Supercomputer as a driver of innovation
State-of-the-art high-tech at Paderborn University
The new supercomputer 'Noctua 2' is the centerpiece of the Paderborn University's high-performance computing center, which was inaugurated in 2022. Researchers nationwide use the computing capacities of the high-performance computer systems to perform the most sophisticated computer simulations and science at the highest level.
Computational sciences
Observing nature, conducting and evaluating experiments, working theoretically: This is how researchers used to work before the first computer was invented in order to understand certain processes, procedures and developments. With the latest computer programs, scientists today also have the opportunity to get closer and closer to reality through precise simulations and models. The advantages of the so-called computational sciences are undeniable: Experiments that would be extremely complex, expensive, dangerous or simply impossible to conduct can be simulated on the computer. So instead of, for example, carrying out many elaborate experiments to find out which material they want to use in the first place, researchers can first calculate what potential functionality different materials have - before they enter the experimental phase. Simulations can also be used to make predictions about future developments, for example in the area of climate change. Various scenarios are used to calculate: What changes if our average temperature rises by three degrees Celsius? What if it only rises by two degrees? Other advantages include perfect reproducibility and the fact that results obtained on the computer can provide new explanations as to why something happens. With supercomputers, researchers have the ability to sift through huge data sets and identify patterns within a short period of time.
High-performance computing as a key technology for numerous scientific fields
Since the beginning of 2022, a total of 140,000 compactly installed processor cores in the 'Noctua 2' supercomputer in Paderborn have been working on highly complex tasks. The computer is operated by the Paderborn Center for Parallel Computing PC2, a central scientific institute of the university. The institute's expertise lies in calculations of atomistic simulations, computational physics and optoelectronics. Specialized simulation programs are also developed and applied in Paderborn for these purposes.
"With a supercomputer it is like having a time machine, because it allows us to calculate today what would otherwise only be possible in one or two decades with future conventional computers. We have to make efficient use of this head start in time and devote ourselves to topics that are relevant to our society - such as sustainable energy technology,"
explains Prof. Dr. Thomas Kühne, theoretical chemist and vice chairman of PC2. " Like using a microscope, we can get closer to things with the help of 'Noctua 2' - except that here, for example, we can see individual atoms and their interactions, which is better than with any microscope in the laboratory. And the whole process is extremely fast: We can run a large number of simulations within a very short time and thus test out millions of structures. This deeper insight into processes helps us understand, for example, how certain chemical reactions take place at the atomic level."
Building X - Home of the supercomputer

Sustainable energy systems
The research questions tackled with the supercomputer in Paderborn are diverse. Tomorrow's research will focus, among other things, on sustainable energy conversion using solar cells and environmental-friendly water splitting to generate hydrogen.
In photocatalytic water splitting, the energy of sunlight is used to produce hydrogen - an important part of the energy transition. In this process, catalysts can be used to speed up the reaction, make it more efficient and obtain more hydrogen. But which catalyst is the best? "With simulation programs we have developed, on 'Noctua 2' we can observe why some catalysts work better or worse, and we can also find new compounds that our experimentally working colleagues can then produce in the lab and that are at the same time more efficient than catalysts already in use," explains Kühne.
Another example are so-called tandem solar cells, whose individual layers can both efficiently convert the energy of sunlight and are translucent, so that light still reaches the second and possibly third layer, which can be exploited. In this way, scientists hope to increase the efficiency of the cells. With the previous supercomputer, 'Noctua 1', the Paderborn scientists were already able to propose various materials for tandem solar cells with an efficiency of 23 percent.
Paderborn is a national high-performance computing center
Many universities and non-university research institutions in Germany have local computing centers for their own use. These are often financed by their own budgets or by the German Research Foundation. In addition, there are national high-performance computing centers in a network called NHR, currently nine (as of May 2022), of which PC2 is one. They make their supercomputers available to users from universities throughout Germany. The federal government and the state of North Rhine-Westphalia have funded the Paderborn supercomputer and the new high-performance computing center with grants totaling 25.4 million euros from the federal Research Infrastructure Program. Additional funding of up to 75 million euros will flow from 2021 to 2030 as part of the NHR network funding.
"With 'Noctua 2', we have expanded to a new dimension" - we are now among the top 10 academic German computing centers,"
enthuses Prof. Dr. Christian Plessl, computer scientist and chairman of the PC2 board. Only the three national supercomputing centers that make up the Gauss Centre for Supercomputing are larger.
