Design and Implementation of Open MPI over Quadrics/Elan4

Publisher Website
Google Scholar
Abstract
Open MPI is a project recently initiated to provide a fault-tolerant, multi-network capable implementation of MPI-2 (1997), based on experiences gained from FT-MPI (G. E. Fagg et al., 2003), LA-MPI (R. L. Graham et al., 2003), LAM/MPI (J. Squyers et al., 2003), and MVAPICH projects. Its initial communication architecture is layered on top of TCP/IP. In this paper, we have designed and implemented open MPI point-to-point layer on top of a high-end interconnect, Quadrics/Elan4. The restriction of Quadrics static process model has been overcome to accommodate the requirement of MPI-2 dynamic process management. Quadrics queued-based direct memory access (QDMA) and remote direct memory access (RDMA) mechanisms have been integrated to form a low-overhead, high-performance transport layer. Lightweight asynchronous progress is made possible with a combination of Quadrics chained event and QDMA mechanisms. Experimental results indicate that the resulting point-to-point transport layer is able to achieve comparable performance to Quadrics native QDMA operations, from which it is derived. Our implementation provides an MPI-2 compliant message passing library over Quadrics/Elan4 with a performance comparable to MPICH-Quadrics.

This publication has 8 references indexed in Scilit: