Insider Brief
- Chinese researchers write in a paper on arXiv that their Zuchongzhi 3.0 quantum computer performed a computational task that would take the Frontier supercomputer over 6.4 billion years to complete.
- The Zuchongzhi 3.0 system features 105 qubits with high fidelities and achieved one million samples from an 83-qubit random circuit in mere seconds, surpassing prior benchmarks set by Google’s Sycamore processor.
- While Zuchongzhi demonstrates quantum advantage through scale and speed, comparisons with Google’s Willow processor emphasize diverging research focuses, with Willow advancing fault tolerance critical for practical quantum applications.
In what seems, at the least, to show how global quantum computing research is intensifying, Chinese researchers have reported that their Zuchongzhi 3.0 superconducting quantum computer has performed a computational task that would take the world’s fastest supercomputer, Frontier, over 6.4 billion years to replicate.
This achievement not only advances quantum computational capability in general, but the team behind Zuchongzhi 3.0 tells the South Chinese Morning Press that their latest experiment is keeping China on par with the U.S. in quantum computing research.
Led by the father of China’s quantum computing program, Jian-Wei Pan, the team writes that Zuchongzhi 3.0, which features 105 qubits with high operational fidelity, achieved a significant milestone by performing random circuit sampling with 83 qubits over 32 cycles. The researchers completed one million samples within a few hundred seconds, representing an improvement over previous records set by Google’s Sycamore processor. The Zuchongzhi 3.0’s performance represents a six-order-of-magnitude leap in simulation complexity compared to Google’s recent experiments, the researchers added.
The findings, published on the preprint server arXiv, emphasize the processor’s capabilities in tackling problems that classical computers cannot feasibly solve.
“This leap in processing power places the classical simulation cost six orders of magnitude beyond Google’s SYC-67 and SYC-70 experiments,” the researchers wrote, referring to Google’s Sycamore advances.
While it’s being portrayed as a horse race between Willow and Zuchongzhi, the teams’ experiments seem to be measuring different things, and perhaps expressing diverging priorities in how to advance quantum computing so that it can solve real-world solutions. Zuchongzhi 3.0 focuses on scale, for instance, relying on 83 qubits for random circuit sampling, achieving fidelities above 99% and demonstrating computational tasks six orders of magnitude harder to simulate than Google’s earlier Sycamore experiments. On the other hand, Willow emphasizes fault-tolerance, with surface code logical qubits reaching error rates below 0.15% per cycle and surpassing physical qubit performance.
While Zuchongzhi establishes quantum advantage through sheer computational power, the Willow team’s advance focuses on scalable, error-corrected quantum systems needed for practical applications. The comparison highlights the dual paths in quantum progress: pushing hardware limits versus refining error correction for future-ready quantum algorithms.
The Chinese team hints at the need for error correction in the SCMP: “Google’s Willow achieved major advances in quantum error correction, a crucial step towards making quantum machines more reliable. The Chinese scientists have announced plans to incorporate similar techniques for Zuchongzhi 3.0 in the coming months.”
Implications for Quantum Computing
Still, the work represents serious implications for quantum computing. First, the researchers write that by pushing the boundaries of quantum computational advantage — a term coined to describe when quantum computers outperform their classical counterparts — the Zuchongzhi 3.0 is aimed at future practical applications in optimization, machine learning and drug discovery.
The researchers write that this progress “lays the groundwork for a new era where quantum processors play an essential role in tackling sophisticated real-world challenges.”
While most current quantum tasks are designed to highlight computational supremacy rather than solve practical problems, the researchers argue that scaling up both qubit count and circuit complexity could lead to innovations in fields requiring immense computational power.
Methods and Technology
As mentioned, the Zuchongzhi 3.0’s performance stems from a combination of increased qubit count and improved qubit fidelity. The processor houses 105 superconducting qubits, configured in a two-dimensional rectangular lattice. For the experiments, the team used 83 qubits, optimized for performance and error rates.
Operational fidelities for single-qubit gates, two-qubit gates, and readout operations are 99.90%, 99.62%, and 99.18%, respectively. These figures represent significant improvements over the Zuchongzhi 2.0, which the scientists attribute mainly to refined circuit designs, noise reduction techniques and improved fabrication methods.
