Scalability Model Based on the Concept of Granularity

Abstract: In the recent years it can be observed increasing popularity of parallel processing using multi-core processors, local clusters, GPU and others. Moreover, currently one of the main requirements the IT users is the reduction of maintaining cost of the computer infrastructure. It causes that the performance evaluation of the parallel applications becomes one of the most important problem. Then obtained results allows efficient use of available resources. In traditional methods of performance evaluation the results are based on wall-clock time measurements. This approach requires consecutive application executions and includes a time-consuming data analysis. In the paper an alternative approach is proposed. The decomposition of parallel application execution time onto computation time and overheads related to parallel execution is use to calculate the granularity of application and then determine its efficiency. Finally the application scalability can be evaluates.