NVIDIA extends CUDA-Q for fault-tolerant quantum systems

NVIDIA added a new CUDA-Q Logical layer to its open CUDA-Q platform for developing and testing fault-tolerant quantum applications on logical qubits. Fermilab cut its architecture development cycle from five months to three weeks, and Iceberg Quantum showed that 1000 logical qubits can be implemented on ~150,000 physical qubits—10 times fewer than previous Diraq estimates.

• CUDA-Q Logical allows modeling algorithms, error correction codes, and QPU components
• Fermilab sped up fault-tolerant architecture development 7x—from 5 months to 3 weeks
• Iceberg Quantum: 1000 logical qubits on ~150,000 physical, 10x fewer than Diraq estimates
• Added support for the QUOPS benchmark from Sandia National Laboratories