Working groups & Tasks

The aim of this task is to provide a management of the project between the mains interfaces and to ensure a global coordination between the WP and the participants Read more

This tasks aims to follow up the progress, especially for the production of deliverables, administrative issues and all kind of meetings. The details are following : Read more

The aim of this task is to investigate and propose initiatives and needs for education and training in computational science and computer technologies in relation to Exascale computing and in particular: Read more

The aim of this task is to investigate and describe funding structures and strategies of public bodies (worldwide) with Exascale computing projects or plans towards such projects. This will be done by establishing and maintaining a global network of expertise and funding bodies in the area of Exascale computing. The follow up of this network should permit to have a relatively good view of overall funding effort of the different regions and by this way, should lead to give orientations and recommendations which will be reported in the deliverables of the global vision of WP7. Read more

All actions of this task do lead to suggestions of roadmap and recommendations which will feed the deliverables of the global vision of WP7, in particularly concerning international collaboration of R&D programs taking in account Europe strengths and weaknesses, coordination of international Exascale centres, and specific Educative programs. Suggestions could also be provided on Europe role in an International Exascale Software initiative. Read more

The aim of this task is to investigate the landscape of co-design centers in the area of HPC and specifically Exascale computing and propose some recommendations. Read more

At Exascale high-resolution, Climate and Earth system models will be feasible to better simulate the evolution and the interactions among the different complex physical and biological processes controlling natural phenomena. Innovative approaches for modelling solid Earth processes will be also achievable for a better understanding of the physical processes controlling earthquakes, volcanic eruptions and tsunamis as well as those driving tectonics and Earth surface dynamics. Unified next-generation atmospheric simulation systems will be feasible for Weather and Climate scientists, allowing a strong reduction in climate prediction uncertainty. This task will identify the main challenges for the roadmap for Weather, Climatology and Earth Sciences modeling at Exascale level, taking into account several challenges like... Read more

Modern life science and medical research is characterized by the need to perform computations on very large data sets. Such computations cover a large number of different application areas and use a large variety of different algorithms and software. The work group of converting this software to run on special purpose hardware in order to benefit from peta and exacale computational resources is very considerable indeed. It will raise challenges in terms of optimisation, scalability and resilience of the various codes as well as that of training people to do it. Read more

The fields of physics and chemistry span a diversity of demanding computational disciplines that share a lot in terms of models and computing kernels, such as nano-physics, solid state physics, computational fluid dynamics, astrophysics, particle physics, material science, soft matter physics, theoretical chemistry, biophysics and plasma physics, to name some important representatives. Large length- and time-scales and multi-scale integration are frequently encountered challenges in these areas. Read more

Engineering simulations are inter-disciplinary and have multiple scales, leading to a growing requirement for supercomputing resources. Solving challenges of using Peta or Exascale resources requires to develop new software approaches, for multi-level parallelism, load-balancing with automatic and adaptive meshing, fault-tolerance and data management. There is also a strong need for increased efficiency in the field of numerical methods, algorithms, libraries, standard and novel programming tools for optimal implementations on many-core systems or massively hybrid (CPU/GPU) computers. Read more

Software or algorithmic innovation that may eventually supplant an existing dominant traditional software or algorithm is understood as disruptive technology. It is considered to have the potential to significantly reshape the high performance computing and data analysis in the near future, but which is not standard in today’s use. Read more

This transverse task will work on setting up and following the 5 working groups of WP3, defining common management guidelines for working group activity and interactions with the others work packages and elaborate a gap analysis methodology between current situation and Exascale objectives. Read more

The core software technologies of today are ill-equipped to handle peta and Exascale systems. One of the major difficulties will be to manage massively parallel systems, composed of millions of heterogeneous cores. The challenge is particularly severe for multi-physics, multi-scale simulation platforms that will have to combine massively parallel software components developed independently from each others. Another difficult issue is to deal with legacy codes, which are constantly evolving and have to stay in the forefront of their disciplines. Scalability and load balancing of scientific applications strongly depend on the quality of the spatial discretisation. Meshing tools adapted to massively parallel computing, including parallel meshing, mesh healing, CAD healing for meshing and dynamic mesh refinement are still mostly missing software pieces. One example from the field of fluid dynamics serves to illustrate the scale of the problem – to run a simulation on one million cores requires a computational mesh with many billions of grid nodes. Most grid generation software is provided by ISVs and is serial. How do we create a mesh with 50+ billion nodes? Assuming we can generate the mesh, how can we partition the grid to run on a million cores? How do we minimise load imbalances? Read more

There are many challenges in the domain of the working group on numerical libraries, solvers and algorithms. Some arise directly because of the new architectural features in Exascale computers and others arise because of the increasing complexity of computations that application scientists and engineers believe can be solved on these forthcoming machines. In the first category numerical algorithms and libraries will have to address memory bandwidth limitations, efficient use of many thousands of multicore chips, the increasing number and variety of errors and faults and the increasing complexity of the architecture (hierarchical, heterogeneous). In this second category, more research and development on algorithms and software for very large three  dimensional multiscale and multiphysics problems is required. Optimization algorithms will increasingly face very large scale structured problems, that require new solution paradigms. Additionally, application researchers increasingly want to solve inverse problems rather than the direct forward problem, for example in seismic modelling and shape optimization, and new algorithms will be required to efficiently utilize already developed forward solvers within this environment. Read more

