The Nighthawk r2 quantum processor, developed by IBM, managed to execute a computational task in just 19 seconds. Researchers estimate that the same operation would require approximately 110 years if performed on a traditional supercomputer.
This experiment was conducted using the company's cloud quantum computing platform. The test employed a methodology known as Random Circuit Sampling (RCS), a technique designed to verify the potential benefits of quantum computers compared to conventional machines. It is important to note that this study is still awaiting peer review and is available on arXiv.
The Nighthawk r2 is equipped with 120 qubits, which function as the fundamental unit of information in quantum systems, analogous to bits in common machines. For this trial, scientists selected 61 of these qubits and executed random circuits with progressive complexity, reaching up to 40 operation cycles.
The most notable performance was observed with 36 cycles. In this specific configuration, the circuit utilized 918 two-qubit gates, which are operations responsible for modifying and combining the states of these quantum units. It was in this scenario that the processor generated one million samples in only 19 seconds.
A relevant aspect of the test lies in its execution. The team chose not to use specially calibrated experimental equipment for the research, opting instead for the standard operational flow of the IBM cloud quantum computing platform. This approach simplifies the process of replicating the experiment, as the researchers made the codes, samples, and circuits used in the analysis available for other users to verify the results.
The comparison with supercomputers was not done by directly running the same program on a classical machine, as this would be unfeasible. Instead, the team applied a technique called tensor network contraction, used to calculate the workload a traditional computer would need to employ to replicate the behavior of the quantum processor.
Details of the Comparison Calculation
The calculation revealed that generating the 1 million samples would require about 1.2 × 10^27 computational operations, an extremely high number of individual calculations. Based on a more cautious estimate of the sustained performance of Frontier, the researchers arrived at a projection of approximately 110 years of processing time.
However, this disparity must be analyzed with caution. The result depends on the method chosen to perform the simulation and does not establish an absolute limit to the capacity of classical computing. The researchers themselves admit that more optimized algorithms could reduce the workload required to reproduce the experiment. Thus, the study's conclusion is not that classical computers will never reach Nighthawk r2, but rather that, in this specific performance test, the quantum processor demonstrated a significant advantage.
Additionally, the demonstration stands out for having occurred on commercially accessible hardware, rather than on apparatus developed solely for laboratory environments. To ensure the work can be audited and repeated, the authors made the necessary details available. Nevertheless, the work still needs to pass through the peer review process, meaning that until then, the finding represents proof of quantum advantage on commercial hardware, but its conclusions remain open to scientific community evaluation.