Abstract
A major milestone in the era of noisy intermediate scale quantum computers is quantum supremacy claimed on the Sycamore quantum processor of 53 qubits, which can perform a random circuit sampling task within 200 seconds while the same task is estimated to require a runtime of 10,000 years on Summit. This record has been renewed with two recent experiments on the Zuchongzhi 2.0 (56 qubits) and Zuchongzhi 2.1 (60 qubits) quantum processors. On the other front of quantum supremacy comparison, there has also been continuous improvements on both the classical simulation algorithm as well as the underlying hardware. And a fair justification of the computational advantages for those quantum supremacy experiments would require to practically simulate the same problems on current top supercomputers, which is still in lack. Here we report the full-scale simulations of these problems on new generation Sunway supercomputer, based on a customized tensor network contraction algorithm. Our benchmark shows that the most challenging sampling task performed on Sycamore can be accomplished within 1 week, thus collapsing the quantum supremacy claim of Sycamore. Additionally, we show that the XEB fidelities of the quantum supremacy circuits with up to 14 cycles can be verified in minutes, which also provides strong consistency check for quantum supremacy experiments. Our results redefine quantum supremacy baseline using the new generation Sunway supercomputer.
Introduction
In 2019, Google’s 53-qubit Sycamore processor performed a Random Circuit Sampling (RCS) task in 200 seconds, leading to a claim of "quantum supremacy" based on an estimated classical runtime of 10,000 years on the Summit supercomputer. However, quantum supremacy is not a static milestone but a moving target defined by the performance gap between quantum hardware and the most advanced classical algorithms. This paper addresses the lack of a fair justification for these claims by performing full-scale simulations on the new generation Sunway supercomputer, demonstrating that optimized classical systems can significantly narrow the gap.
Methodology
The researchers developed a customized Tensor Network Contraction (TNC) algorithm specifically optimized for the Sunway supercomputer's architecture. The approach centers on several key innovations to overcome memory and computational bottlenecks.
Two-Level Parallelization and SWTT
The algorithm employs a two-level parallelization strategy. At the "outer level," the large tensor network is sliced into independent chunks distributed across different CPUs. At the "inner level," the contraction of each slice is parallelized across the 384 cores of the Sunway SW26010P CPU. To address memory bandwidth limitations, the team implemented the Sunway Tensor Transpose (SWTT) algorithm, which fuses tensor permutations and multiplications. This minimizes data movement and pushes the efficiency of skewed tensor contractions toward theoretical hardware limits.
Slicing and Mixed-Precision
To handle circuits beyond the 45-qubit limit of standard Schrödinger-style algorithms, a "slicing" scheme trades computation time for memory space. Furthermore, the use of mixed-precision arithmetic (half and single precision) significantly accelerates the simulation while maintaining result validity.
Key Results
The study evaluated performance using circuits from the Sycamore and Zuchongzhi experiments. The results redefine the baseline for classical simulation of quantum circuits.
- Collapsing the Sycamore Claim: The most challenging task from Google’s 2019 paper—sampling 1 million bitstrings with 0.2% fidelity for the Sycamore-20 circuit—can be completed in just one week on the Sunway system.
- Performance Gains: The implementation achieved a 25x to 100x speedup over standard software packages like
jax. A single "perfect" sample for the Sycamore-20 circuit was generated in only 276 seconds. - Verification: Cross-Entropy Benchmarking (XEB) fidelities for 14-cycle circuits can now be verified in minutes, providing a robust consistency check for quantum experiments.
Discussion and Conclusion
While this work effectively "collapses" the 2019 Sycamore claim, quantum hardware continues to scale. Newer experiments like Zuchongzhi 2.1 (60 qubits) remain extremely challenging, with an estimated classical runtime of 5 years for a single perfect sample.
Future improvements in classical simulation may come from more intelligent tensor contraction ordering and the reuse of intermediate tensors, which could potentially reduce the Sycamore-20 simulation time from a week to approximately 1.5 hours. Additionally, moving this algorithm to architectures with higher memory bandwidth, such as NVIDIA A100 GPUs, could yield further performance boosts. Ultimately, quantum supremacy remains a continuous competition between the best quantum and classical systems.
Summary updated