24/03/2026
From climate change prediction to urban planning, supercomputing has become a key tool for science and society. Sergi Girona, CIO of the Barcelona Supercomputing Center, explains why we’re only now beginning to feel the impact.
Super computation came about in the scientific field in response to a very specific need to execute tasks and solve problems that a single computer couldn’t tackle efficiently, according to Sergi Girona, head of technology infrastructure at the Barcelona Supercomputing Center (BSC).
“When you have complex systems, with many unknowns or enormous volumes of data, what you do is divide the problem into small pieces and distribute them among different processors,” he explains. In this way, each party resolves the task assigned to it, but since the parties aren’t independent, they end up exchanging information and, from that interconnection, super computation was born.
“Ramón y Cajal was a great researcher, but his work was only possible because he tools in the form of large-scale pieces of equipment that allowed him to see cells with great clarity.” Supercomputers do the same thing: “They allow for much more detailed simulations of reality,” he explains.
Tech that’s constantly evolving
Supercomputing capabilities have existed since the 1980s and have evolved constantly since then. In Spain, the technological milestone happened in 2004 with the creation of the National Supercomputing Center, (in Barcelona), which is responsible for both researching and providing advanced calculation services to the scientific community.
Two years later, in 2006, a key step was taken with the creation of the Spanish Supercomputing Network (RES), with the aim of structuring access to smaller-scale computing resources for researchers, thereby reserving the large machines for the most demanding problems. “With a single access committee, researchers can request access and are assigned the most suitable resources,” explains Girona, who is in charge of coordinating the different nodes of the RES.
That model was replicated at the European level in 2010, when PRACE (Partnership for Advanced Computing in Europe) was born, an initiative to coordinate the major super computation centers on the continent. And in 2018, the definitive jump came with EuroHPC, a public European initiative involving more than 35 countries that brings together the great machines of the continent. MareNostrum, the BSC’s supercomputer, which covers the Spanish super computation network, is part of that European network of seven super computers.
Multidisciplinary support
Sergi Girona explains that access to super computation is open to all scientific branches. It was first used in areas such as physics, mathematics, chemistry, and astrophysics, but with the incorporation of AI services it’s now much more accessible and has expanded to fields like sociology, demographics, and political science, without abandoning the classical disciplines.
The BSC and the National Center focus on four main research areas:
- Computers and chip design, seeking to boost European technology independence.
- Earth sciences, with applications in climate change, air quality, and urban pollution.
- Life sciences,including genomics, proteins, personalized medicine, and organ simulation.
- Engineering, which encompasses energy (wind, nuclear, fusion, and fission), transport, flight modeling, and other complex systems.
The expert emphasizes that supercomputing has a direct impact on society, and can measure air quality, analyze public health, optimize medical treatments, plan cities, or improve energy efficiency.
“We have pollution alerts and carpool recommendations thanks to high-performance computing (HPC) simulations,” he explains, while in medicine, the examples are transcendental in their impact: “We can simulate how a stent is placed in an artery before the operation and check if the blood flow will be sufficient.”
Supercomputer characteristics
From a technical point of view, a supercomputer works by dividing problems into many small parts that are processed in parallel. Its power is measured by various factors: the number of computational units (CPU and GPU), the memory, and the interconnection network.
“CPUs are general-purpose processors and GPUs act as accelerators, and are used especially in scientific computing and AI,” Girona explains. To achieve large capacity, thousands of nodes, each with their own many nuclei and memory, are used together. At the BSC, for example, there is a supercomputer with 6,480 nodes, each with 112 cores and 256 GB of memory, providing an enormous total capacity.
For everything to work as a single system, a high-speed communications network with high bandwidth and very low latency is essential. “Each node is connected to a network of 200 gigabits per second, with microsecond latencies,” Girona explains, all of which is concentrated in a data center measuring just 900 square meters, with a consumption of between 8 and 9 megawatts.
This high calculation density requires a lot of energy, which in turn necessitates an advanced closed circuit, water-based cooling system that reuses the energy it consumes.
Cyberattacks and security
As with any critical infrastructure, the BSC is a target for cyberattacks. “It’s a non-stop thing,” recognizes Girona, although he emphasizes that they’re not very different from those suffered by banks or public administrations, although at times, interest is greater due to the type of research being conducted.
The protection protocols in place aren’t exceptional either, and since the connection of the supercomputer with the outside world is limited, the risks are reduced. The Spanish center complies with the security policies of the National Intelligence Center, the National Security Center, and RedIRIS, which allows for constant monitoring of connections.
In May 2020, several European supercomputers had their integrity compromised by cybercriminals who used stolen access credentials to take control of the systems. These incidents, which were repeated in later years, didn’t result in irreparable losses.
“The main impact was the loss of time,” says Girona. The machines were isolated, the intrusion was analyzed, and activity resumed. “We don’t work toward commercial ends. If the science that was supposed to come out in five years is delayed by two days, it’s not a big deal,” he summarizes.
The future: AI and quantum computing
For Girona, the potential of super computation is continuous and far from being exhausted. As far the future goes, AI has already been fully integrated, mainly through the development of software and appropriate tools. The next challenge is quantum computing and for several years, the BSC has been experimenting with integrating it into HPC systems.
“We don’t believe it’ll replace classical computers,” Girona clarifies, “but it can solve very specific parts of certain problems.” For now, they are experiments, but they mark the path of a technology that, like supercomputing, promises to continue expanding the limits of what’s possible.
“The goal of super computation is to do science, but science ends where technology transfer begins, which is when companies adopt these developments,” he concludes.
Article collaborators
Sergi Girona has been the Director of Operations at the Barcelona Supercomputing Center-National Supercomputation Center (BSC-CNS) for more than 20 years. He also serves as coordinator for the Spanish Supercomputing Network (RES).
The BSC-CNS is Spain’s national super computation center, specializing in high-performance computing (HPC) and managing the supercomputer MareNostrum, one of the most powerful in Europe. Founded in 2005, it offers HPC services to the international scientific community and industry alike, promoting competitiveness in science and engineering.
The center actively participates in European super computation initiatives, including PRACE and EuroHPC, and manages the Spanish Supercomputing Network (RES). The multidisciplinary team conducts research in five areas: computational sciences, life sciences, earth sciences, computational applications in science and engineering, and social sciences and computational humanities.
BSC-CNS combines Tier 1 facilities with collaboration programs with the EU, Spain, and leading companies, establishing itself as an international benchmark in e-science and attracting talent from around the world.
