A Grid‐Based Distributed Simulation of Plasma Turbulence

Abstract

Grid technology is widespreading, but most grid-enabled applications just exploit shared storage resources rather than computational ones, or utilize static remote allocation mechanisms of Grid platforms. In this paper the porting on a Globus equipped platform of a hierarchically distributed-shared memory parallel version of an application for particle-in-cell (PIC) simulation of plasma turbulence is described, based on the hierarchical integration of MPI and OpenMP, and originally developed for generic (non Grid) clusters of SMP nodes. INTRODUCTIONGrid technology is gaining more and more widespread diffusion within scientific community. Despite the original motivations behind many Grid initiatives, that is sharing and cooperatively using computational resources scattered through the globe in a transparent way with respect to their physical location, most grid-enabled applications just exploit shared storage resources rather than computational ones, or utilize static remote allocation mechanisms in order to transparently select and allocate sequential or parallel tasks, running, in any case, on a single grid node. Nowadays most Grid platforms are able to present computational and storage resources spread all over the world and managed by different entities, as they were a single, virtually homogeneous, parallel machine. Parallel tasks and applications could, at least in principle, be managed and executed over widespread, heterogeneous platforms.