All of us dream, but I imagine that very few of us ever dream about supercomputers. After all, to the untrained eye a supercomputer is little more than a room full of banks of flashing lights and cables, an ironic throwback to the earliest days of computing when even the least powerful computer was the size of a small room, and certainly nothing very much to be excited about.
But let me make a confession: supercomputers do make me dream. Not of course because of what they look like, but because of what they can do and the places they can potentially take us.
Science fiction has in fact been taking us to new places with the help of supercomputers for years. In the 2017 film The Martian, for example, when Matt Damon’s character is left stranded on Mars with the rest of his crew already heading home on their long journey to Earth, NASA scientists use the Pleiades supercomputer to figure out a way to turn the spacecraft around and slingshot it back to Mars to pick him up. Written down like that, it sounds almost unbelievable, but in fact the filmmakers worked very closely with NASA on the accuracy of every part of that film, and it is precisely because of its plausibility that the movie was so well received.
The Martian is set in 2035 - just around the corner in technological terms. And while we may not have that sort of supercomputing power yet, or put it to quite those uses, we are very close to being able to. It may surprise you to know that there is a global ranking of supercomputers - there are so relatively few of them around the globe that it is still easy to compare and contrast them all. Competition between each supercomputer centre (and the country or region where they are based) is fierce - in turn pushing every centre towards bigger and better computers and bigger and better achievements.
Part of our aspirations is for Europe to become a world leader in terms supercomputing power, and now I am convinced that this will become a reality. Just last month, eight different supercomputer sites in eight different EU countries were selected to become the future hosts of some of the world’s biggest and best supercomputers. When the computers are installed sometime next year, they will be the result of a dream shared by Member States and the European Union, together investing €900 million to make that dream come true. Combined, these new supercomputers will boost European computing by four to five times compared to current levels, a shot in the arm that will make Europe the world’s leading supercomputing power by 2020. But it will do much than just lift us up the global ranking: it is also expected to generate investment of more than €10 billion in applications made possible or improved by the increase computing power, with benefits for all areas of industry, including small and medium-sized enterprises. The EU-funded Fortissimo project has already given many such enterprises in a range of fields the opportunity to use supercomputers to make major discoveries and advances.
And this is where the real dream comes true. Being the world’s number one in supercomputing power is important, not least as it sends a clear message to the rest of the world about our ambitions. But having all this supercomputer power is not an end in its own right: it is what it will allow us to achieve that really excites me. Supercomputers are already behind some of the most innovative and ground-breaking applications in the world today, in almost any area you can think of from personalised medicine to advanced research on cancer, from cybersecurity and natural disasters’ prediction to environmental protection and road safety. I have invited the heads of three of those centres to share some of those results with you below.
But the additional computing power these new sites will bring leads to a wealth of new opportunities to really push the boundaries of what we can achieve. These supercomputers will be accessible to any private or public users in the European Union. They will serve over 800 application areas including in critical domains such as artificial intelligence and big data analytics, cybersecurity and blockchain, accelerating the digital transformation of our economy and society. They will support Europe’s researchers and industry in developing new applications and products in many areas, such as personalised medicine, new drugs, material design, bio-engineering and manufacturing, weather forecasting, understanding climate change and many, many more.
Science fiction has always enabled us to dream about the future, and some of the most visionary writers have even wondered whether we are the only ones who can dream. “Do Androids Dream of Electric Sheep?” asked writer Philip K Dick - a question that was not entirely answered in the film of his book, better known as Blade Runner. We may be many years away yet from creating machines that look, think, experience and dream like we do, and I wonder if that would be a good idea, but this great leap forward in supercomputing power might just be the moment that those machines began to allow us to dream of a better world.
Sanzio Bassini, Director of Supercomputing Applications and Innovation Department, Cineca
Showing how supercomputing can make a difference to society – in Italy
Since 1969, Cineca has supported Italian and European public and private scientific communities, providing access to advanced supercomputing systems and high-level specialised support. Scientific and socio-economic challenges, collaborations, co-design, are the themes that shape Cineca's activities, which are carried out in collaboration with national and European institutions, the technology industry, and national and local governments.
The supercomputing systems most recently built by Cineca, FERMI and its successor MARCONI, were ranked 7th and 12th in the TOP500 global supercomputing ranking at the time of their entry into production. MARCONI will be replaced by the LEONARDO pre-exascale system, which will aim to be one of the top five systems in TOP500 when it enters into production. This is scheduled for the beginning of 2021. LEONARDO is one of the three pre-exascale machines co-funded by the EuroHPC Joint Undertaking – in this case, jointly with the Italian Ministry of Education University and Research.
Cineca is heavily involved in supporting excellence in scientific research, and is dedicated to pushing the boundaries of knowledge. For example, it has carried out support and supercomputing activities for the international research projects Virgo, which is aiming to detect gravitational waves, Mars Express, which is looking for water on Mars, and the Event Horizon Telescope, which produced the first ever image of a black hole.
Cineca has also been part of projects that seek to boost Europe’s competitiveness and improve the lives of its citizens. These include Antarex, a revolutionary and scalable approach to accelerating the discovery of new drugs with a particular focus on responding to epidemics, Mistral, which is developing a portal for personalised processing of weather forecasts for companies and citizens, and Highlander, a project aimed at simulating the impact of climate change scenarios to support agricultural production, with particular reference to southern Italy. Support for innovation for industrial production and technology transfer is also at the centre of Cineca services: many projects also draw on IoT and artificial intelligence methods that address industrial problems, like developing predictive maintenance techniques and creating “digital twins”, virtual models of products which are used to spot potential problems in manufacturing processes and make them more efficient.
