Nvidia GTC: DDN Delivers 4TB/s of Accelerated Storage Performance in Eos AI Supercomputer

At Nvidia GTC, DDN (DataDirect Networks Inc.) announced its implementation of EXAScaler AI storage (A³I) in Nvidia Eos, a TOP10 supercomputer based on Nvidia DGX SuperPOD. 

Incorporating 576 Nvidia DGX H100 systems, Quantum-2 IB networking and Nvidia AI Enterprise software within the DGX SuperPOD reference architecture, the company has architected a performance, power-efficient storage solution with 48 A³I appliances in less than 3 racks and 100kW of power consumption.

The result is 12PB of DDN storage delivering 4TB/s of data speed to serve an 18.4 exaflops of FP8 Nvidia AI performance. NvidiaEos powers the largest AI workloads, large language models, recommender systems, and simulations delivering Nvidia’s latest AI innovations.

“Enabling leading whole-system efficiency is our aim, not just delivering efficient storage,” said Paul Bloch, president and co-founder, DDN. “By integrating with the Nvidia Eos supercomputer, DDN is demonstrating our ability to deliver shorter time to value while reducing risks for both on-prem and cloud partners using one of the most powerful GPU architectures in the world.”

“Built with Nvidia DGX H100 systems, Nvidia‘s Eos supercomputers power advanced AI research and development,” said Charlie Boyle, VP, DGX platform, Nvidia. “Nvidia DGX SuperPODs like Eos provide a full-stack blueprint for turnkey enterprise AI supercomputing, and DDN A³I delivers high performance, power efficient storage for Eos to deliver the data needed for advanced AI training.”

DDN storage in Nvidia Eos highlights what organizations and sovereign AI entities in data centers and the cloud can achieve. Delivering the benefits of generative AI and AI frameworks at scale in a cost-effective and power-efficient way is now a reality, from life sciences research and and drug discovery to autonomous driving and personal assistant chatbots and beyond.