Intel, in partnership with Argonne National Laboratory and Hewlett Packard Enterprise (HPE), said its Aurora supercomputer broke the exascale barrier with 1,012 exaflops and also became the “world's fastest artificial intelligence system dedicated to artificial intelligence for open science” by achieving 10.6 exaflops of AI.
Aurora is a little behind the Frontier system at Oak Ridge National Laboratory in Tennessee, which previously reached 1,206 exaflops, but being the second supercomputer to break the exascale barrier is a huge achievement, especially considering Aurora is still working. and it is not yet completely complete.
HPC Cable ranked it No. 2 on its Top 500 list in November 2023. But that ranking was achieved with only “half of Aurora running the HPL benchmark.” In its most recent ranking, announced in May 2024, Aurora was still in second place (behind Frontier), but its new HPL score improved significantly from the 585.34 petaflops previously recorded.
Paving the way to tomorrow's discoveries
Designed from the ground up for AI research, Aurora comprises 10,624 compute blades, 21,248 Intel Xeon CPU Max Series processors, and 63,744 Intel Data Center GPU Max Series units, all housed in 166 racks.
Ogi Brkic, vice president and general manager of data center AI solutions at Intel, said: “The Aurora supercomputer that goes beyond exascale will pave the way to the discoveries of tomorrow. From understanding weather patterns to unraveling the mysteries of the universe, supercomputers serve as a compass that guides us toward solving truly difficult scientific challenges that can improve humanity.”
Future plans for Aurora range from climate change modeling to advances in fusion energy, atomistic simulations, and solving memory-related problems in the design of future power plants.
If you're interested in seeing what Aurora looks like, Argonne National Laboratory provided an all-access tour that you can check out below.