"We are already working on plans for 'Noctua 3'," reveals Plessl. The new Building X at Paderborn University, which was built for 'Noctua 2', was specially designed to provide twice as much space as the current computer. Cooling and fire protection concepts, power supply and office space are also designed for expansion stages. Sustainability played a special role in the construction of the building: 100 percent of the electricity for 'Noctua 2' is generated from hydropower and is therefore CO2-free. The hot water cooling system is highly efficient and the waste heat is used to heat X and other buildings. The chillers are all free of chlorofluorocarbons, which were often used as refrigerants in the past but are harmful to the ozone layer.
'Noctua 2' user operation
Employees of all German universities are eligible to apply for computing time on the supercomputer. Regardless of whether researchers work at Paderborn University itself, at other universities in the state of NRW or nationwide: In an application, they specify what is to be computed, how many processor hours are expected to be needed and what the goal of their research is. "A typical computing time proposal has a scope of six million processor hours that can be used on our supercomputer within one year. This is equivalent to the aggregated computing time of a workstation computer with eight processor cores over 85 years," Plessl explains. The smaller applications are reviewed and approved internally by PC2. However, "For very large projects with more than twelve million processor hours, the scientific quality and appropriateness of the use of resources is evaluated by external reviewers."
Once a project has been approved, applicants can send their program and the data to be computed to the 'Noctua 2' queue. As soon as the required number of processors becomes available, calculation starts. Afterwards, users receive their results via download. "All these processes usually run automatically. Assigning the free capacities to the projects in the queue is a bit like Tetris, because thousands of jobs are being calculated at the same time and a suitable gap has to be searched for," says Plessl.
Computer Systems Research in Paderborn
The Paderborn Center for Parallel Computing has its roots in theoretical computer science and has continued to develop over the past 30 years. Today, the PC2 acts on the one hand as a service provider for various users, offering consulting and services in the field of high-performance computing. On the other hand, the PC2 focuses on computer systems research itself by researching particularly efficient hardware accelerator technologies.
The processors in the computer are like tools with which tasks are processed: There are specific and non-specific processors, specifically "central processing units," or CPUs. These can be thought of as either an entire toolbox (non-specific) or, for example, a wrench (specific). Now, there are tasks for which only a wrench is needed - it is not very energy-efficient to carry the whole box. Depending on the requirements, corresponding CPUs are used. However, there are also tasks that are so user-dependent that hardware components have to be specially programmed for them. To stay with the image, this could be a combination of wrench and tweezers. These are the FPGAs, "Field-Programmable Gate Arrays". FPGAs are fully specialized arithmetic units, interconnection networks and memories that allow massive parallel computing in the field, with thousands of simultaneous operations. The Paderborn computer scientists are world leaders in this field: they are investigating how to simplify the programming of FPGAs, how adaptable they are, whether there are alternative methods for doing so. "We have a Europe-wide unique installation of so-called FPGAs here. This makes it possible to completely specialize the computing unit for the given task at hand. We consider this to be an extremely promising technology for the computer systems of tomorrow," summarizes Plessl. CPUs, GPUs and FPGAs are all installed in 'Noctua 2' - a unique and, above all, productively used high-performance computer on which not only basic research but also very application-oriented work is carried out - in other words, a supercomputer.
Author: Gesa Seidel, Department of Press, Communications and Marketing
Good to know
Conventional computers, like typical personal computers, perform "simple" calculations by executing one or very few tasks at the same time. Supercomputers use "parallel" processing, because the tasks, which they must master, are extremely complex. For this purpose, they use a very large number of parallel processing structures, each of which takes care of one work package that has been created by splitting the complex task into many smaller work packages. Parallel computing can thus be compared to the transportation of people: In this analogy, the cars represent the processors. In the past, there was a car that drove from A to B over and over again to transport people. At some point, buses were used because more people could fit in them, but they too drove from A to B over and over again. Today, however, so many people want to travel the route from A to B that several buses and even trains are used simultaneously. So, by analogy, there are many processors working in parallel in a supercomputer because they can do so much more than with a simple processor. The biggest challenge for researchers who want to work with the help of a parallel computer like 'Noctua 2' is to consider which parts of the problem are sufficiently independent from each other can be computed in parallel, i.e. separately.
The development of computers is extremely dynamic and has seen an exponential growth in computing capabilities: whether mainframe computers or cell phones, the devices will be twice as fast in two years as they are today.
One of the first commercially distributed supercomputers was called "Cray 2" and was developed in 1985. Its computing power was about the same as Apple's iPad 2. Six years later, the technology of massive parallel computing was used for the first time in the "Connection Machine CM-5" with more than 1024 processors. To date, there have been countless further developments. The largest supercomputer today (as of May 2022) is in Japan: "Fugaku" is the most powerful in the world; its computing power is 283 million times that of the "Cray 2". It consists of over 7.5 million processor cores.
Today, supercomputers are primarily used in the natural sciences and engineering. They are not yet as widely used in the humanities or social sciences.