Nvidia unveils first look at Eos, its latest data center-scale supercomputer

AI SaaS

Nvidia Corp. today provided the first public look at the architecture powering Eos, the company’s latest data center-scale supercomputer designed for accelerating artificial intelligence development, described as an “AI factory.”

Revealed in November during the Supercomputing 2023 trade show, Eos is named after the Greek goddess who is the personification of dawn. The supercomputer is built with 576 Nvidia DGX H100 systems combined with Quantum-2 InfiniBand networking and software to provide a total of 18.4 exaflops of FP8 AI performance.

Nvidia said it built Eos based on prior knowledge from previous supercomputer architectures, including the Nvidia SaturnV and the modular supercomputer Selene. The entire system architecture is built into an extremely large-scale DGX SuperPOD, which allows system architects to rapidly scale up AI data center solutions with extremely high performance.

Under the hood, every DGX H100 supports eight H100 Tensor Core APUs and Eos itself features a total of 4,608 H100 GPUs. This means that the entire system will be able to handle tremendous workloads to train large language models, AI recommenders, large-scale analytics, quantum simulations and more, the company said.

According to Nvidia, the architecture for Eos has been optimized for AI workloads, which demand ultra-low latency and high throughput across large clusters of accelerated computing nodes. Using Nvidia’s own networking technology, the supercomputer can support data transfer speeds of up to 400GB/s, which are needed to move extremely large datasets for training AI models.

Software integrated into the Eos supercomputer includes offerings such as Base Command and AI Enterprise. Base Command provides powerful AI workflow management, cluster management and libraries for accelerated compute, storage and network infrastructure for running AI workloads. AI Enterprise is a cloud-native platform designed to help accelerate and streamline the development and deployment of AI applications, including generative AI by being the “operating system” for enterprise AI, the company said.

In November, Eos won the ninth slot on the TOP500 list of the world’s fastest supercomputers.

Photo: Nvidia

Your vote of support is important to us and it helps us keep the content FREE.

One click below supports our mission to provide free, deep, and relevant content.  

Join our community on YouTube

Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts.

“TheCUBE is an important partner to the industry. You guys really are a part of our events and we really appreciate you coming and I know people appreciate the content you create as well” – Andy Jassy

THANK YOU

AI SaaS

Leave a Reply

Your email address will not be published. Required fields are marked *