Amir Shehata from Oak Ridge National Laboratory has outlined a software architecture that would integrate emerging quantum computers with the world’s fastest supercomputers, such as ORNL’s exascale machine, Frontier, which achieved speeds of more than 1 quintillion calculations per second in 2022. The design introduces a unified resource-management system, a flexible quantum-programming interface, and a quantum-platform-management interface that together enable a quantum accelerator to be deployed alongside Frontier, potentially allowing the supercomputer to model the equivalent of 50–60 qubits in a single run. By prioritising data traffic and providing a performance-portable hybrid application tool chain, the framework aims to unlock an exponential increase in the scale of problems that can be tackled, from high-resolution digital simulations to complex scientific modelling.
ORNL Blueprint for Quantum HPC Integration at Oak Ridge
On August 29, 2025, Oak Ridge National Laboratory (ORNL) released a comprehensive study that outlines a software architecture designed to integrate quantum computers with the world’s fastest supercomputers, notably the exascale machine Frontier. Lead author Amir Shehata, a senior software engineer at ORNL, framed the initiative as a way to “get ahead of the curve and drive development with as many people as possible participating.”
The blueprint builds upon a 2024 ORNL paper that outlined foundational strategies for quantum-HPC integration, and it now provides concrete design guidelines for translating those strategies into operational systems. Central to the proposal is a unified resource-management system that coordinates quantum and classical resources, a flexible quantum-programming interface that abstracts hardware-specific details, and a quantum-platform-management interface that simplifies the integration of diverse quantum hardware. A comprehensive tool chain for quantum‑circuit optimisation and execution is also included, together with a quantum controller that translates between the classical HPC environment and the quantum accelerator and a scheduling algorithm that directs data traffic to maximise throughput.
The study quantifies the potential of the architecture by noting that Frontier’s peak performance exceeded 1 quintillion calculations per second in 2022; however, it could model only about 50–60 qubits, as each additional qubit doubles the computational demand. In contrast, contemporary quantum machines can, in theory, support hundreds of qubits, a gap that the ORNL design seeks to bridge. The architecture is deliberately modular, allowing it to accommodate future quantum technologies, such as neutral-atom arrays, trapped-ion chains, superconducting circuits, and other emerging platforms, without requiring a redesign of the software stack. The study is supported by the Department of Energy’s Advanced Scientific Computing Research programme and ORNL’s Laboratory‑Directed Research and Development programme, and it is carried out within the Oak Ridge Leadership Computing Facility (OLCF), a user facility of the DOE Office of Science that UT-Battelle manages. ORNL’s overarching mission is to harness emerging quantum technologies to accelerate scientific discovery while maintaining leadership in high‑performance computing.
Flexible Software Stack to Couple Quantum Accelerators with Frontier Exascale
The software stack proposed by ORNL is engineered to maximise the synergy between quantum accelerators and Frontier’s exascale capabilities. At its core lies a unified resource‑management system that dynamically allocates compute cycles, memory, and network bandwidth across the quantum and classical layers, ensuring that neither domain becomes a bottleneck. Complementing this is a quantum-programming interface that presents developers with a high-level, language-agnostic API, allowing them to express quantum algorithms without requiring intimate knowledge of the underlying qubit technology. The quantum-platform-management interface, meanwhile, provides a thin abstraction that exposes the essential control primitives of each quantum device—whether it is a neutral-atom array, a trapped-ion chain, or a superconducting circuit—so that the same orchestration logic can be reused across platforms. Together, these interfaces feed into a sophisticated scheduling engine that prioritises data traffic, balances load, and guarantees that any quantum invocation delivers a net performance benefit. The tool chain for quantum‑circuit optimisation and execution further refines the process by automatically tailoring gate sequences to the idiosyncrasies of each device, reducing latency and error rates. Tom Beck, co‑author of the study and head of the Science Engagement Section at ORNL’s National Centre for Computational Sciences, highlighted the transformative potential of the approach, noting that “harnessing quantum advantage could pump up our problem‑solving capacity.” This perspective underscores the study’s ambition to unlock new scientific horizons by marrying quantum speed with classical scale.
