“Our objective is to support innovation, primarily for small and medium-sized enterprises, and to leverage excessive computing capabilities of large companies,” says WUT Professor Michał Śmiałek, D.Sc., Eng. of the Faculty of Electrical Engineering, who is the BalticLSC Consortium Coordinator.
According to a report by PRACE (Partnership for Advanced Computing in Europe), the European Union loses 2% to 3% GDP annually due to lagging in the application of advanced computing methods. Those indicators are even worse for the Baltic states. The BalticLSC project is to improve this.
Leveraging the potential
“Our project focuses on three main areas: providing computing capabilities to those who need them; identifying areas were these capabilities could boost competitiveness; and managing excessive capabilities to the benefit of the providers,” explains WUT Professor Bartosz Sawicki, D.Sc. Eng. of the Faculty if Electrical Engineering, who leads the BalticLSC system implementation team.
The project participants aim to create a system of interrelations and interdependencies. Companies and institutions who have insufficient knowledge of large-scale computing or no access to high-performance computing capabilities will be provided with tools which facilitate formulating questions and with an infrastructure which accelerates the delivery of results as required. Those businesses who are properly equipped but make a limited use of their in-house capabilities will also benefit as the available computing potential will be offered to others.
Ultimately, BalticLSC intends to develop a business model to ensure that the created system is efficiently used. “We do not want it to be purely commercial,” says Professor Śmiałek. “We want it to incorporate various financing methods and support non-profit organizations and government agencies and to be open to new companies and institutions.
How does it work?
A computer generates a specific number of computing operations per second. That number represents its computing performance. “A standard computer can compute eight threads at a time while for supercomputers this figure is in the range of tens, hundreds or even thousands,” explains Krzysztof Marek, a student of the Faculty of Electrical Engineering, who is involved in BalticLSC. “Computers can be grouped into clusters and scaled up virtually infinitely.”
Such clustering will be the strength of the Consortium led by the Warsaw University of Technology. Other Consortium Members are: RISE Research Institutes of Sweden AB, Sweden; IMCS University of Latvia, Latvia; EurA AG, Germany; the Municipality of Vejle, Denmark; the Lithuanian Innovation Centre, Lithuania; Machine Technology Center Turku, Finland; and Tartu Science Park Foundation, Estonia.
The WUT scientists are responsible for coordinating all work on the project. They also handle software development.
“The challenge is to design software so that it can perform computing operations on thousands or millions of computers in parallel,” says Professor Sawicki. “We call it parallelization.”
What is it about practically? Let us take an example.
“Say there is a shipbuilder designing a new yacht hull,” says Professor Sawicki. “Engineers determine the optimization parameters and use simulation software to test their values. For instance, a lean hull is fast but unstable whereas a wide hull is more stable but also slower. And what if you have thousands of such parameters? This means millions of instances to compute. Normally, it would take weeks or months to fully review the spectrum of solutions.
“With our system, you can compute multiple variants simultaneously instead of processing them one by one,” explains Professor Śmiałek.
“All you need to do is to upload your task, which will then be forwarded to other computers and in a matter of hours you will have your result,” clarifies Professor Sawicki.
No background in large-scale computing or software developer skills will be needed to use the system. “This will be our job; our task is to accelerate the process,” says Krzysztof Marek. “Anyone willing to enter the network will just have to know what they want to calculate.” The field expertise will be the baseline.
“The applications of our system will include classical engineering but also the now popular fields such as data mining, machine learning or big data,” says Professor Śmiałek.
Assistance to scientists
The project is to be tailored to the needs and expectations of its users as much as possible. To gain an insight and a better understanding of what these are, scientists are holding free workshops “Big Computing for Smaller Entities”. The information harvested this way will help them design solutions to address the actual challenges and issues faced by companies, institutions and innovators.
Promotion and Information Office
Learn more about the project at https://www.balticlsc.eu/.
Project duration: January 1, 2019 to September 30, 2021.
The project is funded under the Interreg Baltic Sea Region Program and co-funded under the European Regional Development Fund.