© University of Kent - Contact | Feedback | Legal
The University of Kent, Canterbury, Kent, CT2 7NZ, T +44 (0)1227 764000
Distributing process-oriented programs across a cluster of machines requires careful attention to the effects of network latency. The MPI standard, widely used for cluster computation, defines a number of collective operations: efficient, reusable algorithms for performing operations among a group of machines in the cluster. In this paper, we describe our techniques for implementing MPI communication patterns in process-oriented languages, and how we have used them to implement collective operations in PyCSP and occam-pi on top of an asynchronous messaging framework. We show how to make use of collective operations in distributed processoriented applications. We also show how the process-oriented model can be used to increase concurrency in existing collective operation algorithms.
Download publication 173 kbytes (PDF)
@inproceedings{2912,
author = {John Markus Bjørndalen and Adam T. Sampson},
title = {Process-Oriented Collective Operations},
month = {September},
year = {2008},
pages = {309--328},
keywords = {occam-pi python pycsp mpi distributed},
note = {},
doi = {},
url = {http://www.cs.kent.ac.uk/pubs/2008/2912},
publication_type = {inproceedings},
submission_id = {4613_1243431344},
ISBN = {978-1-58603-907-3},
booktitle = {Communicating Process Architectures 2008},
editor = {Peter H. Welch and Susan Stepney and Fiona A.C. Polack and Frederick R.M. Barnes and Alistair A. McEwan and Gardner S. Stiles and Jan F. Broenink and Adam T. Sampson},
volume = {66},
series = {Concurrent Systems Engineering},
address = {Amsterdam, The Netherlands},
organization = {WoTUG},
publisher = {IOS Press},
ISSN = {1383-7575},
refereed = {yes},
}