Enabling Hybrid HPCQC Workflows with a Heterogeneous Software Stack

Enabling Hybrid HPCQC Workflows with a Heterogeneous Software Stack demonstrates the integration of quantum processors into production HPC systems using a two-level scheduling model. This approach treats Quantum Processing Units (QPUs) as scheduler-managed accelerators via SLURM Generic RESources (GRES), while the Munich Quantum Software Stack (MQSS) handles just-in-time compilation and circuit dispatch.

The system integrates three key components: The SuperMUC-NG petascale supercomputer at the Leibniz Supercomputing Centre (LRZ). A 20-qubit superconducting quantum processor (Q-Exa) provided by IQM Quantum Computers. * The Munich Quantum Software Stack (MQSS) from Munich Quantum Valley (MQV).

Experimental results show that hybrid HPCQC workflows execute with no significant latency overhead compared to conventional classical workloads, establishing a portable integration model for future HPE Cray platforms.

This work presents an end-to-end demonstration of hybrid High-Performance Computing–Quantum Computing (HPCQC) workflows on a production petascale system. The demonstration integrates the SuperMUC-NG supercomputer at the Leibniz Supercomputing Centre, a 20-qubit superconducting quantum processor from IQM Quantum Computers, and the Munich Quantum Valley’s Munich Quantum Software Stack. Rather than treating the quantum processor as an isolated experimental device, the study frames it as a compute resource that can be embedded within conventional HPC workflows, with classical processing, orchestration, and post-processing handled on the supercomputer.

The key contribution is systems-level: a heterogeneous software stack that enables practical coupling between classical HPC infrastructure and a remote quantum processor. This includes abstractions for workflow execution, resource scheduling, communication between the classical and quantum sides, and management of the data and control flow needed to run hybrid jobs. The insight is that the main barrier to HPCQC is not only quantum hardware capability, but also the software and operational integration required to make quantum devices usable within existing production HPC environments. By demonstrating this on a real petascale system, the work provides evidence that hybrid workflows can be made reproducible, schedulable, and operationally manageable.

This matters because it moves HPCQC beyond proof-of-concept demonstrations toward a reusable infrastructure pattern. As near-term quantum processors remain small, noisy, and often accessed remotely, the ability to integrate them cleanly into classical HPC pipelines is essential for realistic applications. The work offers a reference point for building hybrid software stacks, informs design choices for future quantum-aware HPC systems, and supports the longer-term goal of quantum-assisted scientific and engineering workflows that can scale as both classical and quantum hardware improve.