Multi-Institutional Consortium Expands QRMI to Standardize Quantum-HPC Integration Across Major Workload Managers

A multi-institutional consortium comprising IBM Quantum, Pasqal, Alice & Bob, CINECA, Lawrence Livermore National Laboratory (LLNL), Oak Ridge National Laboratory (ORNL), and the STFC Hartree Centre has detailed the expansion of the open-source Quantum Resource Management Interface (QRMI).

Detailed in a research paper (arXiv:2607.19591), QRMI establishes a vendor-agnostic middleware layer that exposes quantum processing units (QPUs) as first-class schedulable resources alongside classical CPUs and GPUs. Building on IBM’s reference architecture for quantum-centric supercomputing, the updated framework extends QPU co-scheduling beyond Slurm to encompass six primary HPC workload managers and orchestrators: Slurm, PBS Professional, IBM Spectrum LSF, Open Cluster Scheduler (Grid Engine), Kubernetes, and the Flux Framework.

Integrating heterogeneous QPU hardware into classical HPC data centers presents abstraction and scheduling challenges due to vendor-specific APIs, dynamic calibration windows, and queue latency. QRMI decouples application logic and batch schedulers from hardware APIs through a unified middleware layer. The integration enforces a standardized three-step job lifecycle across all supported schedulers: