Last month, a leaked paper from Google provided evidence that the 53-qubit Sycamore quantum processor had achieved 'quantum supremacy'. This was after the processor was able to solve a problem in under three minutes and 20 seconds that would have allegedly taken the most powerful classical computers today over 10,000 years to solve.
Google’s experiment is an excellent demonstration of the progress in superconducting-based quantum computing, showing state-of-the-art gate fidelities on a 53-qubit device, but it should not be viewed as proof that quantum computers are “supreme” over classical computers.
The researchers' root for disagreement stems from the fact that they believe that Google's findings failed to account for the repertoire of techniques and resources that classical computers have up their sleeves. These include software assets, a vast knowledge base of algorithms, and a hierarchy of memories. The team argued that any fair comparison between quantum and leading classical computers would be incomplete without considering that repertoire. And that Google's experiment might have not prepared the most powerful classical computer simulation since it missed out on one key aspect:
When their [Google's] comparison to classical was made, they relied on an advanced simulation that leverages parallelism, fast and error-free computation, and large aggregate RAM, but failed to fully account for plentiful disk storage.
When extended primary and secondary memories were incorporated in IBM's simulation, along with other performance-enhancing techniques like circuit partitioning, tensor contraction deferral, cache blocking, among others, the researchers claim that the task that would have allegedly taken 10,000 years to solve previously, can be solved in under 2.5 days with conservative, worst-case estimates:
We argue that an ideal simulation of the same task can be performed on a classical system in 2.5 days and with far greater fidelity. This is in fact a conservative, worst-case estimate, and we expect that with additional refinements the classical cost of the simulation can be further reduced.
The team further claimed that this performance can be increased substantially if optimizations like GPU-direct communications were utilized as well.
IBM's researchers advised the public to take quantum supremacy with a pinch of salt "due to the complicated nature of benchmarking an appropriate metric," as outlined above. Finally, they went on to give their take on the current state of quantum supremacy and how quantum computers might fare against their classical compatriots in the future, saying that the two will work in harmony, and that "quantum computers will never reign “supreme” over classical computers".