Skip to content
Quantum.world

The Signal

NVIDIA says Slurm 26.11 will schedule quantum processors

Slurm 26.11 is set to add support for scheduling quantum processors alongside supercomputing workloads. NVIDIA says the change will let hardware builders reuse integrations across centers and support hybrid quantum-classical jobs.

NVIDIA says Slurm 26.11 will schedule quantum processors

Slurm 26.11 is set to let supercomputing centers schedule quantum processors alongside conventional computing resources, bringing quantum hardware into the workload manager’s existing scheduling and resource-management workflows. NVIDIA described the planned change in a Nov. 16 post, saying it is designed to give quantum processor builders a shared integration route across centers.

A shared interface for quantum hardware

The update builds on work with IBM, Pasqal and the wider quantum computing community. It supports the Quantum Resource Management Interface (QRMI) and Quantum Device Management Interface (QDMI), which NVIDIA’s post identifies as existing quantum-HPC interfaces that can help hardware builders reuse integrations across deployments.

NVIDIA said the expansion is already being adopted by Alice & Bob, IBM, IQM, Pasqal, Quantinuum and MQSC. The release is intended to reduce the need for a separate software integration for each combination of processor and supercomputing center; NVIDIA says centers will be able to add QPUs to established scheduling workflows without building bespoke QPU-specific solutions.

How the integration works

IBM’s account of the work provides more detail on one implementation: IBM and NVIDIA are integrating the open-source QRMI with Slurm’s Hierarchical Resources, or HRES, feature. That work is intended to give Slurm greater visibility into quantum resources and workloads.

MQSC describes a complementary route using QDMI. Its MQSC Orchestrator combines HRES with QDMI, compilation and hybrid execution; the company says the Slurm 26.11 release candidate used in its integration adds plugin support for external services. The details describe the software components behind these integrations, not a claim that every named quantum processor is already operating through the new Slurm release.

Quantum jobs in supercomputing workflows

NVIDIA says quantum processor deployments at supercomputing centers have grown nearly sevenfold over the past five years, with nearly 30 centers around the world in the process of deploying over 50 processors. Many of those centers already use Slurm for scientific workloads, the company said, including work by computational chemists, nuclear physicists and materials scientists.

The company’s stated aim is to let researchers submit quantum-classical applications through familiar HPC workflows, with Slurm coordinating access to the processing resources those jobs need. That includes applications developed with NVIDIA CUDA-Q, which the company describes as a platform for hybrid quantum-classical computing.

Part of a broader quantum-GPU platform

NVIDIA presented the Slurm expansion as an orchestration layer in its quantum-GPU platform, alongside CUDA-Q tools for quantum error correction, application development and quantum hardware control. Those announcements included a new decoder server for researchers prototyping error-correction ideas and expanded fault-tolerant primitives for application builders.

Other work described in the post spans software and large-scale simulation: Quantum Motion integrated its QUICHE chemistry framework with CUDA-Q, while Altera and Oxford Quantum Circuits designed an FPGA-based controller connecting qubit control and readout to GPUs through NVQLink. Separately, QunaSys used cuQuantum and NVIDIA Hopper GPUs on ABCI-Q to simulate quantum operations for a four-hydrogen-atom model.

Slurm 26.11 was scheduled to become available later in November.

Sources

  1. NVIDIA Blog
  2. IBM Quantum
  3. MQSC

Companies in this story

Related stories

Explore

More articles