Quantum computers are getting a place in the supercomputer queue
Slurm’s next release aims to make shared quantum hardware easier to schedule. Early integrations show how jobs can wait for a device instead of starting while it is unavailable.
Reading tools for text size, copying, printing and your reading list

NVIDIA announced on November 16 that Slurm, software used to allocate work on supercomputers, is expanding its support for quantum processors. The practical goal is simple: let researchers request quantum and conventional computing through the same familiar job system, rather than a separate arrangement for each machine. Slurm 26.11 is due later in November.[1]
A scientific application may prepare data on conventional processors, send a selected task to a quantum device, then analyze the measurements. Those stages need coordination. A quantum processor can be shared by several users and may be unavailable even while the conventional machines are running.[2]
The new integration makes device availability part of the scheduling decision. If a quantum service is unavailable, work can remain queued without tying up the computing nodes that would otherwise start it. Munich Quantum Software Company says its Orchestrator has tested waiting through outages, resuming after recovery and enforcing access limits using Slurm’s release candidate.[2]
IQM, whose quantum computers already run alongside classical systems at centers including Germany’s Leibniz Supercomputing Centre, says its engineers are evaluating the new capability and plan validation as centers upgrade. That is an early integration programme, not a claim that every IQM installation already runs the new release.[3]
Pasqal and IBM have built another connection using their shared quantum-resource interface. Pasqal reports tests with an emulator and simulated local deployment, plus a run on its physical FRESNEL_CAN1 processor in Canada. It also identifies a useful distinction: CINECA’s existing production access to the Sol quantum computer uses an older plugin, not this new scheduling path.[4]
The advance is in operating shared equipment, rather than making its intrinsically better. For computing centers, the useful next milestone is a dependable upgrade under everyday workloads: keeping access rules, device status and job cleanup together while researchers continue using their usual submission tools. QubitWire’s earlier openQSE report examines a separate integration effort.[2][4]