The Slurm workload manager is adding support for quantum processing units, allowing supercomputing centers to schedule quantum and classical resources through one familiar system. The change gives quantum-hardware makers a common interface that can be reused across facilities instead of building a separate integration for every deployment.
The expansion supports the QRMI and QDMI interfaces and is being adopted by Alice & Bob, IBM, IQM, Pasqal, Quantinuum and MQSC. Scientists will be able to run hybrid applications built with platforms such as Nvidia CUDA-Q while Slurm coordinates access to CPUs, GPUs and quantum processors. Slurm 26.11 is due later this month.
Nvidia says nearly 30 supercomputing centers are deploying more than 50 quantum processors, almost seven times the number five years ago. Shared scheduling does not remove the technical limits of today’s quantum hardware, but it can make experimental devices easier to expose to researchers through existing queues and resource policies.
Related CUDA-Q updates add tools for error-correction research, fault-tolerant primitives, reusable quantum algorithms and pulse control. Nvidia says Fermilab used the new workflow to complete studies in three weeks that would otherwise have taken five months, a vendor-reported result that will depend on the specific workload and hardware.