The NVIDIA Eos supercomputer delivers a staggering 18.4 Exascale (FP8) performance for AI applications, while also delivering about 9 Exascale in FP16 computing and 275 Pflops in FP64. The company announced its development almost a year ago and presented it in detail at the Supercomputing 2023 exhibition last November, and now it is ready for commissioning. The developers point out that Eos is capable of handling the most demanding AI workloads, including training large language models, recommendation algorithms, quantum modeling, and more. The supercomputer is ranked ninth in the TOP500 rating.
The high performance of NVIDIA Eos is powered by powerful NVIDIA accelerators and interfaces, as well as software such as NVIDIA Base Command and NVIDIA AI Enterprise.
“The Eos architecture is optimized for AI tasks that require minimal latency and high bandwidth connections within the accelerator cluster, making it ideal for enterprises looking to scale their AI systems. As companies and developers around the world seek to harness the power of AI, Eos is emerging as a key resource that promises to accelerate the development of AI-powered applications for every organization.“, NVIDIA said