Single sided MPI implementations for SUN MPIr
Proceedings of the 2000 ACM/IEEE conference on Supercomputing
MPI-The Complete Reference, Volume 1: The MPI Core
MPI-The Complete Reference, Volume 1: The MPI Core
An Architecture for Using Multiple Communication Devices in a MPI Library
HPCN Europe 2000 Proceedings of the 8th International Conference on High-Performance Computing and Networking
Flattening on the Fly: Efficient Handling of MPI Derived Datatypes
Proceedings of the 6th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
Implementing MPI's One-Sided Communications for WMPI
Proceedings of the 6th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
Runtime Checking of Datatype Signatures in MPI
Proceedings of the 7th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
MPI-2 Process Creation & Management Implementation for NT Clusters
Proceedings of the 7th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
WMPI - Message Passing Interface for Win32 Clusters
Proceedings of the 5th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
| Hi-index | 0.00 |
This paper describes the implementation of MPI-2 one-sided communications (OSC) for the forthcoming WMPI II product, aimed at clusters of workstations (Windows workstations, for now). This implementation is layered directly on top of the WMPI Management Layer (WML), rather than being on top of the MPI layer and as such can draw more performance from the new features of the WMPI's WML. The major features of this implementation are presented, including the synchronization operations, the remote memory operations and the datatype handling mechanism. Performance benchmarks were taken, comparing the message passing and the one-sided communication models, as well as to compare this implementation with one layered on top of MPI.