PRACE User Forum (PRACE) Session 1
Time and Date: 16:40 - 18:20 on 1st June 2015
Chair: Derek Groen
|742|| Understanding Scientific Application’s Performance using BSC tools [abstract]
Abstract: he current trend in supercomputer architectures is leading scientific applications to use parallel programing models such Message Passing Interface (MPI) to use computing resources properly. Understanding how these applications behave it is not straightforward and its crucial for achieving good performance and good efficiency of their codes. Here we present a real case of study of a cuttingedge scientific application. NEMO is a stateoftheart Ocean Global Circulation Model (OGCM) hundreds of users around the world. It is used for oceanographic research, operational oceanography, seasonal forecast and climate studies. In the framework of PRACE, the projects HiResClim and HiResClim II, related to High Resolution Climate projections, uses NEMO as a oceanic model and had conceded more than 38 million core hours in the 5th PRACE Regular Call for Proposals and 50 million core hours in the 7th PRACE Regular Call for Proposals, both in the Tier0 machine Marenostrum. That huge amount of computation time justifies the effort to analyze and optimize the application’s performance. Using the performance tools developed at Barcelona Supercomputing Center (BSC) it is possible to analyse the behaviour of the application . We studied different executions of the NEMO model and performed different analysis of the computational phases, analyzing how cpu and memory behaves, and also communication patterns. We also did strong and weak scaling tests to find bottlenecks constraining the scalability of the application. With this analysis, we could confirm some of the envisaged problems in previous performance analysis of the application and further see other problems not identified before. Using Paraver is both possible to see with high detail the internal behaviour of the application (we can see for example when, who and to where every message is sent) or to compute metrics to extract useful information (such parallel efficiency, load balance or many more). Dimemas allows us to simulate the behaviour of the application under different conditions. It could be useful to analyze the sensibility to network parameters, and for example it could be useful to analyze if one application could run properly in cloud computing. Other tools being developed at BSC and used in this work are Clustering and Folding. The clustering tool uses a data mining technique to identify regions of code with similar performance trends. This make possible to group together and study different iterations, using the folding tool, in order to get instantaneous performance metrics inside the routines, finding areas of interest that have a poor hardware usage. To demonstrate the power of these tools we will show some success stories for NEMO using BSC tools, reporting how we identified specific bottlenecks, proposed some solutions and finally confirmed the impact of the changes.
|Oriol Tinto, Miguel Castrillo, Kim Serradel, Oriol Mula Valls, Ana Cortes and Francisco J. Doblas Reyes|
|746|| Using High Performance Computing to Model Clay-Polymer Nanocomposites [abstract]
Abstract: Using a three-level multiscale modelling scheme and several Petascale supercomputers, we have been able to model the dynamical process of polymer intercalation into clay tactoids and the ensuing aggregation of polymer-entangled tactoids into larger structures. In our approach, we use a quantum mechanical and atomistic descriptions to derive a coarse-grained yet chemically specific representation that can resolve processes on hitherto inaccessible length and time scales. We applied our approach to study collections of clay mineral tactoids interacting with two synthetic polymers, poly(ethylene glycol) and poly(vinyl alcohol). The controlled behavior of layered materials in a polymer matrix is centrally important for many engineering and manufacturing applications, and opens up a route to computing the properties of complex soft materials based on knowledge of their chemical composition, molecular structure, and processing conditions. In this talk I will present the work we have performed, as well as the techniques we used to enable the model coupling and the deployment on large infrastructures.
