作者: P.K. Jimack , G. Romanazzi , C.E. Goodyer
DOI:
关键词: Cluster (spacecraft) 、 Code (cryptography) 、 Simplicity 、 Distributed memory 、 Parallel computing 、 Multi cluster 、 Benchmarking 、 Computer science 、 Numerical analysis 、 Performance prediction
摘要: We propose a model for describing and predicting the performance of practical parallel engineering numerical software on multi-cluster environment with different distributed memory architectures. The goal is to allow reliable predictions to be made as execution time given code large number processors of a parallel system, by only benchmarking small numbers processors. The tested using engineering multilevel code. Despite its simplicity demonstrates accurate robust respect cluster architectures considered.