Reliable performance prediction for parallel scientific software in a multi-cluster grid environment

We propose a model for describing and predicting the performance of practical parallel engineering numerical software on a multi-cluster environment with different distributed memory architectures. The goal of the model is to allow reliable predictions to be made as to the execution time of a given code on a large number of processors of a given parallel system, by only benchmarking the code on small numbers of processors. The model is tested using a a practical engineering multilevel code. Despite its simplicity the model demonstrates to be accurate and robust with respect the cluster architectures considered.