Maciej Malawski 1 Micha Placek 3 Marian Bubak
Maciej Malawski 1, Michał Placek 3, Marian Bubak 1, 2 of Computer Science AGH, Mickiewicza 30, 30 -059 Kraków, Poland 2 Academic Computer Centre CYFRONET, Nawojki 11, 30 -950 Kraków, Poland 3 Faculty of Physics and Applied Computer Science AGH Al. Mickiewicza 30, 30 -059 Krakow, Poland {bubak, malawski}@agh. edu. pl, placek@fatcat. ftj. agh. edu. pl 1 Institute • Clusters of atoms – Very interesting forms between isolated atoms or molecules and solid state – Important for the technology of constructing nanoscale devices. • Modeling of clusters – Several energy minimization methods such as • MOCCA – Common Component Architecture compliant distributed framework – Based on H 2 O resource sharing platform • Features: – Facilitated deployment - easy mechanisms for creation of components on distributed shared resources - using H 2 O; MDSA or L-BFGS, – Choosing an empirical potential – Highly compute-intensive – The optimal result depends on the number of possible iterations and initial configurations for each simulation run. – Efficient communication - both for distributed and local From sequential code to distributed components Component application distributed on multiple H 2 O kernels components – using RMIX; – Flexible - allow flexible configuration of components and various application scenarios; – Support native components, i. e. components written in non-Java programming languages and compiled for specific architecture – on-going work • Advantages of component-based approach – Flexibility of composition: from local to distributed configurations – Additional minimization methods pluggable as components – Multiple inputs and outputs possible: text file or GUI (future Example application deployment scenario work) • Experiences with distributed environment – Multiple annealing components running over many machines – Support for multiple ports and connections in MOCCA • Future improvements – From static do dynamic deployment configuration – Tests in Peer-to-Peer environment – Application performance tuning – Native components Performance tests on a PC cluster – Athlon MP 1800 MHz – 8 CPUs – Fast Ethernet – SUN Java J 2 SE 1. 4. 2 Example results References 1. 2. 3. 4. European Research Network on Foundations, Software Infrastructures and Applications for Large Scale Distributed, GRID and Peer-to-Peer Technologies. http: //www. coregrid. net/ M. Malawski, D. Kurzyniec, V. Sunderam, MOCCA - Towards a Distributed CCA Framework for Metacomputing, Proceedings of 19 th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Joint Workshop - HIPS-HPGC, April 4 -8, 2005, Denver, Colorado, USA, IEEE Computer Society Press, 2005, pp. 174 a. N. T. Wilson and R. L. Johnston: Modeling Gold clusters with an Empirical Many-body Potential, Eur. Phys. J. D 12, 161 -169 (2000) CCA forum. The Common Component Architecture (CCA) Forum home page, 2005, http: //www. cca-forum. org/. http: //www. icsr. agh. edu. pl/mambo/mocca This research is partly funded by the European Commission Project „Core. GRID”
- Slides: 1