ACM Gordon Bell Prize-winning team develops revolutionary simulation for tsunami prediction
Innovative “Digital Twin” yields Ten-Billion-Fold Speedup Over Existing State-of-the-Art Methods
Association for Computing Machinery
image:
The digital twin developed by this year’s winning team enables real-time, data-driven tsunami forecasting with dynamic adaptivity to complex source behavior.
view moreCredit: Association for Computing Machinery
ACM, the Association for Computing Machinery, named an eight-member team drawn from US institutions as the winner of the 2025 ACM Gordon Bell Prize for their project, “Real-time Bayesian inference at extreme scale: A digital twin for tsunami early warning applied to the Cascadia subduction zone.” The ACM Gordon Bell Prize tracks the progress of parallel computing and rewards innovation in applying high-performance computing to challenges in science, engineering, and large-scale data analytics.
Existing state-of-the art high-performance computing simulations for early tsunami warning are developed primarily through models which process seismic data. The drawbacks of these approaches include: 1) They do not allow for enough warning time, as destructive tsunami waves can arrive onshore in under ten minutes, and 2) They fail to capture the complexities of earthquake ruptures which cause the Tsunamis.
The Gordon Bell Prize-winning team created a far more predictive early warning framework by developing a full-physics Bayesian inversion framework—popularly called “digital twin.” A digital twin is a virtual simulation of a physical process (or object) that uses real-time data from sensors to match its physical counterpart. The digital twin developed by this year’s winning team enables real-time, data-driven tsunami forecasting with dynamic adaptivity to complex source behavior.
With this approach, they achieved the fastest time-to-solution of a partial differential equation (PDE)-based Bayesian inverse problem with 1 billion parameters in 0.2 seconds, a ten-billion-fold speedup over the existing state-of-the-art. This is the largest-to-date unstructured mesh finite element (FE) simulation with 55.5 trillion degrees of freedom (DOF) on 43,520 GPUs, with 92% weak and 79% strong parallel efficiencies in scaling over a 128× increase of GPUs on the full-scale El Capitan system—the world’s largest supercomputer.
The team simulated a Tsunami in an area in the Pacific Ocean called the Cascadia Subduction Zone, which stretches 1000 km from northern California to British Columbia. This area has been eerily quiet for over 300 years—but is considered overdue for a magnitude 8.0–9.0 megathrust earthquake.
The members of the ACM Gordon Bell Prize-Winning team are Stefan Henneking, Sreeram Venkat, Milinda Fernando, and Omar Ghattas (all of The University of Texas at Austin); Veselin Dobrev, John Camier, Tzanio Kolev (all of Lawrence Livermore National Laboratory); and Alice-Agnes Gabriel (University of California San Diego).
Honorable Mention
This year an Honorable Mention for the ACM Gordon Bell Prize was given to a 10-member team from ETH Zurich for their project “Ab-initio Quantum Transport with the GW Approximation, 42,240 Atoms, and Sustained Exascale Performance.” Team members include Nicolas Vetsch, Alexandros Nikolaos Ziogas, Alexander Maeder, Vincent Maillou, Anders Winka, Jiang Cao, Grzegorz Kwasniewski, Leonard Deutschle (also affiliated with NVIDIA), Torsten Hoefler, and Mathieu Luisier.
The ACM Gordon Bell Prize was presented today during the International Conference for High-Performance Computing, Networking, Storage and Analysis (SC25) in St. Louis, Missouri.
About ACM
ACM, the Association for Computing Machinery is the world’s largest educational and scientific computing society, uniting computing educators, researchers, and professionals to inspire dialogue, share resources and address the field’s challenges. ACM strengthens the computing profession’s collective voice through strong leadership, promotion of the highest standards, and recognition of technical excellence. ACM supports the professional growth of its members by providing opportunities for life-long learning, career development, and professional networking.
About the ACM Gordon Bell Prize
The ACM Gordon Bell Prize is awarded each year to recognize outstanding achievement in high-performance computing. The purpose of this recognition is to track the progress over time of parallel computing, with particular emphasis on rewarding innovation in applying high-performance computing to applications in science. The prize is awarded for peak performance as well as special achievements in scalability and time-to-solution on important science and engineering problems and low price/performance. Financial support for the $10,000 awards is provided by Gordon Bell, a pioneer in high-performance and parallel computing.
