Abstract: ne of the key missions of Computational Science is to drive to a solution the challenges posed by several scientific and technological problems. Satisfactory solutions to these problems, however, can be worked out only through massive computational campaigns exploiting the innovative features of platforms based on a large number of cpus that can be offered only by highly parallel machines or distributed Grid infrastructures.
Recent European projects have boosted the latter technology by gathering around the public network a large number of heterogeneous computers belonging to different scientific institutions. This is, indeed, the goal of the EGEE European project that has assembled a computer Grid strong of about one hundred thousand cpus and has gathered on this platform the efforts of many scientists mainly from physics disciplines but also from other scientific sectors like biology, medicine, chemistry, geology, etc.
Molecular and Material Science and Technology (MMST) is, indeed, one of these fields of research in which the possibility of using Grid infrastructure as an advanced high throughput platform is fostering an explosive advance in concurrent programming and distributed implementation of computational packages. This is, in particular, being exploited through the design and implementation of multi-scale and multi-physics realistic simulations (of which advanced molecular modelling is an integral part) exploiting also the basic cooperative nature of Grid computing. As a matter of fact, the COST in Chemistry and Molecular Science and Technology (CMST) domain has been pioneering these efforts and launched at the end of 1999 a first Action (D23: Metalaboratories for complex computational applications in Chemistry) that has represented the first distributed computing cooperative endeavour in MMST. At present D23 has evolved into GRIDCHEM: Grid Computing in Chemistry (D37).
Along this line COST has funded the first edition of the "HANDS ON TRAINING SCHOOL ON MOLECULAR AND MATERIAL SCIENCE GRID APPLICATIONS'' jointly organized within EGEE by the EU-IndiaGrid project and the COMPCHEM Virtual Organization.
This hands-on school aimed at presenting the Grid activities, tools and applications developed and implemented in Molecular and Material Science within the COST D37 Action in collaboration with the EGEE/COMPCHEM and EU-IndiaGrid projects. Target audiences were young scientists active in computational molecular and material science interested in performing their activities on Grid infrastructures. For this purpose, during the school, basic software tools, computer programs and some successful case studies were illustrated and discussed. During the hands on sessions participants were encouraged to bring their own application and enable them on the Grid infrastructures.
This is the reason why the lectures at the school have been articulated in six half days around some MMST subjects and on the introduction to some basic Grid technologies as follows: In the first afternoon an introduction to the school and an illustration of Grid computing (including some introductory hands on tutorial sessions in the laboratory) were given. At the same time a session was devoted to the presentation of the problems proposed by the participants. The following morning some lectures on porting experiences of computational applications on the Grid, implementing on the Grid molecular simulators, electronic structure computational packages were delivered whereas in the afternoon more practical sessions like the illustration of formats for ab initio data, tools for running concurrent applications, practice on the use of Grid-enabling procedure (on the problems presented by the participants). The day after lectures on Biased Exchanged Metadynamics and the Grid, Reactive Scattering and Molecular Dynamics calculations, the Quantum Espresso suite of codes, multiple shooting algorithms for an extended phase space sampling and long time dynamics calculations on the Grid were given in the morning and were followed in the afternoon by practical sessions on profiling applications on the Grid as well as participants work on the implementation on the Grid on their own codes. On the final day some lectures on the Indian Grid experiences in Computational Chemistry and material science and the report on the work performed by the participants on their own problems were delivered.
We consider the school quite succesful: the selected participants were strongly motivated and the shape of the school with all the lectures and hands-on done in a computer lab allowed them to be very focused on the practical side. We got the favourable impression that almost everybody was able to perform some real work on his/her specific problem. The interaction among computational scientists coming from India was also very positive.
This favorable impact has also suggested the collection of the relevant material in this book which may turn useful for subsequent editions of the school and as important reference for other people interested in this field.
Abstract: We present a detailed comparison of the performance of synthetic test programs such as DGEMM and STREAM as well as typical atomistic simulation codes DIPROTEIN and CPMD, which are extensively used in computational physics, chemistry and biology on a large number of high-performance computing platforms. With an eye toward maximizing the price/performance ratio for applications in atomistic simulations, we examine a wide class of commonly used 64-bit machines and discuss our results in terms of the various aspects of the machine architecture, such as CPU speed, SMP memory performance and network. We find that although the Intel EM64T machines show superior performance for applications that extensively exploit the MKL library the Opteron-based machines show superior performance with less optimized codes. Moreover, for large memory applications such as electronic structure codes, the SMP performance of the Opteron is superior. An overview of which architecture is suitable for which applications and a comparison of AMD dual-core CPU technology to Intel hyper-threading are also discussed.