This WG will attack specific disruptive technologies in the field of software eco systems and numerical analysis. Several topics must be tackling with breakthroughs: Read more

The aim of this task is to investigate state of the art and trends related to an Exascale roadmap in the HPC hardware and software industry and based on these inputs propose initiatives for Europe. Complementary to WG4.2 of Scientific software engineering, software industry refers here to vendor operating software, compilers, debuggers, fault tolerance software, memory management, data management; this will be achieved by establishing and maintaining a global network of contacts with vendors in the HPC industry. Read more

This transverse task will work on setting up and following the 4 working groups of WP4, defining common management guidelines for working group activity and interactions with the others work packages and elaborate a gap analysis methodology between current situation and Exascale objectives. Read more

Resilience is one of the most challenging issues to address before Exascale. A first task will be to fix concrete and reasonable objectives in terms of performance overhead and reliability considering the current state of the art in this domain. We also need to define a reasonable limit in the number of tolerated simultaneous failures. Read more

Engineers and researchers coping with quantitative predictions using deterministic models deal actually with several sources of uncertainties affecting the inputs (and eventually the model itself) which are transferred to the outputs, i.e. the outcomes of the study. Read more

EESI identified power efficiency as one of the most important issues to address in order to achieve Exascale systems. The approach in this task will be to analyze power and performance issues from two different directions: top-down, starting from the complete hardware and software system view, and bottom-up, looking at each of the components of the system individually. These two different approaches will flush out different problems, including component-level inter-dependence and any resulting impact on energy efficiency. Read more

This task will propose R&D actions on one of the major challenge of Exascale application: Data management. Memory by core will decrease dramatically and at the same time data to be treated will increase dramatically too. This WG will ensure a coherent approach toward the evolution on management of I/O, such as Big data transfer, Storage, compression, massively // I/O, also Memory access, memory storage... Sensibilisation and Incentive actions will be provided on following but non exhaustive items: Read more

The objective of this task is to seek out new technologies with the potential to disrupt the HPC landscape as we know it, that means drastic innovations in current practices Disruptive technology could come both from within the HPC market (e.g. the introduction of a new kind of interconnection, a new kind of processor technology) as it was for RISC vs Vector processors of high performance switching technologies, or from other markets (e.g. Gaming, embedded system, etc...).The same is true for what concerns the software technologies. Therefore in this task actions should be taken in order to examine new architectures or single components that will significantly impact the high-performance computing field throughout the next five to 15 years, but have not yet emerged in current systems. Experts of this WG will try to discover emergent disruptive technologies on the following: Read more

This transverse task will work on setting up and following the 5 working groups of WP5, defining common management guidelines for working group activity and interactions with the others work packages and elaborate a gap analysis methodology between current situation and Exascale objectives. Read more

This task will develop and document a methodology for estimating the level of maturity of Exascale software components in cooperation with experts from the ESSI2 work packages 3 ("Application"), 4 (“Enabling technologies”) and 5 (‘Cross-cutting issues”). The results of this task will be documented in deliverable D6.1. It will be the input and prerequisite for T6.2. Read more

The purpose of this task is to identify 3 Exascale software components from current and near future EU Exascale projects and then to evaluate them based on the methodology developed in task 6.1, and to suggest potential structures of such centre. In detail, it will Read more

This task will collect all outputs and feedback provided through the project to update vision and roadmap. The vision of the update Exascale integrating technical, scientific, strategic aspects will be presented through a gap analysis methodology especially for the key issues to be solved (Resilience, Load balancing, Programming, Mathematical solvers and tools, engineering tools and platforms for multiphysics, R&D international collaboration, … Uncertainties quantification, … Exascale Code Maturity, strategy for co-design center, educational programs). Read more

This task will bring together European and international HPC stakeholders, to share information on various aspects, including funding models of relevant research and needs/challenges with respect to efficient software for future Exascale systems. It will organize two technical workshops including parallel sessions for the various Working Groups. The Working Groups will be able to lay out their working methodology and results. The workshops will be held at Ter@tec and CINECA premises. Each workshop is expected to have about 100 participants. The organization of the workshops will be done together with the currently running EC-funded European Exascale projects. Read more

This task will be responsible for the public presentation of projects results to the stakeholders, scientists and policy makers. It will give opportunity to EC-funded Exascale projects to present their findings and results, and will summarize the findings and recommendations of the EESI-2 project. The conference will be organized in Amsterdam, as a one-day event. The number of participants expected for this conference is around 200. During the event, contributions from the Working Groups, EC-funded Exascale projects and international speakers will be included. The conference is planned at M29 at the end of the month of January. Read more

At project start a detailed communication strategy and contributions for the dissemination plan (D8.1) will be established and agreed upon. It will define the target audiences, objectives, communication channels, tools and support material (flyers, web-site), activities, responsibilities and timing. Throughout the lifetime of the project, the selected dissemination actions will be implemented according to the plan. All partners will directly participate to the implementation of the dissemination actions. The goal of this task is to provide to the largest public possible an easy access to general information on the current Exascale developments. Read more

Two kind of events are expected: first, participate in International workshops about Exascale, second, participate in conferences about supercomputing. Read more