A standard door into quantum supercomputing
NVIDIA said it has expanded Slurm, the workload manager widely used at supercomputing centers, to support quantum processing units (QPUs). According to the company, nearly 30 supercomputing centers worldwide are deploying more than 50 quantum processors, and many of them run Slurm.
What changes
Previously, every new QPU and every new center required a separate software integration. With the new release, QPU builders can connect to all centers through a single file using the QRMI and QDMI interfaces. Centers can add quantum processors to their existing scheduling and resource-management workflows.
- Slurm 26.11 will be available later this month.
- Alice & Bob, IBM, IQM, Pasqal, Quantinuum and MQSC have adopted the integration.
- Quantum-classical applications built with CUDA-Q can run through Slurm.
Updates across CUDA-Q
At SC26, NVIDIA also updated layers of its CUDA-Q platform. CUDA-Q QEC gains a new decoder server for researchers working on error correction; the company says the new TCBP-JOBS decoder is 20,000x faster than the industry standard. Fermilab used CUDA-Q to complete studies in three weeks that would normally take five months.
Examples from the ecosystem
Quantum Motion integrated its QUICHE chemistry framework with CUDA-Q. Altera and Oxford Quantum Circuits designed an FPGA-based controller that connects qubit control to GPUs through NVQLink. QunaSys simulated more than 300 million quantum operations on ABCI-Q using 1,024 NVIDIA Hopper GPUs.



