The latest version of Zuchongzhi includes 105 transmon qubits — devices made from metals such as tantalum, niobium, and aluminum that have increased resistance to noise. (Image credit: D. Gao et al. [2])
Researchers in China have developed a quantum processing unit (QPU) that is 1 quadrillion (10¹⁵) times more efficient than the best supercomputers on Earth.
Scientists at the University of Science and Technology of China (USTC) in Hefei said a new prototype 105-qubit chip called “Zuchongzhi 3.0” that uses superconducting qubits represents a significant advance in quantum computing.
It rivals the benchmark results achieved by Google's newest quantum processor, Willow, in December 2024, which allowed scientists to claim quantum supremacy – a state where quantum computers outperform the fastest supercomputers in lab tests.
Scientists used the processor to solve a problem on the widely used random-sampling circuit (RSC) quantum computing benchmark in just a few hundred seconds, according to a new study published March 3 in the journal Physical Review Letters.
The test, which involved sampling a random circuit with 83 qubits and 32 layers, was completed 1 million times faster than the result set by Google's previous-generation Sycamore chip, published in October 2024. By comparison, the world's second-fastest supercomputer, Frontier, would only take 5.9 billion years to complete the same task.
While the results suggest that QPUs can achieve quantum supremacy, the specific RCS benchmark used favors quantum methods. Additionally, improvements in the classical algorithms that drive traditional computing could reduce that advantage, as happened in 2019 when Google scientists first announced that a quantum computer had outperformed a classical one — in the first use of the RSC benchmark.
“Our work not only expands the horizons of quantum computing, but also creates the foundation for a new era in which quantum processors will play a key role in solving complex real-world problems,” the scientists noted in their study.
Competing with Google's Leading Quantum Processor
The latest version of Zuchongzhi contains 105 transmon qubits — devices made from metals such as tantalum, niobium and aluminum that are less sensitive to noise — in a 15-by-7 rectangular lattice. This builds on the previous chip, which contained 66 qubits.
One of the most critical areas determining the viability of quantum computing in real-world settings is coherence time, which measures how long a qubit can maintain its superposition and use the laws of quantum mechanics to perform calculations in parallel. Longer coherence times allow for more complex operations and calculations.
Another major improvement concerns gate accuracy and quantum error correction, which has previously been a barrier to building useful quantum computers. Gate accuracy measures how accurately a quantum gate performs its intended operation, where a quantum gate is analogous to a classical logic gate, performing a specific operation on one or more qubits and manipulating their quantum state. Higher qubit accuracy means fewer errors and more reliable computations.
Zuchongzhi 3.0 demonstrated an impressive 99.90% parallel single-qubit gate accuracy and 99.62% parallel dual-qubit gate accuracy. Google's Willow QPU slightly outperformed it, achieving 99.97% and 99.86%, respectively.
These advances were made possible in large part by engineering improvements, including improved manufacturing methods and more optimized qubit designs, the scientists reported in their study. For example, the latest version lithographically defines qubit components using tantalum and aluminum bonded together using a process
Sourse: www.livescience.com
