IBM, Fermilab, and NVIDIA Unveil Modular Quantum Co-Design Platform, Slurm HRES Integration, and GPU QEC Tools at SC26
AI-summarised brief · reviewed before publication
At the SC26 supercomputing conference in Chicago, IBM, NVIDIA and the U.S. Department of Energy’s Fermilab announced a modular quantum‑classical co‑design platform that will co‑locate NVIDIA DGX B200 systems with IBM Quantum Nighthawk processors. Fermilab will manage researcher access and hybrid workflows through its HEPCloud infrastructure, initially emulating inter‑module communication to study latency and routing errors in multi‑QPU runs, with a later phase adding ultracoherent superconducting RF cavities as quantum memories and entanglement buffers. NVIDIA also released GPU‑accelerated quantum error‑correction (QEC) tools, including the TCBP‑JOBS decoder that achieves millisecond‑scale latencies for surface and qLDPC codes, and integrated native quantum resource scheduling into Slurm 26.11’s Hierarchical Resources (HRES) scheduler via QRMI and QDMI interfaces. The joint effort aims to streamline quantum resource allocation across more than 50 processors at roughly 30 supercomputing centers worldwide.
💡 Why It Matters
- · By embedding quantum scheduling directly into Slurm, the consortium removes a major software barrier, enabling seamless, policy‑driven access to quantum hardware alongside traditional CPUs and GPUs.