The Institute of Physics Belgrade (IPB) has established a long-term collaboration with Hewlett Packard Enterprise (HPE) in the design, deployment, and evolution of the PARADOX high-performance computing cluster, the national flagship supercomputing system supporting research and innovation in Serbia.
Through this collaboration, HPE provided advanced HPC technologies and expertise, contributing to the implementation of a scalable, energy-efficient, and high-availability computing infrastructure tailored to the needs of scientific research, data-intensive workloads, and emerging AI applications. The PARADOX cluster was designed to support a wide range of disciplines, including physics, materials science, chemistry, life sciences, climate modeling, and artificial intelligence.
The partnership involved close collaboration on defining the system architecture and performance requirements, followed by the deployment of HPE’s compute, storage, and interconnect technologies, together with ongoing knowledge transfer and technical support to ensure efficient operation, maintenance, and future upgrades of the system.
One of the key contributors within this long-standing partnership is Đorđe Ristić from Hewlett Packard Enterprise, operated by Selectium, who spoke in a recent interview about the decades-long cooperation between HPE and the Institute of Physics Belgrade, as well as the continuous development of high-performance computing systems in Serbia.
He recalled his first professional interactions with the Institute at the end of the 1990s, when, as a young engineer, he was involved in activities related to numerical simulations and early computational research. Reflecting on this long trajectory, he emphasized the importance of sustained technological partnership and support for scientific computing infrastructures, highlighting how collaboration with IPB has accompanied the evolution of high-performance computing in Serbia—from its early research clusters to today’s modern, faster, and more energy-efficient systems. This enduring partnership, he noted, has played a significant role in strengthening national research capacities and supporting the advancement of computational science in the country.
HPE has developed some of the world’s most powerful supercomputers, with several used by private companies and most operating within research institutions. We spoke with Milan Avramović from HPE (handled by Selectium) about why HPE supercomputers are ranked among the top systems worldwide.
"HPE is a world-famous company producing datacenter infrastructure, servers, storage, switches... But we are also present in supercomputing, high-performance computing, and AI fields", says Milan. "HPE acquired Cray Supercomputers in 2019, and with that, we gained a lot of expertise in building these kinds of supercomputers, used in scientific and other fields. And currently, as of this year (2025), HPE has seven systems out of the top ten of the Top500 Supercomputer list. We are proud of that, of course. Basically, the first three fastest supercomputers with the highest performance ratings come from HPE".
"Supercomputing is not only about raw horsepower, what is in the box: the CPU, the GPU, the accelerator cards, which come with it. But it's also about how to build a massively parallel system", Milan explains. Through the acquisition of Cray, HPE has developed specialized knowledge in building massively parallel computers that function smoothly and support high levels of parallelism for various workloads. Additionally, as he notes, HPE has expanded its expertise to include management systems, software control solutions, and related services.
This "Cray DNA", as Avramović describes it, underpins three of the fastest supercomputers in Europe: one belongs to the Italian National Petrol Company, ENI, and two are based in research institutions, Swiss National Supercomputing Centre and LUMI in Finland. In South-Eastern Europe, HPE also has several installations both in private enterprises and in academia based in Croatia, Bosnia, Bulgaria and Serbia.
