Title: Groundbreaking Advances in High-Performance Computing Highlighted by Gordon Bell Prize Finalists
The prestigious Gordon Bell Prize, known for recognizing exceptional achievements in high-performance computing (HPC), has announced its five finalists for this year. These finalists have leveraged NVIDIA-powered supercomputers to make significant strides in various scientific fields, including climate modeling, materials science, fluid simulation, geophysics, and electronic design. These advancements were revealed during the Supercomputing Conference 2025 (SC25) and are accessible to the public on the open-access platform ArXiv. Here, we take a closer look at the projects and the supercomputers that have powered these groundbreaking achievements.
The Role of Supercomputers
The finalists’ work has been driven by some of the most advanced supercomputers available today, including:
- Alps – Located at the Swiss National Supercomputing Centre (CSCS), this supercomputer is powered by over 10,000 NVIDIA GH200 Grace Hopper Superchips. It has been instrumental in advancing scientific discoveries by providing unprecedented computational capabilities.
- Perlmutter – Hosted at the National Energy Research Scientific Computing Center (NERSC), Perlmutter is an NVIDIA-accelerated supercomputer that supports a wide range of scientific research.
- JUPITER – As Europe’s first exascale supercomputer, JUPITER is housed at the Jülich Supercomputing Centre (JSC) and utilizes the NVIDIA Grace Hopper platform along with Quantum-X800 InfiniBand networking to deliver exceptional performance.
These supercomputers have enabled researchers to push the boundaries of what is possible in scientific computation, leading to breakthroughs that were previously unimaginable. Thomas Schulthess, director of CSCS, emphasized the transformative impact of these technologies, stating, "Pushing computational boundaries turns bold targets into reality, delivering scientific revolutions that will redefine our world."
The Finalists and Their Projects
Let’s delve into the projects that have earned these finalists their place in the spotlight:
ICON: Modeling Earth at Kilometer-Scale
The ICON Earth system model, developed by a collaboration of researchers from the Max Planck Institute for Meteorology, German Climate Computing Centre, CSCS, JSC, ETH Zurich, and NVIDIA, represents a significant advancement in climate modeling. This model enables the simulation of the Earth’s systems at kilometer-scale resolution, offering a detailed view of energy, water, and carbon flows through the atmosphere, oceans, and land.
By achieving an unprecedented temporal compression, ICON can simulate approximately 146 days in a single 24-hour period. This allows for efficient climate simulations projecting decades into the future. According to Daniel Klocke, group leader at the Max Planck Institute for Meteorology, this level of resolution provides critical insights into the implications of future warming on both human and ecological systems.
ORBIT-2: Exascale Vision Foundation Models for Weather and Climate
ORBIT-2, developed by Oak Ridge National Laboratory in collaboration with NVIDIA, is an AI foundation model for weather and climate downscaling. Running on the Alps supercomputer, ORBIT-2 demonstrates unparalleled scalability and precision by utilizing exascale computing and algorithmic innovation.
This model overcomes challenges faced by traditional climate models by employing spatial hyper-resolution downscaling, a technique that generates high-resolution data from lower-resolution sources. As a result, researchers can capture and predict localized phenomena, such as urban heat islands and extreme weather events, with greater accuracy. Prasanna Balaprakash, director of AI programs at Oak Ridge National Laboratory, highlights the exceptional scalability and impact of ORBIT-2 at the intersection of AI and HPC.
QuaTrEx: Advancing Transistor Design Through Nanoscale Device Modeling
A team from ETH Zurich has pioneered QuaTrEx, a package of algorithms designed to enhance nanoscale electronic device modeling. Running on the Alps supercomputer with NVIDIA GH200 Superchips, QuaTrEx can simulate devices with over 45,000 atoms, enabling faster and more accurate design of next-generation transistors, known as NREFTs.
Mathieu Luisier, full professor of computational nanoelectronics at ETH Zurich, notes that access to the Alps supercomputer was instrumental in achieving these simulations, which were previously unimaginable just a few months ago.
Simulating Spacecraft at Record-Breaking Scales With the MFC Flow Solver
The Georgia Institute of Technology, in collaboration with NVIDIA, has developed MFC, an open-source solver for fluid flow simulation. This tool, running on the Alps supercomputer, achieves fluid flow simulation four times faster and with over five times greater energy efficiency compared to previous world records.
MFC’s capabilities allow for detailed simulations of rocket engines, paving the way for more accurate design of critical spacecraft components. Spencer Bryngelson, assistant professor at the Georgia Institute of Technology, highlights the significant improvements in simulating complex fluid flows, which enable unprecedented scale simulations of rocket engine plumes.
A Digital Twin for Tsunami Early Warning
A collaboration between the University of Texas at Austin, Lawrence Livermore National Laboratory, and the University of California San Diego has resulted in the development of a digital twin for tsunami early warning. This system, applied to the Cascadia subduction zone in the Pacific Northwest, performs complex computations at unprecedented speeds.
The digital twin combines real-time sensor data with full-physics modeling and uncertainty quantification, providing a predictive framework for emergency-response systems across various hazards. Omar Ghattas, professor of mechanical engineering at UT Austin, emphasizes the potential for this framework to offer actionable insights before disasters occur.
Conclusion
The finalists for the Gordon Bell Prize have demonstrated the transformative potential of high-performance computing in advancing scientific understanding across a range of fields. By leveraging NVIDIA-powered supercomputers, these researchers have achieved breakthroughs that promise to redefine our understanding of the world and our place within it. As these projects continue to evolve, they will undoubtedly contribute to the development of solutions for some of the most pressing global challenges of our time.
For those interested in learning more about these advancements, the Supercomputing Conference 2025 offers an opportunity to explore the latest developments in supercomputing technology and its applications in various scientific domains.
For more Information, Refer to this article.
