THE CRUNCH

NVIDIA has expanded Slurm, the job scheduler many supercomputing centres use to manage their workloads, so it can handle quantum processors alongside classical hardware. The new release builds on earlier work with IBM, Pasqal and the wider quantum community, and lets QPU builders expose their hardware to every supercomputing centre through a single interface defined in a simple file.

Until now, plugging a quantum processing unit into a supercomputing centre meant solving a fresh software integration problem for every new QPU or every new centre. By supporting the widely used QRMI and QDMI interfaces, the update lets hardware builders reuse their integrations across deployments, while centres can add quantum processors to their existing scheduling workflows without bespoke engineering.

Adoption is already underway, with Alice & Bob, IBM, IQM, Pasqal, Quantinuum and MQSC signed on. The timing reflects how quickly quantum hardware is spreading: deployments at supercomputing centres have grown nearly sevenfold over the past five years, and nearly 30 centres worldwide are in the process of deploying more than 50 quantum processors.

The change also fits into NVIDIA's broader quantum-GPU platform announced at SC26. Scientists can run hybrid quantum-classical applications built with CUDA-Q, with Slurm coordinating access to all the processing they need. NVIDIA says Fermilab has already used CUDA-Q to complete studies in three weeks that would normally have taken five months.

WHAT HAPPENS NEXT

Slurm 26.11 is due later this month, so centres and QPU builders can start using the new interface then.