Riverlane and NVIDIA are collaborating on the use of artificial intelligence and accelerated computing to make quantum error correction calibration faster and more adaptive as quantum systems scale.
Quantum error correction is designed to detect and correct errors caused by quantum computers’ extreme sensitivity to noise. Future systems could require real-time processing of terabytes of information every second, creating significant computational requirements for the classical systems supporting quantum hardware.
A key part of QEC is the decoder, which interprets signals generated by a quantum processor to determine which errors most likely occurred and what corrections should be applied. Decoders depend on numerical settings known as weights, which reflect the error characteristics of the underlying hardware.
Those characteristics can change over time as qubit frequencies shift, defects emerge or disappear, and gate performance changes. As a result, decoder settings can become less accurate unless they are regularly recalibrated. Frequent calibration can consume valuable operating time that otherwise could be used for experiments or computational workloads.
Riverlane has developed an AI-based calibration loop that learns improved decoder settings directly from experimental data. As part of the collaboration, the company is using NVIDIA H100 GPUs to evaluate different decoder settings and determine which configurations produce fewer logical errors without requiring engineers to manually adjust individual parameters.
The companies demonstrated a substantial improvement in calibration speed using Riverlane’s Local Clustering Decoder adapted for high-throughput GPU processing. A surface-code training cycle that previously required approximately three minutes on a single CPU core was completed in 2.8 seconds using an NVIDIA H100 GPU.
The H100 was also able to decode more than four million surface-code syndromes per second, compared with roughly 59,000 per second on a single CPU core. Riverlane said the performance could allow calibration to occur before individual experiments rather than only periodically, while also enabling longer training cycles within the same amount of time.
Faster calibration could reduce idle time for quantum computers and help maintain logical performance as hardware conditions change. Riverlane also sees the workload as relevant to high-performance computing centers and quantum control-system providers already deploying GPUs alongside quantum hardware.
The collaboration will also investigate broader applications of AI and accelerated computing to QEC, including AI-enhanced error correction, different computing architectures for future quantum processors, and ways Riverlane’s technology could operate alongside NVIDIA quantum and accelerated computing platforms.
KEY QUOTES:
“Quantum error correction is the key technology that will enable useful quantum computers. As quantum systems scale, QEC becomes one of the world’s most demanding real-time computing challenges. Through this collaboration with NVIDIA, we are exploring how advances in quantum computing, AI and accelerated computing can come together to support the next generation of quantum systems.”
Steve Brierley, Founder And CEO Of Riverlane
“Accelerated computing is becoming the engine of useful quantum computing, transforming how quantum computers are built and operated. Riverlane is using AI to help error correction keep pace with changing quantum hardware. Accelerating that learning process opens the door to more frequent calibration, so researchers can spend less time tuning their quantum computers and more time doing science.”
Sam Stanwyck, Director Of Quantum Product At NVIDIA
