With the increasing prevalence of architectures based on massively parallel and multi-core processor topologies, simulation scientists are now compelled to take scalability into account when developing new models or when porting long-established codes to new machines. To meet this challenge, the Jülich Supercomputing Centre (JSC) in Germany, one of the major European players in supercomputing, has pioneered the introduction of community-oriented support units for scientific computing, so called Simulation Laboratories. Progress in establishing these units and first success stories are reported on ( www.fz-juelich.de/ias/jsc/EN/Expertise/SimLab/simlab_node.html ).

Besides its efforts in strengthening the computational sciences, JSC is actively investigating future architectures and their application, primarily in cooperation with hardware vendors. Since 2010, JSC has established joint Exascale Laboratories with IBM, Intel, and NVIDIA, to give such collaborations a more formal framework with binding resource commitments, multi-year work plans, and agreements on intellectual property rights. First-hand experiences with this new kind of collaboration and some highlights are presented ( www.fz-juelich.de/ias/jsc/EN/Research/HPCTechnology/ExaScaleLabs/_node.html ).

back

This talk will address:
- What is Horizon 2020 (H2020)? – an introduction [1],
- General information about funding research in H2020 [2],
- Presentation of topics related to eHealth [3],
- Funding by grants or accepting the challenge? – Prize instrument.

References
[1] ec.europa.eu/programmes/horizon2020/h2020-sections
[2] ec.europa.eu/programmes/horizon2020/en/h2020-section/health-demographic-change-and-wellbeing
[3] ec.europa.eu/research/participants/data/ref/h2020/wp/2014_2015/main/h2020-wp1415-health_en.pdf

back

Phenomics is concerned with detailed description of all aspects of organisms, from their physical foundations at genetic, molecular and cellular level, to behavioural and psychological traits. Neuropsychiatric phenomics tries to understand mental disease from such broad perspective. It is clear that learning sciences also need similar approach that should integrate efforts to understand cognitive processes from the perspective of the brain development, in temporal, spatial, psychological and social aspects. A new branch of science called neurocognitive phenomics, or neurophenomics, is proposed, treating the brain as a substrate shaped by the genetic, epigenetic, cellular and environmental factors, in which learning processes due to the individual experiences, social contacts, education and culture take place. A brief review of selected phenomena relevant to neurophenomics will be presented, from genes to learning styles. Central, peripheral and motor processes in the brain will be linked to learning styles and mental disorders, such as autism and ADHD.

back

Efficient parallelization and optimization for modern parallel architectures is a time-consuming and error-prone task that requires numerous iterations of code transformations, tuning and performance analysis which in many cases have to be redone for every different target architecture.
In this talk we introduce an innovative auto-tuning compiler named INSIEME which consists of a compiler component featuring a multi-objective optimizer and a runtime system. The multi-objective optimizer derives a set of non-dominated solutions, each of them expressing a trade-off among different conflicting objectives such as execution time, efficiency, and energy consumption. This set is commonly known as Pareto set. Our search algorithm, which explores code transformations and their parameter settings, dynamic concurrency throttling (DCT), and dynamic voltage and frequency scaling (DVFS) is based on differential evolution. Additionally, rough sets are employed to reduce the search space, and thus the number of evaluations required during compilation. We demonstrate our approach by tuninng loop tiling, the nr. of threads, and clock frequency in cache sensitive parallel programs, optimizing for runtime, efficiency and energy. 
www.insieme-compiler.org

back

"The new EGI vision statement for an ‘Open Science Commons’ reads: “Researchers from all disciplines have easy and open access to the advanced digital services, data, knowledge and expertise they need to collaborate to achieve excellence in science, research and innovation”.

Open Science refers to the opening of knowledge creation and dissemination towards a multitude of stakeholders, including the society in general. Nowadays, open science, sometimes called Science 2.0, is driven by the digitalization of the research process and by the globalization of the scientific communities.

Commons is a resource management principle by which a resource is shared within a community in a way that allows non-discriminatory access, while ensuring adequate controls to avoid congestion or depletion when the capacity is limited.

The next step is to recognize that e-Infrastructures are a component of the whole scientific process that should be managed as a commons. The new Open Science Commons vision will be enabled by the richness of the EGI ecosystem, where different actors participate with different roles to deliver value to the processes involved into research and innovation.

back

The speech addresses the technical foundations and non-technical framework of Big Data. A new era of data analytics promises tremendous value on the basis of cloud computing technology. Can we perform predictive analytics in real-time? Can our models scale to rapidly growing data? How do we efficiently deal with bad data? Some practical examples as well as open research questions are discussed.

back

Scientific data repositories have been growing larger than ever. Yet it is still surprisingly unhandy to easily share files between researchers, access the data from mobile devices or personal desktop computers. At the same time in the open consumer market simple file sharing and synchronisation services have been extremely successful: Dropbox, Box, Wuala are just few big names. At CERN we have lately looked into extending the idea of simple file sharing and synchronization service to help researcher do their science in easier and sometimes innovative ways. High Energy Physics (HEP) is an interesting target as it is a highly distributed and fairly large user community (more than 10K users in over 100 countries) with very large scientific data repositories (above 50PB). As often for other sciences, the data analysis workflows in HEP are tightly integrated with processing and data services – batch farms, analysis clusters and storage systems provided by Computing Centers. This creates an additional challenge but also offers an interesting opportunity to explore file sharing and synchronization in this context. In this talk we will review recent initiative at CERN in the area of file sharing and synchronization which is also an invitation for collaboration.

back

Significant advances in the scope of e-Science during the last two decades and resulting sophistication in the e-Infrastructures have evolved them as critical information infrastructure for our society. Providing reasonable security services to these infrastructures is no longer an optional feature. Each e-Infrastructure's operations team has specialised security professionals to ensure effective protection of their assets. They also have CERT/CSIRT members who are tasked with maintaining a vigilant lookout to discover and remediate security breaches. These cyber firefighters play vital role in the smooth and secure functioning of e-Infrastructures. Their task is so overwhelming that in the most of the situations, they have to limit their scope to the detection and removal of security cracks, instead of pursuing these incidents as cybercrime investigations. However, post-accident analysis of these attacks is becoming a necessity with the ever-increasing reliance of scientific and industrial communities on e-infrastructures. This talk will describe best practices and guidelines for the e-Infrastructures' cyber incident response teams to cope with the inevitable challenges posed by cybercriminals. Benchmarks for evaluating the operational readiness of these teams will also be presented.

back

Since decades, HP has traditionally provided our HPC users with a set of traditional, yet innovative server technologies. The emergence of blade-based designs has allowed HP to reach a very significant position in the TOP500. HP is today the clear leader amongst all other vendors in terms of number of installed systems. With the recent announcement of the densest and easy-to-deploy hyperscale products, HP sees a massive opportunity to develop a set of dedicated offerings for its most demanding customers. We also will touch our current directions for HPC, and the trends that HP can see on this market. Finally, we will also mention the R&D directions that HP is pursuing.

back