The Argon National Laboratory (ANL) of the US Department of Energy announced the availability of the Aurora Exflopse supercomputer for researchers around the world. As indicated in the press release, thanks to the wide possibilities of modeling, AI and data analysis, Aurora will contribute to breakthroughs in a number of areas, including designing aircraft, cosmology, the development of drugs and research in the field of nuclear energy.
Michael Papka, director of Argonne Leadership Computing Facility (Alcf), computing Center for Science of the US Department of Energy, noted that the first projects using Aurora have demonstrated its huge potential. “We look forward to how a wider scientific community will use the system to transform its research,” he said.
Aurora has already established itself as one of the world’s leaders in the AI performance, taking first place in the HPL-MXP benchmark in November 2024, ANL noted. The capabilities of the machine to perform AI-Zadach are used by scientists to open new materials for batteries, develop new drugs and accelerate research in the field of thermonuclear energy. Before deploying it, the ANL team under the leadership of Aurora, using it for teaching AI models for protein modeling.
Image source: Anl
Among the first projects implemented with the help of Aurora are the development of high -precision models of complex systems, such as the human circulatory system, nuclear reactors and supernova stars. In addition, the supercomputer’s ability to process huge data sets is crucial for the analysis of growing data streams from large research plants, such as an improved photon source (APS) of the Argon National Laboratory, Scientific Objects of Science of the US Department of Energy (DOE) and a large adron -adronical collider of European Organizations of nuclear research (CERN).
In order to guarantee Aurora’s readiness for use for scientific research from the first day of launch, the so -called joint design was used when creating it. Using this approach, the Aurora team has developed hardware and software in tandem to optimize performance and ease of use. This required many years of cooperation between Alcf, Intel, HPE and researchers throughout the country participating in the Exascale Computing Project project of the US Department of Energy and the Aurora Early Science Program (ESP) of the Center.
While the Aurora installation work was carried out, the ECP and ESP commands launched apps for stress testing of equipment, simultaneously optimizing their code for the most efficient operation in the system. As a result, dozens of scientific applications, as well as a wide range of software and development tools, were ready even before Aurora was commissioned, the press release said.