HPE to Build Next-Gen Exascale Supercomputer and AI Cluster for Oak Ridge
▼ Summary
– HPE will build two new systems for the U.S. Department of Energy’s Oak Ridge National Laboratory to advance American leadership in AI and supercomputing for science, energy, and national security.
– The systems include “Discovery,” a second-generation exascale supercomputer, and “Lux,” a dedicated AI cluster that will provide cloud-like access for researchers.
– Discovery will be based on the new HPE Cray Supercomputing GX5000 platform and is expected to increase select application productivity tenfold compared to its predecessor, Frontier.
– Lux will be built on HPE ProLiant Compute XD685 with AMD components and serve as a sovereign AI factory for training and inference, expanding researcher access to AI resources.
– Discovery will feature advanced capabilities including greater density with 25% less data center space, a high-performance HPE Slingshot interconnect, and 300% more IOPS per storage rack than Frontier.
Hewlett Packard Enterprise has secured a major contract to construct two advanced computing systems for the U.S. Department of Energy’s Oak Ridge National Laboratory. This initiative supports the DOE’s objective of strengthening American leadership in artificial intelligence and supercomputing for scientific, energy, and national security applications. The project includes a next-generation exascale supercomputer named Discovery and a specialized AI cluster called Lux.
Discovery represents the successor to ORNL’s Frontier system, which was also built by HPE and previously broke the exascale speed barrier. It will be based on the new HPE Cray Supercomputing GX5000 platform, designed specifically for leadership-class systems. This platform integrates a unified AI and high-performance computing architecture to simplify operations across the entire site and among distributed clusters.
The system will be enhanced by the new HPE Cray Supercomputing Storage Systems K3000, a storage solution built on DAOS (Distributed Asynchronous Object Storage). Discovery is projected to boost productivity for certain applications by up to ten times compared to its predecessor, enabling scientists to accelerate progress in critical fields such as precision medicine, cancer research, nuclear energy, and aerospace engineering.
Antonio Neri, President and CEO of HPE, emphasized the importance of this next step. “When we delivered Frontier for Oak Ridge National Laboratory and ushered in the exascale era, it marked a historic achievement in supercomputing and a victory for the United States,” he stated. “We are honored to build on that legacy of innovation and our strong public-private collaboration with the DOE, ORNL, and AMD to bring Discovery and Lux to life, propelling the next wave of scientific and AI breakthroughs.”
Lux will function as a dedicated AI system built on the direct liquid-cooled HPE ProLiant Compute XD685. It will incorporate AMD Instinct MI355X GPUs, AMD EPYC CPUs, and AMD Pensando networking components. Designed to broaden access to AI resources, Lux will offer researchers across the nation cloud-like entry to a sovereign AI factory specifically equipped for training and inference tasks.
Bronson Messer, Director of Science for the Oak Ridge Leadership Computing Facility, expressed enthusiasm for the new systems. “We are eager for Discovery and Lux to expand the scope of research possible at Oak Ridge,” he commented. “Discovery will establish a new standard for converged HPC, AI, and quantum computing capabilities, while Lux significantly widens researcher access to specialized AI resources. We anticipate both systems will drive a paradigm shift in our productivity, achieving unprecedented advances in vital areas of scientific investigation.”
Dr. Lisa Su, Chair and CEO of AMD, highlighted the longstanding partnership. “For over ten years, AMD and HPE have collaborated to redefine the boundaries of high-performance computing, creating solutions that enable world-changing discoveries,” she noted. “Working alongside Oak Ridge National Laboratory, we are advancing the next generation of AI systems with Discovery and Lux, empowering researchers to accelerate innovation and reinforce America’s leadership in science and technology.”
The HPE Cray Supercomputing GX5000 platform underpinning Discovery represents decades of supercomputing evolution, tracing back to the Cray-1 introduced in 1975. It is engineered for the converged AI and HPC era, incorporating cutting-edge capabilities across CPUs, GPUs, accelerators, networking, software, storage, and liquid cooling.
The HPE GX5000 platform brings exascale performance in a more compact form, using 25% less data center space per rack than previous generations. It’s supported by HPE’s end-to-end supercomputing services, built on years of expertise in AI and high-performance computing. With a unified management system that simplifies infrastructure and application oversight, the platform enables organizations to boost efficiency, reduce operational complexity, and stay focused on driving innovation and achieving their core business goals.
(Source: ITWire Australia)
