TY - JOUR
T1 - An MPI prototype for compiled communication on Ethernet switched clusters
AU - Karwande, Amit
AU - Yuan, Xin
AU - Lowenthal, David K.
N1 - Funding Information:
This work was partially supported by NSF Grants, ANI-0106706, CCR-0208892, CCR-0342540. ∗ Corresponding author. Tel.: +850 644 9133; fax: +850 644 0058. E-mail addresses: [email protected] (A. Karwande), [email protected] (X. Yuan), d [email protected] u (D.K. Lowenthal).
PY - 2005/10
Y1 - 2005/10
N2 - Compiled communication has recently been proposed to improve communication performance for clusters of workstations. The idea of compiled communication is to apply more aggressive optimizations to communications whose information is known at compile time. Existing MPI libraries do not support compiled communication. In this paper, we present an MPI prototype, CC-MPI, that supports compiled communication on Ethernet switched clusters. The unique feature of CC-MPI is that it allows the user to manage network resources such as multicast groups directly and to optimize communications based on the availability of the communication information. CC-MPI optimizes one-to-all, one-to-many, all-to-all, and many-to-many collective communication routines using the compiled communication technique. We describe the techniques used in CC-MPI and report its performance.The results show that communication performance of Ethernet switched clusters can be significantly improved through compiled communication.
AB - Compiled communication has recently been proposed to improve communication performance for clusters of workstations. The idea of compiled communication is to apply more aggressive optimizations to communications whose information is known at compile time. Existing MPI libraries do not support compiled communication. In this paper, we present an MPI prototype, CC-MPI, that supports compiled communication on Ethernet switched clusters. The unique feature of CC-MPI is that it allows the user to manage network resources such as multicast groups directly and to optimize communications based on the availability of the communication information. CC-MPI optimizes one-to-all, one-to-many, all-to-all, and many-to-many collective communication routines using the compiled communication technique. We describe the techniques used in CC-MPI and report its performance.The results show that communication performance of Ethernet switched clusters can be significantly improved through compiled communication.
KW - Clusters of workstations
KW - Collective communication
KW - Communication optimization
KW - Compiled communication
KW - MPI
UR - http://www.scopus.com/inward/record.url?scp=24144487162&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=24144487162&partnerID=8YFLogxK
U2 - 10.1016/j.jpdc.2005.04.014
DO - 10.1016/j.jpdc.2005.04.014
M3 - Article
AN - SCOPUS:24144487162
SN - 0743-7315
VL - 65
SP - 1123
EP - 1133
JO - Journal of Parallel and Distributed Computing
JF - Journal of Parallel and Distributed Computing
IS - 10
ER -