Brief overview
- Researchers at the Shanghai Artificial Intelligence Lab in China have demonstrated that their classical computer system can perform complex calculations faster and more energy-efficiently than Google’s Sycamore machine.
- The team used their classic apparatus to perform a task similar to that made famous by Google’s quantum supremacy experiment in 2019.
- Experts suggest that the results of the experiment show both the need to rigorously evaluate quantum advantages and to continuously promote innovation in classical and quantum computing.
A classical computer has outperformed a quantum computer at a task once thought to be the preserve of a quantum machine. Researchers at the Shanghai Artificial Intelligence Lab in China have demonstrated that their classical computing system, powered by more than 2,300 Nvidia A100 GPUs, can perform a complex calculation faster and more energy-efficiently than Google’s Sycamore quantum processor.
While it may seem like a battle royale between classical computers, supercomputers, and quantum computers, scientists say this is exactly the kind of innovation needed to fuel competition and drive further innovation across all computing approaches.
A confrontation, not a slowdown
Some background — In 2019, Google’s Sycamore quantum computer made headlines when it achieved what was called quantum supremacy, performing in 200 seconds a calculation that would have taken the world’s fastest supercomputer at the time 10,000 years. The achievement was hailed as a historic moment that demonstrated that quantum computers could solve problems that were beyond the reach of classical systems.
However, the idea of quantum supremacy can mistakenly give the impression that innovations in classical computer hardware design, algorithms, and other techniques will remain static. In fact, rapid advances in algorithms and classical computer hardware have since narrowed the gap.
In a recent study, published on the arXiv preprint serverResearchers led by Rong Fu have achieved impressive results using classical computing power. Their system completed a computational task—simulating and sampling random quantum circuits (RQC), similar to Google’s supremacy experiment—in just 14.22 seconds, consuming 2.39 kilowatt-hours (kWh) of energy.
In another configuration, they managed to perform the same task in 17.18 seconds while consuming only 0.29 kWh.
The hardware behind the success
According to the paper, the classical computing system used over 2,300 Nvidia A100 GPUs, some of the most advanced classical computing chips available, interconnected to create a massively parallel processing system. Each A100 GPU, equipped with 80 GB of memory and a peak FP16 Tensor Core performance of 312 teraflops (TFLOPS), played a crucial role in handling the computations. The GPUs on each node were interconnected via NVLink, providing a one-way speed of 300 GB/s, while the nodes were connected via InfiniBand with a one-way speed of 100 GB/s.
This setup allowed the researchers to overcome the memory constraints and computational bottlenecks typically associated with large-scale tensor network simulations. By implementing a three-tier parallel scheme and a hybrid communication approach, they significantly reduced the overhead of handling large tensors, achieving unprecedented scalability and efficiency. For clarity, a three-tier parallel scheme involves each node processing a portion of the task independently (node level), multiple CPUs or GPUs within each node working together (intra-node level), and different nodes coordinating to combine their results (inter-node level).
The results obtained by the Shanghai team challenge Google’s initial claim of quantum supremacy. Google’s Sycamore completed its task in 600 seconds with an energy consumption of 4.3 kWh. In contrast, the classical system not only completed the task much faster but also consumed less energy, highlighting the potential of classical computers to keep up with quantum systems, or even surpass them under certain conditions.
An article in New scientist provides additional context: Claims of quantum supremacy are often “overblown,” and the true value of quantum computers will come from practical applications that serve users, regardless of whether similar tasks can be performed on classical machines, according to Christopher Monroe of the University of Maryland.
“It’s not surprising that there are esoteric cases where a programmable quantum system can solve some problems that we can’t solve with classical computers,” Monroe tells New Scientist. “But for quantum computing to be effective, you just need to have a user base that is asking to use quantum computers for certain applications – even if it’s possible to do it with classical computers.”
Josh Nunn of quantum computing startup Orca Computing told the magazine: “The 2019 paper is a supremacy result based on classical methods known at the time.”
He stressed that the arms race between quantum and classical machines was valuable for driving innovation. However, he also stressed that each claim to supremacy and each subsequent counter-claim should be closely examined to ensure that the capabilities of the technology justify the investment.
Future consequences
The success of classical computing in this case does not diminish the potential of quantum computing. As quantum technology continues to develop, it is expected to improve exponentially faster than classical systems. Google’s principal scientist Sergio Boixo noted in 2022 that while classical algorithms have improved, quantum circuits are expected to maintain their advantage as quantum technology advances.
As the scientists point out, the Shanghai team’s work highlights the need to rigorously evaluate quantum advantages and continually push the boundaries of both classical and quantum computing.
However, their techniques, particularly in the treatment of large-scale tensor networks, have broader applications than the specific task at hand. In fact, these methods could be extended to a variety of fields, including—somewhat ironically in the scientific sense—quantum computer simulation, condensed matter physics, and combinatorial optimization, potentially making it possible to solve complex real-world problems more efficiently.