Chinese scientists challenge Google’s previous quantum supremacy claims with GPUs

Why it matters: In 2019, a paper leaked claiming physicists at Google used their quantum computer, Sycamore, to run a calculation that would overwhelm the world’s most powerful supercomputer. Chinese scientists recently challenged that claim by successfully running the identical computation in a matter of hours using the computing power of today’s GPUs. Their results prove a supercomputer using today’s technology could likely beat Sycamore’s previous record.

Google’s quantum computing researchers originally ran the complex calculation in 200 seconds (just over three minutes), a feat they claimed would have taken the fastest supercomputer upwards of 10,000 years. Based on this result, the team claimed they had reached a significant milestone known as quantum supremacy. Quantum supremacy is the point at which a quantum device can solve problems that otherwise cannot be solved by classical technology in any reasonable amount of time.

Not everyone was a believer in Google’s self-proclaimed supremacy in 2019. Another major player in the quantum computing space, IBM, challenged Google’s claims from the start. Researchers there claimed the same task could be performed in a matter of days with the right amount of available resources, invalidating Google’s claim of quantum supremacy.

Chinese scientists successfully proved IBM’s point by attacking the original problem using advanced algorithms and compute power from today’s GPUs to complete the calculation. According to a report by science.org,the effort used 512 GPUs, a number that is far from unfathomable when considering how many units have been used in cryptocurrency mining operations over the last several years.

Chinese scientists successfully proved IBM’s point by attacking the original problem using advanced algorithms and compute power from today’s GPUs to complete the calculation. A report in Science notes that the effort used 512 GPUs — a number that is far from unfathomable when considering how many units cryptocurrency mining operations sometimes use.

The GPU compute power combined with advanced algorithms completed the same calculation within several hours. The results, which were unthinkable according to the leaked 2019 research findings, provide evidence to back claims that a large enough supercomputer could, in fact, rival Sycamore’s earlier achievement.

Conventional computing relies on bits, the most basic information units in computing. These bits can exist as only one of two values, either a 0 or 1.

Quantum computing relies on quantum bits, or qubits, comprised of a superposition of 0 and 1. Like a bit, a qubit can equal 0 or 1. However, it has the added property of equaling 0 and 1 simultaneously, resulting in vastly increased computing potential.

The achievement does not invalidate Google’s previous quantum achievements, nor does it mean that standard processing hardware can “catch up” to quantum’s capabilities. Google Quantum AI’s principal scientist, Sergio Boixo, said the original 2019 paper acknowledged the likely future improvement to classical algorithms but doesn’t believe today’s classical computing approach can keep pace with quantum technologies.

Boixo’s statement is accurate given the rate of quantum growth since 2019. Google’s original Sycamore was a 53-qubits processor. In 2021 IBM unveiled their 127-qubit Eagle, and their quantum roadmap looks to break the 1,000-qubit barrier sometime in 2023.