Revisiting Quantum Supremacy: Simulating Sycamore-Class Circuits Using HPC (arxiv.org)

Researchers have developed a new framework for simulating quantum circuits, specifically targeting Google’s Sycamore-class circuits, utilizing high-performance computing (HPC) systems that combine CPU and GPU capabilities. By employing an NVIDIA A100 GPU for quantum state construction along with N parallel CPU jobs for measurement sampling, they achieved a remarkable linear cross-entropy benchmarking (XEB) score of 0.549 for a 53-qubit, 14-cycle Sycamore circuit, significantly surpassing Google's published score of 0.002. This process demonstrates a substantial acceleration in execution time, with the more complex 53-qubit, 20-cycle circuit simulation completing a massive workload in just over an hour—indicating a staggering speedup of 6.95 x 10^7 times compared to Google's classical estimates. The implications of these findings suggest that the concept of "quantum supremacy" is not a fixed milestone but rather a continually evolving benchmark. The successful use of hybrid classical-quantum strategies demonstrates that practical quantum computing capabilities can be achieved sooner than previously anticipated, making advanced simulations and applications more accessible. This research underscores the potential for HPC to enhance our understanding and application of quantum algorithms, paving the way for broader utility in the field of AI and machine learning.