Kimmo Koski, Managing Director of the Finnish IT center for science, CSC
Building Europe’s supercomputing ecosystem – in Finland
Twice a year, the TOP500 list of the world’s most powerful computers is published, sparking competition among supercomputing vendors, countries and continents. Currently, European machines are not as well represented as they could be in the top ten. The EuroHPC initiative was born from the need to bridge this gap, with a target of launching a minimum of two pre-exascale infrastructures available in 2020, with full exascale systems a few years after that.
European competitiveness is at stake, so this ambitious target is definitely worth supporting. There is a huge need for more computational capacity in the European research and innovation system. However, capacity alone is not enough to keep us competitive. Supercomputing excellence will not be won by hardware, but by people with the skills to make the most of the machines. The primary target of EuroHPC should therefore not just be a high score on the TOP500 list, but the ability to efficiently solve the most demanding research challenges with maximum impact for European research and industry.
EuroHPC is a great opportunity to accelerate and intensify collaboration between European countries. Its first eight supercomputers will be hosted by consortia of which a total of nineteen European countries are members. One of them brings together eight countries, led by CSC – IT Center for Science in Finland, to invest together over €100 million in a pre-exascale system, the LUMI (Large Unified Modern Infrastructure). This will be located in Finland but shared by all European researchers. Such a project is unique on a European scale. We consider this the most important outcome of this project – becoming a role model in how European countries can support each other by investing jointly.
Collaboration, trust, infrastructure, software, skills and competencies – these are all parts of an ecosystem. Leveraging local strengths through smart specialisation will create a supercomputing success story for Europe, and as we streamline the ecosystem further we must aim for progress in many different areas. For example, when building this new infrastructure we must make full use of renewable energy, sustainable data centre technologies and reuse of waste heat – not only for cost efficiency, but also to fight climate change.
CSC was established in 1971 and has a long tradition of serving researchers. Today, it is a “one stop shop” providing supercomputing, applications, a research network, data infrastructures, training and consultancy for higher education institutions, research institutes, public administration and industry.
Some examples of CSC’s application development include collaboration with Finnish partners Aalto University on fusion energy simulation code Elmfire, to be used in experiments in generating fusion energy, and University of Helsinki on space physics code Vlasiator: its goal is to study fundamental plasma processes and gain a deeper understanding of space weather. Both codes are of high leading international quality and developed in Finland and they are currently being recoded and developed to fully match the requirements of exascale computing, as part of “Forward looking software solutions work package” of PRACE, the Partnership for Advanced Computing in Europe. This will help to keep these examples of trailblazing Finnish research in computational science internationally competitive. An example of CSC's in-house code development is the Elmer scientific simulation software, an open source multi-physical simulation software mainly developed by CSC. CSC plays a major role in European e-infrastructure development and collaboration, including PRACE, EUDAT, ELIXIR, EOSC, RDA and a number of other international projects.
At national level Finland is currently renewing its national data management, supercomputing and cloud infrastructure, with a €37 million investment from the government. It is building a comprehensive data management and computing infrastructure, where the strong links between data and supercomputing will bring added value to research, industry and the whole of society by enhancing data-intensive research, machine learning and AI development, not forgetting the crucial skills development that is needed. It is not only about the infrastructure, but what you get out of it.
The European Commission and EuroHPC member states have the potential to have a significant impact by putting supercomputing high on the European agenda. Now we need to maximize the impact, by developing the European exascale ecosystem as a whole. This requires solid cross-border and cross-disciplinary collaboration, skills development, and a long-term vision for the role of supercomputing. This is a truly European effort, which has a huge potential for great societal impact. Let’s make the most of it together!
Mateo Valero, Director, Barcelona Supercomputing Centre
Supercomputing as a strategic resource – the Spanish contribution
Since its foundation in 2004, the Barcelona Supercomputing Center (BSC) has aimed for excellence not only in computing architecture and computing science but also in designing supercomputing applications. Indeed, of the 650 scientists and engineers working at the BSC, about 200 work on computer science, architecture and IT services. The rest are part of the departments for life sciences, earth sciences, and engineering, developing and optimising applications that respond to the current needs of research and industry. They are devising ways of meeting the great challenges of our society, such as climate change prediction, personalised medicine, combustion and new renewable energy sources. BSC researchers have, for example, developed an ultra-accurate model of the human heart to simulate its functioning, to help doctors to better understand diseases and design treatments, and are also investigating how drug sprays are absorbed by a patient’s respiratory system, with implications for the design of new medicines. In addition, they are involved in climate forecasting and in analysing the accuracy of previous forecasts. The BSC’s work also includes the modelling of fusion physics devices, contributing to the design of reactors that could potentially generate vast quantities of sustainable energy.
Excellent applications need to be supported by excellent technologies. The EuroHPC Joint Initiative’s goal is not only to build supercomputers but to established a European supercomputing technology supply chain, to ensure that Europe is not dependent on other global regions in this key strategic area.
For example, low-power microprocessors are essential for building supercomputers with pre-exascale and exascale capacity (respectively able to perform 10^17 and 10^18 calculations a second). In 2018, the Commission launched the European Processor Initiative (EPI), which will support the development of the processors we need in Europe. As part of this initiative, the BSC has taken the lead in creating new ways of designing processors. These will be used in a wide spectrum of applications, not just in supercomputing but also other fields, like artificial intelligence and automotive engineering. Several other initiatives will follow and larger investments will be needed, but we should all be proud and grateful for this very important first step.
We firmly believe that in Europe, while it will be necessary to continue using non-European supercomputing technologies for the moment, developing our own supercomputing ecosystem in the next few years will be crucial. The EuroHPC is a very significant step in the process; now we need strong long-term support from national and European funding agencies.