As quantum computing continues to mature, its integration into classical High-Performance Computing (HPC) environments is emerging as a promising strategy to accelerate specific, high-impact workloads.
Rather than replacing classical computation, Quantum Processing Units (QPUs) are expected to operate as specialized co-processors within HPC systems—complementing CPUs and GPUs in heterogeneous computing landscapes.
However, this convergence raises key challenges, particularly in how resources are allocated and orchestrated efficiently across classical and quantum boundaries.
This talk introduces a novel approach that combines process malleability and workflow-driven scheduling to enable adaptive and energy-efficient execution of hybrid HPC-quantum applications.
Leveraging a malleability-aware runtime, our method dynamically releases classical resources when a quantum task is active and reassigns them once quantum execution is complete—improving overall utilization and reducing idle energy overhead.
These decisions are guided by application workflows that delineate classical and quantum execution stages.
We demonstrate our strategy with real-world experiments on a hybrid quantum-classical application, showcasing how dynamic resource management can unlock substantial benefits in performance, resource efficiency, and sustainability.
This work lays the foundation for more intelligent orchestration of quantum-accelerated workloads in HPC clusters and outlines a clear path forward for integrating QPUs into production-level HPC environments.
