NVIDIA said on September 14 that it expanded its open-source CUDA-Q quantum software with CUDA-Q Logical, an orchestration layer for developing fault-tolerant quantum computing applications by modeling algorithms, error correction, and hardware together.

It targets fault-tolerant quantum computing, where calculations can continue accurately despite noisy components. Logical qubits are built from many physical qubits to make operations more reliable.

Such modeling matters because quantum error correction can require large overheads in qubits and runtime, so researchers need resource estimates before large-scale hardware is available.

Nvidia said Fermi National Accelerator Laboratory, or Fermilab, used CUDA-Q Logical to validate earlier results and assess qubit and runtime requirements across different approaches, cutting one workflow from five months to three weeks.

The company also said Sandia National Laboratories developed QUOPS, a cross-platform benchmark for tracking progress toward fault-tolerant quantum computers, and that a QUOPS reference implementation is available in CUDA-Q. CUDA-Q Logical is now available on GitHub.

Similar tools exist elsewhere. IBM’s Qiskit ecosystem includes tools for postselected quantum error correction.

🔗 Source: Nvidia

Recent NVIDIA developments