November 19, 2024 – NVIDIA and Google Quantum AI have announced a collaboration that will utilize the NVIDIA CUDA-Q platform to empower Google researchers in developing large-scale digital models of quantum computers, aimed at addressing design challenges.
According to the joint statement, Google’s Quantum AI division will leverage NVIDIA’s Eos supercomputer and its hybrid quantum-classical computing platform to simulate the physical characteristics of their quantum processors. This collaboration is expected to help Google overcome the hardware limitations of quantum computing and accelerate the development of quantum components.
It’s worth noting that the “hardware limitations” referred to here are what researchers call “noise,” which causes quantum computers to stop computing after a certain number of quantum operations.
Guifre Vidal, a research scientist at Google Quantum AI, stated, “Only by controlling noise while scaling up quantum hardware can we truly develop quantum computers with commercial applications.”
Understanding noise in quantum hardware design requires a series of complex dynamic simulation processes that can fully capture the interaction between qubits in the quantum processor and their environment.
To achieve this goal, NVIDIA has provided a supercomputer based on 1,024 H100 Tensor Core GPUs, “enabling one of the world’s largest and fastest dynamic simulations of quantum devices at a significantly reduced cost.”
NVIDIA claims that with the help of CUDA-Q and H100 GPUs, Google can conduct comprehensive and realistic dynamic simulations of devices containing 40 qubits. The system will provide results within minutes, compared to the previous week-long wait time, and at a fraction of the previous cost (note: specific data or ratios were not disclosed).
