The Aurora Supercomputing System is a supercomputer project of the U.S. Department of Energy’s Argonne National Laboratory. It will be built on the basis of Intel Sapphire Rapids and Ponte Vecchio to achieve ExaFLOP computing power. Previously, the project was delayed due to Intel R&D issues.
Today’s topics will mostly focus on Intel’s release of Alder Lake, but at the “Intel Innovation” summit held in San Francisco, Intel also talked about data centers and supercomputers. At present, Intel expands through the heterogeneous structure of CPU, GPU, application accelerator, interconnection processor, edge computing device and FPGA, and binds it into a complete solution. The goal is to work with the open ecosystem to ensure that developers have optimized tools and software environments to accelerate deployment.
Intel said that the Aurora supercomputing system is designed to handle high-performance computing, AI/ML, and big data analysis workloads, and can achieve 2 ExaFLOP peak computing power. In addition, SiPearl is designing Europe’s first supercomputer with ExaFLOP-level computing power and has chosen to cooperate with Intel, using Ponte Vecchio and CXL (Compute Express Link) for standardized connection computing.
In fact, in the next period of time, there may be multiple ExaFLOP-level supercomputers deployed in the world. It is reported that two sets of new-generation supercomputers have been deployed in China this year, reaching the peak computing power of 1.3 ExaFLOPS and the continuous computing power of 1.05 ExaFLOPS. However, there is relatively little information at present, and none of them are included in the global supercomputing Top500 list. Its computing power is lower than the 1.5 ExaFLOPS of the supercomputing system built with AMD EPYC processor and CDNA 2 architecture Instinct computing card, but the deployment was completed nearly a year earlier.