of applications. We have designed a new model for distributed shared memory that is efficient and consistent. Our approach is based on efficient low-level communication protocols and on a language that allows us to use implementation techniques that existing DSM systems cannot use effectively. We have implemented our DSM system on top of the Amoeba distributed operating system using its fast remote procedure call and group communication mechanisms. Using this implementation we measured the performance of the basic costs for operations on DSM and of several parallel applications.

U2 - 10.1007/978-1-4899-2581-7_5

DO - 10.1007/978-1-4899-2581-7_5

M3 - Conference contribution

SN - 978-1-4899-2583-1

SP - 89

EP - 105

BT - Scientific Computing on Supercomputers III

A2 - Devreese, J.T.

A2 - van Camp, P.E.

PB - Springer

CY - Boston, MA

ER -

Bal HE, Kaashoek MF, Tanenbaum AS. Parallel Programming on Amoeba Using Efficient Distributed Shared Memory. In Devreese JT, van Camp PE, editors, Scientific Computing on Supercomputers III. Boston, MA: Springer. 1992. p. 89-105 https://doi.org/10.1007/978-1-4899-2581-7_5