One key innovation was the use of a “flip-chip” technique to integrate the processor’s components. The “flip-chip” technique is a method of connecting two microchips by flipping one over and bonding their contact points directly to create high-density, efficient integration with minimal signal loss.
Zuchongzhi 3.0 features a sapphire-based architecture, combined with advanced attenuator configurations and tantalum-aluminum components, minimized noise and extended the qubit coherence time. This increased the relaxation time to 72 microseconds and the dephasing time to 58 microseconds — critical for maintaining the processor’s stability during operations.
Experimental Design
To showcase its computational advantage, the researchers undertook a task known as random circuit sampling. This task, which has become a staple for scientists exploring computational advantage, involves executing a sequence of random quantum gates and measuring the output states — a process computationally infeasible for classical computers to simulate at scale.
Zuchongzhi 3.0 executed circuits with 83 qubits over 32 cycles, generating one million samples in mere seconds. For comparison, the researchers point out that Google’s Sycamore processor completed a similar task with 67 qubits and 32 cycles. Frontier, the most powerful supercomputer, would require billions of years to replicate the Zuchongzhi 3.0’s results, even under ideal conditions, the team writes.
As a comparison, and at the risk of adding more hay to the horse race, both Zuchongzhi and Willow systems rely on classical supercomputer benchmarks to highlight the vast computational advances made by their systems. The Chinese team writes that Zuchongzhi 3.0 processor’s random circuit sampling experiment would take an estimated 6.4 billion years to simulate on the Frontier supercomputer. According to the paper in Nature, Google’s Willow processor demonstrates error suppression and logical qubit stability the logical qubits demonstrated on the Willow processor achieve error rates so low that classical devices would require computational times stretching into septillions of years under current supercomputing capabilities.
Limitations and Challenges
Despite its achievements, the study notes several limitations. While the processor demonstrates quantum computational advantage, the tasks remain narrowly focused on proving this advantage rather than solving practical problems. This is a similar limitation in the Willow experiments, it should be added. Expanding the application of quantum processors to broader, real-world challenges will require further advances in error correction and scalability.
Verifying the results of such experiments is another challenge. The researchers used patch circuits — smaller, simplified versions of the full circuit — to estimate fidelity. While the patch results closely matched expectations, fully verifying large-scale quantum computations against classical counterparts is increasingly infeasible as system size grows.
Optimism on Quantum’s Future
The researchers write they are optimistic about the future of quantum computing, emphasizing the need to continue increasing qubit counts and circuit complexity. They propose further work on error correction techniques and the integration of quantum processors into hybrid quantum-classical systems for practical applications.
Scaling quantum systems will also require addressing engineering challenges, such as improving qubit coherence times and minimizing operational errors. Additionally, the development of software and algorithms tailored to quantum architectures will be essential for unlocking their full potential.
While the technology remains in its early stages, its rapid progress offers a glimpse into the potential for quantum processors to revolutionize industries dependent on computationally intensive tasks.
As the researchers conclude: “Our work advances the discourse on quantum computing by providing empirical evidence of the technology’s potential to revolutionize computational tasks. It serves as both a testament to the progress in quantum hardware and a foundation for practical applications. Scaling up in qubits and circuit complexity enhances our capacity to address sophisticated challenges in optimization.”
The researchers note that the paper was published on arXiv, a server that allows researchers to quickly gain informal peer review. They do plan to submit the paper for that official peer review.
For a deeper technical dive that this article can only summarize, please read the paper.
Institutions and companies involved in the study include: the Hefei National Research Center for Physical Sciences at the Microscale, University of Science and Technology of China (USTC); the School of Physical Sciences, USTC; the Shanghai Research Center for Quantum Science; the CAS Center for Excellence in Quantum Information and Quantum Physics, USTC; and the Hefei National Laboratory, USTC. Additional contributors include QuantumCTek Co., Ltd. in Hefei; the Henan Key Laboratory of Quantum Information and Cryptography; the Jinan Institute of Quantum Technology and Hefei National Laboratory Jinan Branch; the National Institute of Metrology in Beijing; the School of Microelectronics, Xidian University in Xi’an; and the CAS Key Laboratory for Theoretical Physics, Institute of Theoretical Physics, Chinese Academy of Sciences, in Beijing.