Unified Resource Management and Quantum Programming Interface for Future Hardware
At the heart of the ORNL design is a resource‑management framework that treats quantum and classical elements as a single, coherent pool. The scheduler embedded within this framework is tasked with prioritising data traffic, ensuring that quantum calls are only dispatched when the classical host can absorb the resulting computational load without incurring a net slowdown. The scheduler also monitors a suite of performance metrics in real time, including execution speeds, qubit fidelity, and gate latency, providing the system with the situational awareness needed to avoid bottlenecks. The quantum-controller component acts as an interpreter, translating high-level orchestration commands into device-specific control signals and routing data back to the classical back-end once a quantum operation is completed. By feeding performance data back into the scheduler, the system can adaptively re‑allocate resources, throttle or pre‑empt quantum jobs, and optimise overall throughput. This tight feedback loop is essential for maintaining the delicate balance between the immense parallelism of Frontier and the comparatively fragile operations of contemporary quantum devices.
Anticipating Quantum Evolution: Modular Design for Emerging Qubit Technologies
The ORNL blueprint recognises that the quantum landscape is rapidly evolving, and it therefore adopts a modular design that can absorb new qubit technologies without necessitating a wholesale redesign of the software stack. The architecture is explicitly compatible with neutral‑atom arrays, trapped‑ion chains, superconducting circuits, and any other platform that may emerge in the coming decade. The study also provides a quantitative assessment of the theoretical limits of coupling: Frontier’s one quintillion‑per‑second throughput, while impressive, can only model roughly 50–60 qubits because each additional qubit effectively doubles the computational demand on the classical host.
In contrast, state‑of‑the‑art quantum machines can, in principle, support hundreds of qubits, a disparity that the modular design seeks to bridge. The design also accounts for the high error rates currently plaguing quantum devices, offering a flexible error-mitigation framework that can be tuned to the specific noise profile of each platform. Internationally, comparable initiatives in Europe and Japan are pursuing similar harmonisation of software standards and benchmarking protocols, recognising that a globally interoperable ecosystem will accelerate progress. By embedding these considerations into the core architecture, ORNL ensures that the system remains future‑proof, capable of integrating any qubit medium that may become viable without requiring a fundamental overhaul of the orchestration logic.
DOE Funding and International Collaboration Driving Hybrid Supercomputing
The ORNL study is underpinned by significant federal investment, drawing on the Department of Energy’s Advanced Scientific Computing Research programme and ORNL’s own Laboratory‑Directed Research and Development initiative. The Oak Ridge Leadership Computing Facility (OLCF), where the research is conducted, is a DOE Office of Science user facility managed by UT-Battelle, and it serves as the primary testbed for the proposed quantum-HPC integration. Beyond domestic funding, the study acknowledges a growing international momentum: parallel projects in Europe and Japan are working to harmonise software standards, share benchmarking data, and co‑develop frameworks that will enable seamless cross‑border collaboration. These efforts include joint experiments that deploy quantum controllers and scheduling engines on shared supercomputing resources, thereby validating the ORNL architecture in a real‑world, multi‑institutional setting. By aligning with these global initiatives, ORNL positions itself at the nexus of a worldwide effort to unlock the full potential of hybrid quantum‑classical computing, ensuring that the United States remains at the forefront of both scientific discovery and high‑performance computing innovation.
Original Press Release
Source: Oak Ridge National Laboratory (U.S. Department of Energy, government agency)
View Original Source
Tags:
Amir Shehata Department of Energy Frontier Supercomputer Oak Ridge National Laboratory Quantum Circuit Optimization quantum-HPC software stack Rafael Ferreira da Silva Tom Beck United States
