Research Activities

HPC-Europa2 provides three JRAs:
JRA1: HPC on Massively Parallel Architectures: Programming models;
JRA2: Virtual Cluster;
JRA3: Tools for Scientific Data Services.

In the next few years, the main European HPC centers, most of whom are involved in HPC-Europa2 will provide HPC infrastructures in the order of several hundreds TeraFlop/s and even PetaFlop/s. For the optimal exploitation of such powerful systems, scientific applications need to be properly adapted. This, in general, can require the adoption of new programming models and new programming strategies and consequently code (re)writing or even reengineering. JRA1 participants have explored these aspects, of paramount importance for the scientists involved in computational sciences, exploring, experiencing, testing and analyzing innovative solutions for parallel computing on massive parallel architectures, icnluding accelerators. Progress update

The aim of JRA2 is to help the TA visitors with the development and parallelization of applications, as well with the actual deployment on the TA site’s machines, which offer quite heterogeneous architectures. In order to ease the deployment of the user’s application, it is important to provide the visitor an early experience on parallel machines and methodologies.
This activity will develop a “HPC-Europa-style” Live CD, that will allow actual hands-on building of a “virtual cluster”. This will assist the users (even before their stay at the TA centres) to get experience and make usage of an ad-hoc cluster with all the tools developed at each HPC-Europa2 partner site and have the developed code integrateable onto the final HPC machine at the TA site. The need for such JRA, was identified from the experience gained with the previous HPC-Europa project where some visitors were quite inexperienced in accessing parallel computing environment and methodologies. Many of their codes were not suitable for the requirements of the specific environment at the TA centre (queuing systems, libraries, environment).
Moreover, this JRA provides an excellent opportunity to integrate in a specific environment the tools and educational material developed at each center and in the other JRAs and NAs. Progress update

The results of numerical investigation produced by actual HPC systems are growing in size and complexity. For dealing efficiently with this amount of data, technical solutions are needed as was identified in a Network Activity of the previous HPC-Europa project. Starting from this indication, JRA3 activity will develop data-grid tools with flexible and powerful meta-data capabilities. In particular, it will develop innovative solutions for scientific data management, search and filtering, and develop tools for high-level data-analysis to investigate and describe complex data-relationships. These tools will improve the data accessibility, reusability and enable the exchange within international research communities. The results of this JRA will greatly improve the way the research communities efficiently operate on data. Progress update

HPC-Europa2 as a whole will undoubtedly enable us to raise the profile of HPC within Europe and of European computational science worldwide, and ensure that users’ interests are given due consideration in the standardisation process, well in the direction of the recommendations provided by ESFRI.