NVIDIA Expands CUDA-Q Platform, Introduces Quantum Fault-Tolerant Computing Orchestration Layer

BlockBeats news, September 14: Nvidia today announced the expansion of its open-source quantum computing platform CUDA-Q, launching the CUDA-Q Logical orchestration layer to provide a programmable and verifiable development approach for fault-tolerant quantum computing applications, supporting researchers in co-designing algorithms, error correction codes, quantum hardware, and other components, and rapidly testing different system configurations.

Nvidia stated that CUDA-Q Logical has been adopted by institutions including Fermi National Laboratory, Infleqtion, IQM Quantum Computers, QCDesign Quantum Motion, and Sandia National Laboratories. Among them, Fermi National Laboratory used the platform to shorten the development cycle of fault-tolerant quantum algorithms from about 5 months to 3 weeks, an approximately 7-fold speed increase. Iceberg Quantum used CUDA-Q Logical to simulate the Diraq qubit architecture, believing that the number of physical qubits required to achieve 1,000 logical qubits could be reduced to about 150,000, roughly a 10-fold decrease from previous estimates.

In addition, the quantum computing benchmark QUOPS launched by Sandia National Laboratories has also joined CUDA-Q to measure progress across platforms in the development of quantum hardware toward practical, fault-tolerant quantum computing. Nvidia is also continuing to advance quantum-GPU supercomputing infrastructure such as NVQLink, promoting close collaboration between quantum processors and GPU supercomputing systems.