DMCA
MagPIe: MPI’s Collective Communication Operations for Clustered Wide Area Systems (1999)
Cached
Download Links
| Venue: | Proc PPoPP'99 |
| Citations: | 172 - 27 self |
BibTeX
@INPROCEEDINGS{Kielmann99magpie:mpi’s,
author = {Thilo Kielmann and Rutger F. H. Hofman and Henri E. Bal and Aske Plaat and Raoul A. F. Bhoedjang},
title = {MagPIe: MPI’s Collective Communication Operations for Clustered Wide Area Systems},
booktitle = {Proc PPoPP'99},
year = {1999},
pages = {131--140},
publisher = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Writing parallel applications for computational grids is a challenging task. To achieve good performance, algorithms designed for local area networks must be adapted to the differences in link speeds. An important class of algorithms are collective operations, such as broadcast and reduce. We have developed MAGPIE, a library of collective communication operations optimized for wide area systems. MAGPIE's algorithms send the minimal amount of data over the slow wide area links, and only incur a single wide area latency. Using our system, existing MPI applications can be run unmodified on geographically distributed systems. On moderate cluster sizes, using a wide area latency of 10 milliseconds and a bandwidth of 1 MByte/s, MAGPIE executes operations up to 10 times faster than MPICH, a widely used MPI implementation; application kernels improve by up to a factor of 4. Due to the structure of our algorithms, MAGPIE's advantage increases for higher wide area latencies.
Keyphrases
collective communication operation clustered wide area system wide area latency collective operation important class link speed single wide area latency slow wide area link good performance mpi application moderate cluster size advantage increase computational grid challenging task application kernel mpi implementation local area network minimal amount parallel application wide area system
