HPE to build next-generation “Discovery” supercomputer with AI cluster codenamed “Lux”

HPE is preparing to build a new exascale supercomputer, “Discovery,” and an AI cluster called “Lux,” for Oak Ridge Labs in the US, to enhance global scientific research capabilities.

November 2, 2568 – HPE announced it has been selected to build two supercomputer systems for the Oak Ridge National Laboratory (ORNL) of the U.S. Department of Energy (DOE) to support the DOE's mission to enhance U.S. leadership in artificial intelligence (AI) and supercomputing for scientific, energy, and national security applications.

The new system includes a second-generation exascale supercomputer, “Discovery,” which succeeds the HPE-built Frontier system that previously broke exascale speed records. HPE will also build a new artificial intelligence cluster, “Lux,” to support the U.S. Department of Energy’s program to develop AI and machine learning across a multi-tenant cloud platform.

The Discovery supercomputer will be built on the HPE Cray Supercomputing GX5000 platform , HPE's next-generation supercomputing platform for leadership-class systems. This platform integrates artificial intelligence (AI) and high-performance computing (HPC) architecture to enhance operational agility both within the center and across distributed clusters. It will be complemented by the HPE Cray Supercomputing Storage Systems K3000, which utilizes Distributed Asynchronous Object Storage (DAOS) technology. Working in conjunction with the HPE Cray Supercomputing GX5000, the Discovery supercomputer will offer groundbreaking capabilities for AI, HPC, and quantum computing.

It is expected to increase the performance of some applications by up to 10 times, helping scientists accelerate major discoveries in fields such as precision medicine, cancer research, nuclear energy, and the aerospace industry.

Antonio Neri, President and CEO of HPE, said, “When we built the Frontier system for Oak Ridge National Laboratory and ushered the world into the era of exascale, we made a landmark in supercomputing history and a source of pride for the United States. We are proud to continue our leadership in innovation and strong collaboration between the public and private sectors with the U.S. Department of Energy (DOE), Oak Ridge National Laboratory (ORNL), and AMD to create the Discovery and Lux ​​systems, which will accelerate a new era of scientific discovery and AI innovation.”

“Lux” will be an AI system built on a liquid-cooled HPE ProLiant Compute XD685 server, featuring an AMD Instinct MI355X graphics processor, an AMD EPYC™ CPU, and the AMD Pensando™ network. Designed to enhance access to AI resources, Lux will provide cloud-based access to researchers across the United States through a secure and sovereign AI Factory platform, allocating dedicated resources for model training and inference.

Discovery will build on the exascale computing capabilities first developed on the Frontier supercomputer HPE built for Oak Ridge National Laboratory (ORNL). With this superior capability, Discovery will open new opportunities for scientific discovery across a wide range of fields, furthering the laboratory's mission to innovate and strengthen the nation's technological security.

Bronson Messer, Director of Science at Oak Ridge Leadership Computing Facility, said, “We are excited that the Discovery and Lux ​​systems will expand the boundaries of research that scientists can conduct at Oak Ridge Labs. Discovery will pave the way for a new era of integrating high-performance computing (HPC), AI, and quantum computing, providing us with much deeper insights through interoperability with other systems, while Lux will broaden researchers’ access to specialized AI resources. Therefore, we expect both systems to be instrumental in creating a paradigm shift in research efficiency, enabling us to achieve unprecedented results in many world-class scientific fields.”

Dr. Lisa Su, President and CEO of AMD, said, “For over a decade, AMD and HPE have collaborated to push the boundaries of high-performance computing, creating solutions that enable new discoveries and transform the world. Partnering with Oak Ridge National Laboratory, we are ushering in a new era of AI with the Discovery and Lux ​​supercomputers, empowering researchers to accelerate innovation and further U.S. leadership in science and technology.”

The scientific advancements of the Discovery supercomputer will be driven by the recently announced HPE Cray Supercomputing GX5000. HPE is developing a new generation of supercomputing infrastructure for the era of AI and high-performance computing, building on supercomputing innovations dating back more than 50 years, since the Cray-1 was launched in 1975.

The HPE Cray Supercomputing GX5000 is specifically designed to support exascale computing and offers a comprehensive, cutting-edge suite of capabilities encompassing CPU, GPU, accelerators, networking, software, storage, and liquid cooling. Leveraging this new architecture, Discovery can deliver:

• Superior performance with maximum space utilization – This new platform is specifically designed to support exascale computing with higher density than its predecessor, reducing data center space per rack by up to 25%.
• High-performance connectivity with HPE Slingshot – The new HPE Slingshot delivers high-performance, modern connectivity to Discovery systems to support high-bandwidth, low-latency workloads for high-performance computing (HPC), machine learning, and analytics applications.
• Industry-leading HPC storage performance – Discovery is powered by the new HPE Cray Supercomputing Storage Systems K3000, which delivers up to 300% more IOPS per rack than Frontier systems to power AI applications. As the industry’s first factory-built storage system with built-in open source Distributed Asynchronous Object Storage (DAOS) software, the HPE Cray Supercomputing Storage Systems K3000 is a cost-effective all-flash storage system that works with the HPE Cray Supercomputing Storage Systems E2000, which uses the Lustre file system, which will also be deployed on Discovery.
• New generation processing with liquid cooling and accelerated compute units – Discovery systems will feature the new AMD EPYC processors (codenamed “Venice”) and AMD Instinct MI430X GPUs, delivering high performance and precision for modeling, simulation, and AI projects, leveraging HPE’s 50+ years of liquid cooling innovation. Discovery compute systems will utilize full liquid cooling to maximize energy efficiency and cost-effectiveness in supercomputing environments.

As the global leader in supercomputing, HPE offers customers end-to-end solutions and services with best-in-class AI and HPC expertise. As a key partner, HPE Supercomputing Services optimizes results through a fully centralized approach to managing enterprise infrastructure and applications, focusing on core business needs and continuous innovation.

Read news related to Technology circles around the world can be found here.