Yazılar

Germany Launches Nvidia-Powered Jupiter Supercomputer to Compete in Global AI Race

German Chancellor Friedrich Merz on Friday inaugurated Jupiter, a Nvidia-powered supercomputer that ranks as the fourth-fastest in the world and Europe’s first Exascale-class machine. Located at the Juelich Research Centre, Jupiter can perform one billion billion (10^18) calculations per second—equivalent to the power of about 10 million laptops.

Built in partnership with France’s Atos and Germany’s ParTec, Jupiter represents a landmark effort for Europe to catch up with the U.S. and China in high-performance computing and AI infrastructure. “We are today witnessing a historic European pioneering project,” Merz said, stressing that Europe has the opportunity not only to close the gap but also to remain competitive in the long run.

Beyond its symbolic importance, Jupiter is expected to advance research in biotechnology, climate modeling, and AI development, reducing Europe’s reliance on foreign digital infrastructure. Ralf Wintergerst, head of German tech association Bitkom, called for easy access to the system for startups and established firms to maximize its impact on innovation.

With Jupiter now operational, Germany positions itself at the heart of Europe’s AI and scientific research ecosystem, while aiming to balance global competition with technological sovereignty.

Supercomputer Frontier Models the Universe with Unprecedented Detail

A groundbreaking cosmic simulation has been achieved using the Frontier supercomputer, one of the most advanced computing systems in the world. This simulation offers an unprecedented level of detail in modeling the observable universe, incorporating not only gravitational forces but also complex interactions involving dark matter, gas, and plasma. The ability to simulate such intricate cosmic phenomena represents a major leap forward in our understanding of the universe’s large-scale structures and evolutionary processes.

The simulation was conducted as part of the U.S. Department of Energy’s Exascale Computing Project, which aims to push the boundaries of computational science. Using the Hardware/Hybrid Accelerated Cosmology Code (HACC), the research team at Oak Ridge National Laboratory (ORNL) leveraged Frontier’s immense processing power to run calculations at speeds nearly 300 times faster than previous cosmological models. This breakthrough showcases the potential of exascale computing in tackling some of the most complex problems in astrophysics.

A key component of this research was the application of hydrodynamic cosmology, which integrates dark matter and energy with traditional gravitational interactions. Previous simulations primarily focused on gravity’s role in shaping the cosmos, but the new model provides a more holistic view by incorporating additional physical factors. To achieve this, the researchers utilized 9,000 computing nodes, each equipped with AMD Instinct MI250X graphics processors, allowing for higher-resolution simulations than ever before.

The success of this simulation underscores the transformative impact of supercomputing on scientific discovery. By replicating the universe’s intricate processes with unparalleled accuracy, researchers can refine existing theories of cosmic evolution and gain deeper insights into fundamental astrophysical questions. As computational power continues to advance, future simulations may unlock even more mysteries about the formation and behavior of the universe on the grandest scales.