20-qubit Superconducting Quantum Computer Integration with HPC Resources Yields Key Facility and Scheduler Lessons

Integrating quantum computers with existing high-performance computing infrastructure represents a crucial advance in modern scientific capability, and a team led by Eric Mansfield and Stefan Seegerer of IQM Quantum Computers, along with Panu Vesanen and colleagues, now demonstrates a practical implementation of this integration at the Leibniz Supercomputing Centre. This work details the successful connection of a 20-qubit quantum computer to a major HPC facility, yielding valuable insights into the challenges and solutions for future deployments. The team’s experience reveals that while quantum computers demand stringent facility requirements, careful planning and site surveys ensure feasible integration, and that automated recalibration controlled by the HPC scheduler is essential for maintaining performance. These findings establish a clear roadmap for other HPC centres aiming to incorporate quantum computing resources, paving the way for a new era of hybrid quantum-classical computation.

This work marks a pivotal step in advancing computational capabilities for scientific research through the incorporation of quantum computers into existing HPC environments, commonly referred to as HPC+QC integration. This practical implementation revealed crucial insights into HPC and quantum computer co-location, demonstrating that incorporating quantum computers into established high-performance computing environments is feasible, though requires careful planning. Facility requirements for quantum computers proved stricter than those for classical systems, yet deployment within an HPC environment was feasible following a rigorous site survey ensuring compliance with environmental controls. Specifically, the electronics cabinet maintained a temperature stability of ±1°C within 12 hours around any set point between 20 and 25°C, with humidity maintained between 0.

25 and 60%, non-condensing. The research team discovered that quantum computers necessitate regular recalibration due to their dynamic nature, a process automatically controlled by the HPC scheduler. To facilitate seamless integration, scientists developed the Munich Quantum Software Stack (MQSS), a custom software architecture supporting both remote API-based access and tightly-coupled in-HPC execution. This stack utilizes a Multi-Level Intermediate Representation (MLIR)-based compiler, enabling homogeneous compilation across diverse quantum targets. The MQSS client automatically detects job origin, routing it to either a REST API for asynchronous access or an HPC client for local accelerator-style submission, without requiring code modifications from the user. The system supports multiple quantum programming frameworks, including CUDAQ, Qiskit, and Pennylane, through modular adapters. A Quantum Device Management Interface (QDMI) delivers hardware-specific performance data to the compiler, enabling just-in-time adaptation of compilation and scheduling strategies, and demonstrated the potential for reducing noise through just-in-time quantum circuit transpilation.

Quantum Integration, HPC Scheduler, and Recalibration

Researchers identified key lessons from this initial implementation, demonstrating that incorporating quantum computers into established high-performance computing environments is feasible, though requires careful planning. The team highlights the necessity of thorough site surveys to ensure quantum systems meet stringent facility requirements, differing from those of classical computers. Furthermore, the study demonstrates that quantum computers necessitate ongoing, automated recalibration managed through the HPC scheduler, reflecting their dynamic nature. Reliable operation also depends on redundant power and cooling infrastructure, essential for maintaining the stability of quantum processors.

The researchers emphasize the importance of comprehensive onboarding programs, designed to support both quantum computing specialists and new users transitioning to this emerging technology. The authors acknowledge that this integration represents an early case study, and further work is needed to optimise workflows and explore the full potential of hybrid quantum-classical computing. They suggest that the lessons learned from this implementation will serve as a valuable roadmap for other HPC centres considering the integration of quantum resources, paving the way for more complex scientific applications.