Molecular dynamics using p-threads
G Chillemi and N Sanna, HIGH-PERFORMANCE COMPUTING AND NETWORKING, 1401, 875-877 (1998).
In this paper we report our work on a SM-MIMD architecture clustered as four SMP machines of the MPI version of the GROMACS V1.5.1. We used a cluster of Digital AS4100 interconnected with Memory Channel and Digital MPI V1.4 as hardware testbed. We developed a new threaded version of MPI GROMACS and it permitted us to measure perfomance and scalability of the new code in real MD simulation runs. The new code open a way to further improve the MD simulations to date giving a way to estimate its scalability on M-SMP clustered machine and similar parallel architectures.
Return to Publications page