The world’s fastest supercomputer dedicated to artificial intelligence (AI) for open science has been officially named Aurora, after surpassing the exascale barrier with a whopping 1,012 exaflops. This impressive achievement was made possible by a joint project between the Argonne National Laboratory in the United States and technology giants Intel and Hewlett Packard Enterprise (HPE).
The main purpose of Aurora is to advance scientific and engineering research powered by AI. One of its key achievements is the training of a generative AI chatbot named Aurora GPT, as well as expanding its capabilities with exascale computing. This means that the supercomputer has the capacity to perform at least one exaflop operation per second, making it a groundbreaking advancement in the field of AI.
During the ISC High Performance 2024 event in Hamburg, Germany, Intel announced that Aurora had broken the exascale barrier by reaching 1,012 exaflops using 9,234 nodes, representing 87 percent of the system’s capacity. This milestone solidifies Aurora’s position as the fastest AI system dedicated to open science.
Aurora is a massive system comprised of 166 racks, 10,624 compute blades, 21,248 Intel Xeon CPU Max Series processors, and 63,744 Intel Data Center GPU Max Series units. It also features an open Ethernet-based supercomputing interconnection with 84,992 HPE Slingshot Fabric endpoints, making it a state-of-the-art AI system for scientific research. In addition to its AI capabilities
+ There are no comments
Add yours