The chip giant has confirmed that it has finally managed to complete the installation of the Aurora Blade supercomputer, a very important project that has been the result of the collaboration between Intel and HPEand which has also had the participation of the United States Department of Energy.
This supercomputer is one of the most powerful in its class, and is prepared to offer a very high level of performance in simulations, data analysis and artificial intelligence. The system incorporates more than 1,024 storage nodes (using DAOS, Intel’s Distributed Asynchronous Object Storage) for a total of 220TB capacityoffers an impressive bandwidth of 31TB/s and leverages the high-performance HPE Slingshot fabric.
Aurora Blade will be the world’s first supercomputer to reach a theoretical peak performance of more than 2 EXAFLOPS. Each EXAFLOP is equivalent to approximately one trillion operations per second, a power level that is truly impressive, and which has been possible to achieve thanks to the integration of 10,624 blades containing the key components of the equipment.
These 10,624 blades add up, in total, 63,744 Intel Data Center Max Series GPUs and total 21,248 Intel Xeon CPU Max Series processors. Aurora will harness the full power of the Intel Max Series family of GPUs and CPUs, designed to seamlessly adapt to the unique needs of the world of High Performance Computing (HPC) and AI.
Each blade consists of two Intel Xeon Max Series CPUs and six Intel Max Series GPUs. Developers are already using oneAPI and AI tools to accelerate HPC and AI workloads and improve code portability across multiple architectures.
It is an important achievement because responds to the real needs of researchers, who are facing ever greater challenges, and who need more advanced technologies and equipment to be able to face them. In this sense we could highlight, for example, the fight against climate change and the search for cures against lethal diseases.
Researchers from ALCF’s Aurora Early Science Program (ESP) and DOE’s Exascale Computing project, migrate their work from the Sunspot benchmark to the Aurora fully installed. This transition will allow them to scale their applications across the entire system. The first users will subject the supercomputer to stress tests and identify possible failures that need to be resolved prior to deployment.
