Skip to Main Content
The global arrays toolkit is a library that allows programmers to write parallel programs that use large arrays distributed across processing nodes. The toolkit provides advantage of one-sided communication operations through the aggregate remote memory copy interface (ARMCI). In this study, we investigate the performance of a parallel application implemented with the toolkit on cluster and grid computing environments as well as comparison of performance of same application implemented with MPI library. In cluster computing environment, the performance of the evaluating application implemented with the toolkit outperforms MPICH implementation when message size is large enough. However, for grid computing environment with lower bandwidth and high latency conditions, the performance of the evaluating application implemented with the toolkit is significantly dropped compared to that of the MPICH-G2 implementation.