|744|| Developing HPC aspects for High order DGM for industrial LES [abstract]
|747|| Introducing the Partnership for Advanced Computing in Europe - PRACE [abstract]
Abstract: The remarkable developments and advances in High Performance Computing (HPC) and communications technology over the last decades made possible many achievements and benefits across a wide variety of academic and industrial branches. Thus, it is well-established that HPC is a key technology and enabler resource for science, industry and business activities, especially for large and complex problems where the scale of the problem being tackled creates challenges or the time of the solution is important. Envisioned to create a world-class competitive and persistent pan-European Research Infrastructure (RI) HPC Service, the Partnership for Advanced Computing in Europe (PRACE) was established in 2010, as a Belgian international not-for-profit association (aisbl) with its seat in Brussels, Belgium. Today, PRACE is one of the world’s leading providers of HPC to research and industry (in particular SME) communities. Out of 25 participating country members within and beyond Europe, 4 “Hosting Members” (France, Germany, Spain and Italy) are in-kind contributors, providing access to 6 leading edge supercomputers in all major architectural classes: JUQUEEN (GCS – FZJ, Germany), CURIE (GENCI – CEA, France), HORNET (GCS – HLRS, Germany), SuperMUC (GCS – LRZ, Germany), MareNostrum (BSC, Spain) and FERMI (CINECA, Italy), who committed a total funding of €400 million for the initial PRACE systems and operations. To keep pace with the dynamic needs of a variety of scientific and industry communities and numerous technical changes and developments, PRACE hosting members' systems are continuously updated and upgraded to make most advanced HPC technologies accessible to European scientists and industry. By pooling national computing resources, PRACE is able to award access to Hosting Members HPC resources, through a unified European open and fair Peer-Review process of proposals calls through a web-tool. Two types of calls for proposals are offered to cover the needs expressed by the research and industry communities and to enable the participating hosting members to synchronize access to the resources, namely the Preparatory Access Call (permanent open call) and the Regular Call for Project Access (twice a year calls). The Preparatory Access is intended for short-term access (2 or 6 months) to resources, for code-enabling and porting, required to prepare proposals for Project Access and to demonstrate the scalability of codes. Project Access is intended for large-scale projects of excellent scientific merit and for which clear European added-value and major impact at international level is expected; and can be used for 12, 24 or 36 months in the case of (Multi-Year Access) production runs. PRACE reserves a level of resources for Centres of Excellence (CoE), selected by the EC under the E-INFRA-5-2015 call for proposals. In 2013, the SME HPC Adoption Programme in Europe (SHAPE) is a pan-European programme to support greater HPC adoption by SMEs was initiated by PRACE. This partnership powers excellent science and engineering in academia and industry, addressing society’s grand challenges. Open to all disciplines of research, and industry for open R&D, the PRACE infrastructure is a vital catalyst in fostering European competitiveness. Up to the 10th PRACE Call for Project Access (February, 2015), PRACE has awarded 10.2 thousand million core hours to 394 R&D projects from 38 countries, to come to fruition and yield unprecedented results. The growing range of disciplines that now depend on HPC can also be observed in the upward trend and evolution of the number and quality of project applications received and resources requested via the PRACE Calls for Project Access. PRACE has supported 2 patents, 158 PhD these, 507 publications (some in the most notable scientific journals) and 719 scientific talks (up to the 5th PRACE Call for Project Access). PRACE is also engaged to provide top-class education and training for computational scientists through the PRACE Advanced Training Centres (PATC), the International HPC Summer School, and PRACE seasonal schools. Until December 2014, PRACE has provided over 200 training events with over 5000 trainees and 19686 person-days of training (attendance-based), with an upward attendance trend from both academia and industry communities. Since mid-2012, PRACE has supported 50 companies, after opening its Calls for Proposals to industrial applicants, in the role of principal investigator or research team member collaborating in an academia-led project. So far, PRACE has awarded 10 SHAPE projects from 6 different countries. PRACE has also published 16 Best Practice Guides and over 200 White Papers. Nowadays it is well-established that HPC is indispensable for Science and Technology advanced in a wide range of scientific disciplines, such as biosciences, climate and health. Success stories and R&D outcomes of PRACE-supported projects shows how joint action and European competitiveness can benefit from a cross-pollination between science and industry (including SMEs), aided by European HPC resources.
|Richard Tavares, Antonella Tesoro, Alison Kennedy and Sergi Girona|