Google’s Sycamore quantum processor hit the headlines in September when a leaked draft paper suggested that the device is the first to have achieved quantum supremacy by solving a problem more than a billion times faster than a conventional (classical) supercomputer. But now, researchers at IBM claim that the Google team has vastly overestimated how long the problem would take to solve on a supercomputer – saying that the actual speed-up offered by Sycamore is more like a factor of 1000.
The final version of the Google paper has since been published in Nature and describes how 53 programmable superconducting quantum bits were used to determine the output of a randomly-chosen quantum circuit. While the calculation has no practical application, it does provide a way of comparing the performance of Sycamore to that of classical computer.
The quantum calculation took 200 s to complete and the Google researchers estimated that the same calculation would take 10,000 years to execute on a state-of-the-art supercomputer.
Days, not millennia
But now, IBM’s Edwin Pednault, John Gunnels, and Jay Gambetta argue that the same task could be performed in just 2.5 days using the Summit supercomputer at Oak Ridge National Laboratory in the US. Writing in the IBM Research Blog, they say that the Summit calculation would be more accurate than the Sycamore calculation and with further optimization, the execution time could be reduced.
The IBM trio say that the Google researchers overestimated the classical calculation time because they failed to fully consider the “plentiful” hard-disk data storage resources available on supercomputer systems.
They write, “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 trio also cautions physicists about the casual use of the term “quantum supremacy”, which is usually understood to describe the ability (at least in principle) of a quantum computer to do a calculation that would take a classical computer an impractically long time to complete.
Google reports quantum supremacy in draft paper
They write, “we urge the [quantum computing] community to treat claims that, for the first time, a quantum computer did something that a classical computer cannot with a large dose of scepticism due to the complicated nature of benchmarking an appropriate metric”.
The term quantum supremacy is attributed to Caltech’s John Preskill, who has recently lamented his choice of words. As well as the unfortunate association with the term white supremacy, Preskill is also concerned about the hype surrounding the word supremacy.
IBM researchers provide a technical description of how the Summit supercomputer calculation could be done in a preprint on arXiv.
- The nascent quantum software industry is profiled by Jon Cartwright in “The hard sell of quantum software“.