NVIDIA and Partners Integrate Quantum Processors into Slurm Scheduler
NVIDIA, IBM, and IQM are integrating quantum processing units into the Slurm workload manager to treat quantum hardware as schedulable supercomputing resources.
At the SC26 conference in Chicago, NVIDIA announced a major initiative to integrate quantum processing units (QPUs) as schedulable resources within the Slurm workload manager. The update in Slurm 26.11 introduces plugin support for Hierarchical Resources (HRES), allowing supercomputing centers to manage QPUs as shared, cluster-level resources similar to CPUs and GPUs.
To implement this, NVIDIA collaborated with IBM and Pasqal to develop the Quantum Resource Management Interface (QRMI), a vendor-agnostic open-source layer. Simultaneously, IQM Quantum Computers joined the initiative, with its partner Munich Quantum Software Company (MQSC) developing the open-source MQSC Orchestrator to schedule systems like the IQM Emerald through the Slurm HRES and Quantum Device Management Interface (QDMI).
In a parallel effort to advance scientific discovery, IBM, NVIDIA, and Fermi National Accelerator Laboratory (Fermilab) are launching a modular co-design platform at Fermilab's Superconducting Quantum Materials and Systems (SQMS) Center. This platform co-locates NVIDIA DGX B200 systems with IBM quantum computers, including the Nighthawk processor, to solve complex problems in high-energy and nuclear physics.
These hybrid capabilities were further demonstrated by IBM, NVIDIA, and RIKEN, who used quantum-classical workflows to simulate an iron-sulfur cluster. The team achieved a 79-fold improvement in accuracy over previous experiments, approaching the precision of leading classical simulation methods.