HPE to build next-generation exascale and AI systems for Oak Ridge National Laboratory

HPE has announced that it has been selected to build two systems for the U.S. Department of Energy’s (DOE) Oak Ridge National Laboratory (ORNL), reinforcing the DOE’s mission to advance American leadership in artificial intelligence (AI) and supercomputing for science, energy, and national security.

The new systems include Discovery, a second-generation exascale supercomputer succeeding ORNL’s HPE-built Frontier, and Lux, a new AI cluster designed to accelerate AI and machine learning innovation on a multi-tenant, cloud-like platform.

Discovery: advancing the converged AI and HPC era

Discovery will be based on the new HPE Cray Supercomputing GX5000, HPE’s next-generation supercomputing platform built for leadership-class systems that unify AI and High-Performance Computing (HPC) architectures. The platform enables streamlined operations across distributed clusters and will be paired with the DAOS-based HPE Cray Supercomputing Storage Systems K3000.

Discovery will deliver new capabilities in AI, HPC, and quantum computing, increasing select application productivity tenfold. Scientists will gain accelerated breakthroughs in areas such as precision medicine, cancer research, nuclear energy, and aerospace.

“When we built Frontier for Oak Ridge National Laboratory and ushered in exascale, we achieved the pinnacle in supercomputing history and a triumph for the U.S.,” said Antonio Neri, President and CEO, HPE. “We are proud to build on that innovation and partnership with the U.S. Department of Energy, ORNL, and AMD to create Discovery and Lux, accelerating the next era of scientific discovery and AI innovation.”

Lux: a dedicated AI system for the future of research

Lux will be a direct liquid-cooled AI system based on HPE ProLiant Compute XD685, powered by AMD Instinct MI355X GPUs, AMD EPYC™ CPUs, and AMD Pensando™ networking. The platform is designed to provide researchers across the U.S. with sovereign, cloud-like access to AI resources for both training and inference.

“Discovery and Lux will expand the science researchers can achieve at Oak Ridge,” said Bronson Messer, Director of Science, Oak Ridge Leadership Computing Facility. “Discovery sets the stage for converged HPC, AI, and quantum computing, while Lux extends researcher access to AI resources. Together, they will transform productivity across critical fields of research.”

“For more than a decade, AMD and HPE have partnered to push the limits of high-performance computing,” said Dr Lisa Su, Chair and CEO, AMD. “With Oak Ridge National Laboratory, we are advancing the next generation of AI systems with Discovery and Lux, empowering researchers to accelerate innovation and strengthen America’s scientific leadership.”

Inside Discovery: the next-generation exascale supercomputer

Discovery’s breakthroughs will stem from the newly unveiled HPE Cray Supercomputing GX5000, extending 50 years of Cray innovation since the Cray-1 in 1975. The GX5000 is purpose-built for exascale with end-to-end capabilities spanning CPUs, GPUs, accelerators, networking, software, storage, and liquid cooling.

Key highlights include:
• Greater performance with optimised space: Scales to exascale performance using 25% less data centre space per rack.
• High-performance interconnect with HPE Slingshot: Provides high-bandwidth, low-latency interconnects for HPC, machine learning, and analytics workloads.
• Industry-first HPC DAOS storage performance: With the HPE Cray Supercomputing Storage Systems K3000, Discovery achieves 300% more IOPS per storage rack than Frontier, boosting AI application throughput. The K3000 is the industry’s first factory-built storage system embedding Distributed Asynchronous Object Storage (DAOS) open-source software.
• Next-generation, liquid-cooled, accelerated compute: Featuring AMD EPYC processors (codenamed “Venice”) and AMD Instinct MI430X GPUs, Discovery’s fully liquid-cooled compute infrastructure delivers high energy efficiency and cost-effectiveness.

Pioneering the next frontier of scientific discovery

As a global leader in supercomputing, HPE delivers end-to-end solutions and services backed by decades of AI and HPC expertise. Through unified management of infrastructure and applications, HPE’s supercomputing services continue to drive performance, innovation, and scientific progress for customers worldwide.