NVIDIA CUDA-Q Logical Targets Faster Quantum Development

NVIDIA Expands Quantum Tools as Fermilab Reports 7x Development Speedup

CUDA-Q Logical helps researchers compare designs for future quantum computers, while Sandia introduces a benchmark for tracking practical capabilities.

NVIDIA has expanded its open-source CUDA-Q platform with CUDA-Q Logical, a software layer designed to help researchers develop applications for fault-tolerant quantum computers. Announced on September 14, the update arrives with a reported result from Fermilab: work that would typically require around five months of specialized infrastructure development was completed in three weeks. NVIDIA announcement

The software addresses a challenge in quantum computing: designing an application also means accounting for the hardware and error correction needed to run it. Changing an algorithm, processor architecture, or error-correction approach can substantially alter the number of qubits and the time a calculation requires.

CUDA-Q Logical gives researchers a programmable way to compare those choices and check their results. Its focus is on systems using logical qubits, which use error correction to protect quantum information from the errors affecting physical qubits. NVIDIA identifies drug discovery, financial modeling, and materials development as potential applications.

At Fermilab, researchers used the software to validate earlier findings and compare qubit requirements, runtimes, and other resources across different system designs. NVIDIA describes the shorter development period as a 7x speedup. That figure measures the research workflow, rather than the execution speed of a quantum computer.

A separate modeling exercise examined how much hardware a future system might need. According to NVIDIA, Iceberg Quantum used CUDA-Q Logical to evaluate its architecture with Diraq qubits. The model indicated that 1,000 logical qubits could be supported by 150,000 physical qubits, roughly one-tenth of Diraq’s previous estimate. This is a projected resource requirement, not a report of an operational machine with that configuration. NVIDIA announcement

The announcement also introduces QUOPS, a benchmark developed by Sandia National Laboratories to assess progress toward quantum computers capable of useful applications. Its approach is intended to work across hardware platforms and help researchers track computational capabilities beyond physical qubit counts alone.

Sandia has shared early QUOPS results for processors from Google, IBM, and Quantinuum in a preprint released ahead of IEEE Quantum Week. NVIDIA says a reference implementation is available through CUDA-Q. NVIDIA announcement

The update accompanies broader efforts to connect quantum processors with GPU supercomputers. NVIDIA says Anyon Computing developed a quantum control system using its NVQLink architecture, while Quandela developed a QPU-GPU architecture. Quantum Machines also demonstrated integration of supercomputing resources with qubits at the Israeli Quantum Computing Center.

NVIDIA says CUDA-Q Logical is available through GitHub, with adoption already including Fermilab, Sandia, Infleqtion, and IQM Quantum Computers. The release expands the software researchers can use to evaluate future systems; it does not announce commercial availability of the fault-tolerant machines being modeled. CUDA-Q repository