PEMPIs: A New Methodology for Modeling and Prediction of MPI Programs Performance

The evaluation and prediction of parallel program performance is becoming more and more important, so that it requires appropriate techniques to identify the factors which influence the application execution time and also the way they interact. In this paper, we present some contributions of our research in this area by describing PEMPIs: a new methodology applied to the analysis and prediction of MPI programs. A new task graph helps us both to understand details of the application and to increase the accuracy of the prediction models. The proposed techniques are detailed and tested through the modeling of a complete application. PEMPIs efficiency has been proved by the results of this application modeling - most tests executed in a cluster of workstations exhibited errors up to 10%.

[1]  Jesús Labarta,et al.  Deriving analytical models from a limited number of runs , 2003, PARCO.

[2]  Ramesh Subramonian,et al.  LogP: towards a realistic model of parallel computation , 1993, PPOPP '93.

[3]  Leslie G. Valiant,et al.  A bridging model for parallel computation , 1990, CACM.

[4]  Duncan A. Grove,et al.  Communication Benchmarking and Performance Modelling of MPI Programs on Cluster Computers , 2004, IPDPS.

[5]  Cho-Li Wang,et al.  Realistic communication model for parallel computing on cluster , 1999, ICWC 99. IEEE Computer Society International Workshop on Cluster Computing.

[6]  Luiz Angelo Steffenel,et al.  Performance characterisation of intra-cluster collective communications , 2004, 16th Symposium on Computer Architecture and High Performance Computing.

[7]  Rajkumar Buyya,et al.  Parallel Programming Models and Paradigms , 1998 .

[8]  Jr. Wagner Meira.  Modeling Performance of Parallel Programs , 1995 .

[9]  W. Press,et al.  Numerical Recipes in C++: The Art of Scientific Computing (2nd edn)1 Numerical Recipes Example Book (C++) (2nd edn)2 Numerical Recipes Multi-Language Code CD ROM with LINUX or UNIX Single-Screen License Revised Version3 , 2003 .

[10]  David B. Skillicorn,et al.  Questions and Answers about BSP , 1997, Sci. Program..

[11]  Arjan J. C. van Gemund.  Performance Modeling of Parallel Systems , 1996 .

[12]  Hélio Marci de Oliveira,et al.  Modeling Repetition Structures in MPI Programs Using a New Graphical Model , 2003, PDPTA.

[13]  Kuan-Ching Li,et al.  Modeling and Predicting Point-to-Point Communications of MPI Parallel Programs in NOW Environments , 2003, PDPTA.

[14]  Mark Crovella,et al.  Performance Prediction and Tuning of Parallel Programs , 1994 .

[15]  Message P Forum,et al.  MPI: A Message-Passing Interface Standard , 1994 .

[16]  Zhiwei Xu,et al.  Modeling communication overhead: MPI and MPL performance on the IBM SP2 , 1996, IEEE Parallel Distributed Technol. Syst. Appl..

[17]  David A. Patterson,et al.  Computer Architecture: A Quantitative Approach , 1969 .