Parallel programming using shared objects
Parallel computers come in two varieties: those with shared memory and those without. The former are hard to build; the latter are hard to program. In this paper we propose a hybrid form that combines the best properties of each. The basic idea is to allow programmers to define objects upon-which us...
Main Authors: | , , |
---|---|
Other Authors: | |
Format: | Text |
Language: | English |
Published: |
1992
|
Subjects: | |
Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.138.5223 http://www.cs.vu.nl/~ast/publications/computer-1992.pdf |
id |
ftciteseerx:oai:CiteSeerX.psu:10.1.1.138.5223 |
---|---|
record_format |
openpolar |
spelling |
ftciteseerx:oai:CiteSeerX.psu:10.1.1.138.5223 2023-05-15T17:53:48+02:00 Parallel programming using shared objects Andrew S. Tanenbaum M. Frans Kaashoek Henri E. Bal The Pennsylvania State University CiteSeerX Archives 1992 application/pdf http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.138.5223 http://www.cs.vu.nl/~ast/publications/computer-1992.pdf en eng http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.138.5223 http://www.cs.vu.nl/~ast/publications/computer-1992.pdf Metadata may be used without restrictions as long as the oai identifier remains attached to it. http://www.cs.vu.nl/~ast/publications/computer-1992.pdf AND BROADCASTING text 1992 ftciteseerx 2016-01-07T14:48:46Z Parallel computers come in two varieties: those with shared memory and those without. The former are hard to build; the latter are hard to program. In this paper we propose a hybrid form that combines the best properties of each. The basic idea is to allow programmers to define objects upon-which userdefined operations are performed, in effect, abstract data types. Each object is replicated on each machine that needs it. Reads are done locally, with no network traffic. Writes are done by a reliable broadcast algorithm. A language for parallel programming, Orca, based on distributed shared objects has been designed, implemented, and used for some applications. Its implementation uses the reliable broadcast mechanism. For applications with a high read/write ratio to the shared objects, we show that our approach can frequently achieve close to linear speedup with up to 16 processors. BLURB FOR CENTER COLUMN Parallel computers come in two varieties: those with shared memory and those without. The former are hard to build; the latter are hard to program. In this paper we propose a hybrid form that combines the best properties of each. Text Orca Unknown |
institution |
Open Polar |
collection |
Unknown |
op_collection_id |
ftciteseerx |
language |
English |
topic |
AND BROADCASTING |
spellingShingle |
AND BROADCASTING Andrew S. Tanenbaum M. Frans Kaashoek Henri E. Bal Parallel programming using shared objects |
topic_facet |
AND BROADCASTING |
description |
Parallel computers come in two varieties: those with shared memory and those without. The former are hard to build; the latter are hard to program. In this paper we propose a hybrid form that combines the best properties of each. The basic idea is to allow programmers to define objects upon-which userdefined operations are performed, in effect, abstract data types. Each object is replicated on each machine that needs it. Reads are done locally, with no network traffic. Writes are done by a reliable broadcast algorithm. A language for parallel programming, Orca, based on distributed shared objects has been designed, implemented, and used for some applications. Its implementation uses the reliable broadcast mechanism. For applications with a high read/write ratio to the shared objects, we show that our approach can frequently achieve close to linear speedup with up to 16 processors. BLURB FOR CENTER COLUMN Parallel computers come in two varieties: those with shared memory and those without. The former are hard to build; the latter are hard to program. In this paper we propose a hybrid form that combines the best properties of each. |
author2 |
The Pennsylvania State University CiteSeerX Archives |
format |
Text |
author |
Andrew S. Tanenbaum M. Frans Kaashoek Henri E. Bal |
author_facet |
Andrew S. Tanenbaum M. Frans Kaashoek Henri E. Bal |
author_sort |
Andrew S. Tanenbaum |
title |
Parallel programming using shared objects |
title_short |
Parallel programming using shared objects |
title_full |
Parallel programming using shared objects |
title_fullStr |
Parallel programming using shared objects |
title_full_unstemmed |
Parallel programming using shared objects |
title_sort |
parallel programming using shared objects |
publishDate |
1992 |
url |
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.138.5223 http://www.cs.vu.nl/~ast/publications/computer-1992.pdf |
genre |
Orca |
genre_facet |
Orca |
op_source |
http://www.cs.vu.nl/~ast/publications/computer-1992.pdf |
op_relation |
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.138.5223 http://www.cs.vu.nl/~ast/publications/computer-1992.pdf |
op_rights |
Metadata may be used without restrictions as long as the oai identifier remains attached to it. |
_version_ |
1766161499464663040 |