AMD EPYC & Instinct GPUs are in Lux supercomputer

Credit: Oak Ridge National Laboratory · newsroom.amd.com

The newly deployed Lux SI + HPC supercomputer uses AMD Instinct MI355X GPUs alongside AMD EPYC processors, Pensando networking and AMD ROCm software. Built by HPE, the system integrates artificial intelligence with high-performance modeling and simulation to bolster U.S. capabilities in science, energy and national security. AMD reports the supercomputer demonstrates a full-stack approach to heterogeneous computing, bringing together hardware and software for demanding challenges. At Supercomputing 2026, AMD is demonstrating how this open computing is expanding the frontier of scientific discovery.

AMD EPYC & Instinct GPUs unlock quantum Lux supercomputer

The Lux SI + HPC supercomputer at Oak Ridge National Laboratory is now operational, integrating AMD Instinct MI355X GPUs to accelerate quantum error correction research. A new joint white paper details how the system achieves 6.23x end-to-end simulation throughput scaling from one to eight of these GPUs, directly linking quantum research with AMD’s high-performance computing infrastructure, DOE says. This scaling was demonstrated using the Pelagos software suite, a key component in the workflow.

The advancement allows for more complex quantum simulations, which are important for developing stable and scalable quantum computers. Built by HPE, the Lux system’s architecture extends beyond raw processing power, incorporating AMD EPYC processors, AMD Pensando networking and the AMD ROCm software platform. AMD reports this combination delivers a full-stack approach to heterogeneous computing, optimizing performance across diverse workloads.

The integration of ROCm is particularly significant, providing a unified software environment for both CPUs and GPUs, and enabling efficient use of the system’s resources. Beyond the Lux supercomputer, AMD is expanding its reach into AI-driven scientific computing globally. The LUMI-AI supercomputer in Finland combines AMD Instinct MI430X GPUs with 6th Gen AMD EPYC processors to increase capacity for artificial intelligence and research.

Similarly, an expansion of the Dawn supercomputer in the U.K. integrates AMD Instinct MI355X GPUs, provided by Dell Technologies, for use by researchers at the national AI Research Resource. AMD and Rancher Government Solutions are also collaborating to bring HPC and AI workloads together on Kubernetes, using a hardened reference architecture built on AMD Instinct GPUs and the RGS Carbide Platform.

Lux SI+HPC: Integrated AMD Architecture for AI & HPC

Ranked number 14 on the TOP500 list with 220.79 petaflops of HPL performance, the system’s design prioritizes a cohesive platform rather than segregated AI and HPC infrastructure. This approach, according to AMD, moves beyond simply adding accelerators to existing systems. Lux demonstrates the breadth of the AMD technology portfolio for AI-driven science, offering a single platform designed for heterogeneous scientific workflows.

The system’s architecture is intended to facilitate the smooth transition between traditional modeling and simulation tasks and the increasingly vital role of AI in scientific discovery, according to DOE. AMD reports that Lux provides the first dedicated U.S. AI Factory for the Genesis Mission, a DOE-led initiative focused on advancing AI-driven science.

The decision to build Lux with this specific AMD technology stack emphasises a commitment to open software ecosystems. AMD highlighted, “AMD Brings Agentic AI to Supercomputing with Open-Source AMD HPC Assistant, Debuting on LUMI and Frontier,” demonstrating the company’s broader strategy of providing accessible tools for scientific advancement. AMD announced, “PsiQuantum and AMD Collaborate to Build the Computing Foundation for Utility-Scale Quantum Computing,” demonstrating an intent to build classical foundations for future quantum systems.

AMD also stated, “AMD Builds the Classical Foundation for Quantum’s Future,” further emphasizing this long-term vision. The design of Lux also reflects a broader trend within the supercomputing landscape.

AMD technology now powers 202 systems on the TOP500 list, representing over 40% of the total and a 14% year-over-year increase. AMD demonstrated, “AMD Showcases 6th Gen AMD EPYC Processors to Lead Next Era of AI and Enterprise Computing,” signaling continued investment in the core components driving these advancements.