###
First-ever full Earth system simulation provides monumental new tool to understand climate change
Breakthrough in climate modelling integrates weather and climate processes with far greater precision than previous frameworks
Association for Computing Machinery
image:
The Gordon Bell Climate Prize-winning team reached a landmark this year by being the first team ever to develop a Full Earth Simulation at 1 km (extremely high) Resolution.
view moreCredit: Association for Computing Machinery
St. Louis, MO, November 20, 2025 – ACM, the Association for Computing Machinery, today presented a 26-member team with the ACM Gordon Bell Prize for Climate Modelling in recognition of their project “Computing the Full Earth System at 1 km Resolution.” The award honors innovative contributions to parallel computing toward solving the global climate crisis.
Climate change is responsible for more extreme hurricanes, more destructive wildfires, severe droughts, and increased human disease, among other harmful outcomes. Experts warn that if carbon emissions are not significantly reduced within a few decades, the damage to the Earth’s ecosystem will be irreversible.
Among the most effective tools scientists have developed to understand climate change are digital simulations of the Earth. These simulations are produced by developing specific algorithms to run on the world’s most powerful supercomputers. But simulating how human activity influences the climate has been an extraordinarily difficult challenge. A mind-boggling number of variables need to be taken into consideration—such as the cycles of water, energy, and carbon, how those factors relate to each other, and how diverse physical, biological, and chemical processes interact over space and time. For these reasons, previous state-of-the-art simulations have not been able to achieve what is referred to as a “Full Earth System” simulation.
The Gordon Bell Climate Prize-winning team reached a landmark this year by being the first team ever to develop a Full Earth Simulation at 1 km (extremely high) Resolution. In their introduction, they explain, “We present the first-ever global simulation of the full Earth system at 1.25 km grid spacing, achieving highest time compression with an unseen number of degrees of freedom. Our model captures the flow of energy, water, and carbon through key components of the Earth system: atmosphere, ocean, and land. To achieve this landmark simulation, the team harnessed the power of 8192 GPUs on Alps and 4096 GPUs on JUPITER, two of the world’s largest GH200 superchip installations.”
The groundbreaking innovations the team employed to make the Full Earth Simulation possible include (1) exploiting functional parallelism by efficiently mapping components to specialized heterogeneous systems and (2) simplifying the implementation and optimization of an important component by separating its implementation in Fortran from the optimization details of the target architecture.
In the conclusion to their paper they write, “This has enormous and enduring potential to provide full global Earth system information on local scales about the implications of future warming for both people and eco-systems, information that otherwise would not exist.”
Team Members
The members of the team are: Alexandru Calotoiu, Torsten Hoefler, Yakup Budanaz, Pratyai Mazumder, Marcin Copik, and Benjamin Weber (all of ETH Zurich); Jan Frederik Engels, Hendryk Bockelmann, and Claudia Frauen (all of Deutsches Klimarechenzentrum), Dmitry Alexeev (NVIDIA), Daniel Klocke, René Redler, Reiner Schnur, Helmuth Haak, Luis Kornblueh, Cathy Hohenegger, and Bjorn Stevens (all of the Max Planck Institute for Meteorology); Fatemeh Chegini (University of Hamburg), Manoel Römmer, Lars Hoffmann, Sabine Griessbach, Mathis Bode, and Andreas Herten (all of Forschungszentrum Jülich); as well as Jonathan Coles, Miguel Gila, and William Sawyer (all of the Swiss National Supercomputing Centre).
The ACM Gordon Bell Prize for Climate Modelling was presented today at the International Conference for High Performance Computing, Networking, Storage and Analysis (SC25), held in St. Louis, Missouri.
About ACM
ACM, the Association for Computing Machinery is the world’s largest educational and scientific computing society, uniting computing educators, researchers, and professionals to inspire dialogue, share resources, and address the field’s challenges. ACM strengthens the computing profession’s collective voice through strong leadership, promotion of the highest standards, and recognition of technical excellence. ACM supports the professional growth of its members by providing opportunities for life-long learning, career development, and professional networking.
About the ACM Gordon Bell Prize for Climate Modeling
The ACM Gordon Bell Prize for Climate Modelling recognizes innovative parallel computing contributions toward solving the global climate crisis. Climate scientists and software engineers are evaluated for the award based on the performance and innovation in their computational methods. A cash prize in the amount of $10,000 accompanies the award, which was conceived and funded by Gordon Bell, a pioneer in high performance computing and researcher emeritus at Microsoft Research. Recipients of the ACM Gordon Bell Prize for Climate Modelling will have their research published in The International Journal of High Performance Computing Applications (IJHPCA).
No comments:
Post a Comment