انجمن کامپیوتر ایران

We report on our work on improving the performance of collective operations in MPICH on clusters connected by switched networks. For each collective operation, we use multiple algorithms depending on the message size, with the goal of minimizing latency for short messages and minimizing bandwidth usage for long messages. Although we have implemented new algorithms for all MPI collective operations, because of limited space we describe only the algorithms for allgather, broadcast,reduce-scatter, and reduce. We present performance results using the SKaMPI benchmark on a Myrinet-connected Linux cluster and an IBM . In all cases, the new algorithms significantly outperform the old algorithms used in MPICH on the Myrinet cluster, and, in many cases, they outperform the algorithms used in IBM’s MPI .