Homogeneous and Heterogeneous Networks of Workstations : Message Passing Overhead
Identifieur interne : 001984 ( Crin/Corpus ); précédent : 001983; suivant : 001985Homogeneous and Heterogeneous Networks of Workstations : Message Passing Overhead
Auteurs : E. Dillon ; C. Gamboa Dos Santos ; J. GuyardSource :
English descriptors
- KwdEn :
Abstract
On account of the growing number of workstations networks, many communication libraries are now also used outside the parallel computers area. Among those communication libraries, Message Passing Interface (MPI) appears to be an attempt of standardisation for distributed-memory computing systems. However, most people still prefer the other libraries, arguing about MPI's efficiency. The aim of our study is to evaluate the overhead introduced by these different communication libraries and to point out the cost of heterogeneous workstations' use. We tested the well-known de facto standards like PVM or P4, and some MPI implementations available in the public domain, on homogeneous and heterogeneous environments. Our results show that MPI implementations, can also achieve efficient results as good as PVM, but that a stronger heterogeneous support will be needed in the future, because of the increasing speed of networks.
Links to Exploration step
CRIN:dillon95bLe document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" wicri:score="51">Homogeneous and Heterogeneous Networks of Workstations : Message Passing Overhead</title></titleStmt><publicationStmt><idno type="RBID">CRIN:dillon95b</idno><date when="1995" year="1995">1995</date><idno type="wicri:Area/Crin/Corpus">001984</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en">Homogeneous and Heterogeneous Networks of Workstations : Message Passing Overhead</title><author><name sortKey="Dillon, E" sort="Dillon, E" uniqKey="Dillon E" first="E." last="Dillon">E. Dillon</name></author><author><name sortKey="Gamboa Dos Santos, C" sort="Gamboa Dos Santos, C" uniqKey="Gamboa Dos Santos C" first="C." last="Gamboa Dos Santos">C. Gamboa Dos Santos</name></author><author><name sortKey="Guyard, J" sort="Guyard, J" uniqKey="Guyard J" first="J." last="Guyard">J. Guyard</name></author></analytic></biblStruct></sourceDesc></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>MPI implementation</term><term>message passing</term><term>performance</term></keywords></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en" wicri:score="2269">On account of the growing number of workstations networks, many communication libraries are now also used outside the parallel computers area. Among those communication libraries, Message Passing Interface (MPI) appears to be an attempt of standardisation for distributed-memory computing systems. However, most people still prefer the other libraries, arguing about MPI's efficiency. The aim of our study is to evaluate the overhead introduced by these different communication libraries and to point out the cost of heterogeneous workstations' use. We tested the well-known de facto standards like PVM or P4, and some MPI implementations available in the public domain, on homogeneous and heterogeneous environments. Our results show that MPI implementations, can also achieve efficient results as good as PVM, but that a stronger heterogeneous support will be needed in the future, because of the increasing speed of networks.</div></front></TEI><BibTex type="inproceedings"><ref>dillon95b</ref><crinnumber>95-R-238</crinnumber><category>3</category><equipe>RESEDAS</equipe><author><e>Dillon, E.</e><e>Gamboa Dos Santos, C.</e><e>Guyard, J.</e></author><title>Homogeneous and Heterogeneous Networks of Workstations : Message Passing Overhead</title><booktitle>{Proceedings MPI Developpers'Conference, University of Notre-Dame (IN,USA)}</booktitle><year>1995</year><month>jun</month><keywords><e>performance</e><e>message passing</e><e>MPI implementation</e></keywords><abstract>On account of the growing number of workstations networks, many communication libraries are now also used outside the parallel computers area. Among those communication libraries, Message Passing Interface (MPI) appears to be an attempt of standardisation for distributed-memory computing systems. However, most people still prefer the other libraries, arguing about MPI's efficiency. The aim of our study is to evaluate the overhead introduced by these different communication libraries and to point out the cost of heterogeneous workstations' use. We tested the well-known de facto standards like PVM or P4, and some MPI implementations available in the public domain, on homogeneous and heterogeneous environments. Our results show that MPI implementations, can also achieve efficient results as good as PVM, but that a stronger heterogeneous support will be needed in the future, because of the increasing speed of networks.</abstract></BibTex></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Crin/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001984 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Crin/Corpus/biblio.hfd -nk 001984 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Wicri/Lorraine
|area= InforLorV4
|flux= Crin
|étape= Corpus
|type= RBID
|clé= CRIN:dillon95b
|texte= Homogeneous and Heterogeneous Networks of Workstations : Message Passing Overhead
}}
| This area was generated with Dilib version V0.6.33. |  |