New competence centre to assist with large-scale data processing
7 June 2021
There is a growing need to use supercomputers for modelling or data processing for use in high performance computer systems. However, to be able to make use of these kinds of computer systems, the software must first be adapted. There, a new national competence centre hosted by Uppsala University fulfils an important function.
"We enable users to make optimal use of their software on European supercomputers," says Lilit Axner, researcher and Director of EuroCC National Competence Centre Sweden (ENCCS).
Since September 2020, there has been a new national competence centre hosted by Uppsala University: ENCCS. It is one of 33 national competence centres across Europe, aiming to provide training and support for users of EU’s supercomputers. The goal is to bring together the skills required to use the cross-border European network of high performance computing systems.
The Swedish competence centre is a joint initiative between the ten largest Swedish universities and the Research Institutes of Sweden (RISE) and is funded by the European High Performance Computing Joint Undertaking (EuroHPC JU), the Swedish Research Council, and Vinnova.
At ENCCS, Lilit Axner and thirteen employees are working to spread information on how their user support can be accessed by academia, as well as business and industry and government agencies.
"We provide comprehensive training so people know how to make optimal use of their software on these supercomputers. We support software that needs to be adapted to work in the new generation processors. In addition, we help people to apply for access to the large-scale computer systems in Europe, access to which is highly competitive,” says Lilit Axner.
Assist users from all disciplines
The advice focuses on three areas: high performance computing (HPC), artificial intelligence (AI), and high performance data analytics (HPDA). Combining these technologies provides tools for calculating, modelling and simulating data for large, complex problems across a wide range of areas. Being able to use e-infrastructure effectively is of interest to all disciplinary domains, as Lilit Axner explains.
"You can calculate global warming, pollution, and the spread of viruses such as COVID-19, which has already been done. It is with the help of supercomputers that researchers have modelled the impact of the COVID-19 vaccine and how consistent this is with real vaccination against COVID-19. If the first calculations you make do not show that the vaccine has a sufficient impact on the virus, you have to make adjustments and new calculations until you see in the modelling that the vaccine will manage to kill the virus.”
Such step-by-step, or iterative, modelling is an important technique, and also a basis for collaboration.
"Examples of large collaborative groups are the Swedish Meteorological and Hydrological Institute’s group and the molecular biosciences groups at SciLifeLab, where many countries share the results of calculations with each other," says Lilit Axner.
How long is the wait time for access to the supercomputers in Europe?
“Every time the application period for a supercomputer opens, there are approximately 200 applications from all over Europe. However, depending on the length of the requested computing time, you usually get a response in two weeks to three months, and then you can run your calculations straight away,” says Lilit Axner.
The duration of the processing depends on the individual software and the quantity of data. However, the user is given access to the computer system for between two months and three years and can run their data as many times as they like during that period. It does not cost the individual researcher anything as the Swedish Research Council and Vinnova pay for Sweden’s membership.
ENCCS also helps Swedish users with access to one of the world's fastest supercomputers, LUMI, which will be launched this summer in Kajaani, Finland. LUMI, or Large Unified Modern Infrastructure, is 3.5 per cent owned by Sweden and can also be used by companies as long as they provide staff who can work with the competence centre, according to Lilit Axner.
Training in Artificial Intelligence
ENCCS is yet to receive a request from researchers at Uppsala University, but it has received requests from a number of external researchers. Some users have also come from government agencies and companies. The ENCCS also regularly holds training activities such as seminars and workshops for academia, the public sector and industry.
“We held an AI boot camp for beginners where we demonstrated the process you go through when applying artificial intelligence to solve a problem for example. We often organise AI events with RISE, and all our activities are free and open to all. Since May, we also have an additional employee who specialises in AI and machine learning, so now we also have the resources to handle these kinds of calculation requests,” says Lilit Axner.
The EuroCC National Competence Centre Sweden (ENCCS) is part of the Department of Information Technology at Uppsala University. The Swedish National Infrastructure for Computing (SNIC) also offers training in modern data processing and the adaptation of selected software for high performance national computing systems in Sweden. Unlike SNIC, however, ENCCS works not only with academia, but also with public administration and industry.
The EU project that coordinates access to the various high performance computing systems is called the Partnership for Advanced Computing in Europe (PRACE).