One week of noise simulation shortened to a few minutes, NVIDIA helps Google accelerate quantum processor design

NVIDIA and Google join forces to accelerate the design pace of Google's quantum computer processors.

On Monday, November 18, Eastern Time, NVIDIA announced a collaboration with Google's quantum computing hardware and software development team, Google Quantum AI, to accelerate the design of Google's next-generation quantum computing devices using simulations powered by NVIDIA's CUDA-Q platform.

Google Quantum AI is using a hybrid quantum-classical computing platform and NVIDIA's Eos supercomputer to simulate the physical characteristics of its quantum processors. This will help overcome the current limitations of quantum computing hardware. Due to what researchers call "noise," quantum computing hardware can currently only perform a limited number of quantum operations before the computation must stop.

A previous report by the domestic Science and Technology Daily referred to noise as the "danger of quantum processors." This is because quantum processors are highly sensitive to noise; even the slightest interference, such as stray photons generated by heat, random signals from surrounding electronic devices, or physical vibrations, can quickly destroy quantum superposition states and severely impact the accuracy of quantum computers. Therefore, whether progress can be made in building practical quantum computers ultimately depends on whether noise can be "tamed."

When announcing the collaboration with NVIDIA on Monday, Guifre Vidal, a research scientist at Google Quantum AI, said:

"Only if we can scale quantum hardware while controlling noise can we develop quantum computers with commercial applications. With NVIDIA's accelerated computing, we are exploring the impact of larger quantum chip designs on noise."

To understand noise in quantum hardware design, complex dynamical simulations are needed to fully capture how qubits in a quantum processor interact with their environment. These simulations typically have a very high computational cost. NVIDIA stated that using its CUDA-Q platform, Google can utilize 1024 H100 Tensor Core GPUs on the NVIDIA Eos supercomputer to perform one of the world's largest and fastest quantum device dynamical simulations at a very low cost.

With NVIDIA's CUDA-Q and H100 GPUs, Google can conduct comprehensive and realistic simulations of devices accommodating 40 qubits. This is the largest-scale simulation performed in its class. The simulation technology provided by CUDA-Q means that noise simulations that originally took a week can now be completed in just a few minutes.

NVIDIA publicly provides software on the CUDA-Q platform to support the aforementioned accelerated dynamical simulations, enabling quantum hardware engineers to quickly scale their system designs.

Tim Costa, NVIDIA's director of quantum and HPC, stated: "AI supercomputing capabilities will help quantum computing succeed. Google's use of the CUDA-Q platform demonstrates its role in advancing quantum computing to help solve real-world problems. GPU-accelerated simulations play a core role."

Regarding the news of Google's quantum computing research and development using NVIDIA's CUDA-Q platform, some netizens commented that Google has succumbed to the power of CUDA-Q. Others remarked that whenever we hear news about the seven tech giants in the U.S. developing their own chips, we always feel that these reports lack thorough research. Manufacturing GPUs like those from companies such as AMD or NVIDIA requires not only money and manpower but also deep expertise