AMD also demonstrated, “AMD AI NIC Technology: AMD AI NICs Now Supports HPC to Accelerate AI for Science,” demonstrating a commitment to optimizing networking for these demanding workloads. Lux, therefore, is not an isolated deployment, but a key example of a wider strategy to deliver integrated, high-performance computing solutions for the future of scientific research.

Lux Supercomputer Debuts at No. 14 on TOP500 List

The Lux SI + HPC supercomputer achieved a measured HPL performance of 220.79 petaflops, securing its position as the No. 14 system on the TOP500 list and demonstrating a significant leap in computational capability for scientific research. This performance stems from the integration of AMD EPYC processors and AMD Instinct MI355X GPUs within a system architected by HPE, a combination designed to accelerate both traditional modeling and the emerging field of AI-driven discovery.

The deployment at Oak Ridge National Laboratory (ORNL) marks the opening of what AMD calls the “First AI Factory” for the DOE’s Genesis Mission, signaling a shift toward systems optimized for both simulation and artificial intelligence. Four of the world’s ten fastest supercomputers, El Capitan, Frontier, HPC7 and HPC6, rely on AMD technology, alongside four of the ten most energy-efficient systems.

This broad adoption emphasises the architectural flexibility and efficiency of AMD’s approach, allowing researchers to tackle increasingly complex problems across diverse scientific disciplines. The company’s commitment to open software is also evident in the integration of AMD ROCm, designed to support adaptation of applications to new computing resources, the company says. Researchers will benefit from the introduction of AMD HPC Assistant, an open-source, agentic AI guide intended to accelerate the process of moving from code development to scientific discovery.

AMD emphasizes the need for software that facilitates collaboration and adaptation, and ROCm is positioned as a key enabler of this goal. AMD states that open software is a core principle guiding its development of both hardware and software solutions for the future of scientific computing.

AMD Powers Over 40% of TOP500 Supercomputers Globally

This widespread adoption demonstrates a clear trend as AMD technology is present in more than half of all new entries to the list, signifying a growing reliance on its processors and accelerators for high-performance computing. Leading the charge are systems like El Capitan, Frontier, HPC7 and HPC6, all using AMD’s architecture to achieve top rankings. The recently deployed Lux SI + HPC supercomputer, ranked No. 14 on the TOP500 list, exemplifies this momentum.

This combination is not merely about raw power, delivering 220.79 petaflops of measured HPL performance, but about creating a platform designed for the evolving demands of scientific research. AMD announced that Lux is now open to scientific users at Oak Ridge National Laboratory, marking a public-private partnership intended to advance AI-driven science.

The system’s architecture, bringing together CPUs, GPUs, networking and open software on a single platform, is intended to support heterogeneous scientific workflows. This approach allows researchers to tackle larger problems and accelerate the pace of discovery by moving from questions to answers more efficiently.

Lux Enables First DOE AI Factory for Genesis Mission

Ranked 14th on the TOP500 list, Lux achieves 220.79 petaflops of measured HPL performance, a figure that emphasises its capacity for complex calculations and simulations. This system was developed through a collaborative effort between AMD and the DOE, demonstrating a commitment to public-private partnerships in advancing national research infrastructure, by the company’s account. The system’s architecture is intended to strengthen U.S. The deployment of Lux aligns with a broader trend of nations adopting AMD technologies to bolster research infrastructure and support scientific advancement.

In France, AMD was selected to build Alice Recoque, the country’s first exascale supercomputer, mirroring the investment in AI and HPC convergence seen with Lux. AMD’s focus on integrating both AI and traditional HPC environments is further exemplified by the AMD Instinct MI430X GPU, the first designed specifically for the convergence of high-precision simulation and AI for Science. AMD stated, “Exascale Was the Beginning: AMD Instinct MI430X GPUs,” signaling a continued push toward more powerful and versatile computing solutions.

Stay current

See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.

Avatar of Dr. Donovan

Latest Posts by Dr. Donovan: