China's Quantum Computer Juizhang Processes AI Tasks 180 Million Times Faster

Saturday - June 10 2023, 04:04 UTC - 1 year ago

A quantum computer, Juizhang, built by a team led by Pan Jianwei, has claimed to complete AI related tasks 180 million times faster than the fastest supercomputer today. Juizhang uses light as a physical medium for calculation and does not need to work at extremely low temperatures to carry out calculations. The research team said that the computations achieved by Jiuzhang could also help researchers apply the technology in areas such as data mining, biological information, network analysis, and chemical modeling research.

A quantum computer, Juizhang, built by a team led by Pan Jianwei, has claimed that it can process artificial intelligence (AI) related tasks 180 million times faster, the South China Morning Post reported. Jianwei is popularly known as the "father of quantum" in the country.

Even as the US celebrates its lead in the list of TOP500 supercomputers in the world, China has been slowly building its expertise in the next frontier of computing - quantum computing. Unlike conventional computing, where a bit- the smallest block of information can either exist as one or zero, a bit in quantum computing can exist in both states at once.

Known as a qubit, it allows basic information to represent all possibilities simultaneously, theoretically, making them faster than conventional computers.

How fast is China's Jiuzhang? .

China's Jiuzhang first shot to fame in 2020, when the research team led by Jianwei performed Gaussian boson sampling in 200 seconds. The same on a conventional supercomputer would take an estimated 2.5 billion years.

Quantum computing is still in its infancy, and researchers worldwide have only begun testing how these systems work and can be used in the future. Pan Jianwei's team, however, decided to use the "noisy intermediate scale" quantum computers to solve real-world problems.

They put Jiuzhang to the test by implementing two algorithms commonly used in AI- random search and simulated annealing. These algorithms can be a challenge even for supercomputers, and the researchers decided to use 200,000 samples to solve it.

At current technological levels, even the fastest supercomputer would take an estimated 700 seconds to go through each sample and a total of five years of computing time to process the samples the researchers had in mind. In sharp contrast, Juizhang took less than a second to process them. That's 180 million times faster than the fastest supercomputer on the planet today.

Advantages of using Jiuzhang .

The US has also been working on quantum computers and has found that the sub-atomic particles involved in the computing process are prone to error even if exposed to the slightest disturbance from the surroundings. This is why quantum computers are operated in isolated environments and at extremely low temperatures.

Jiuzhang, on the other hand, uses light as a physical medium for calculation and does not need to work at extremely low temperatures either. However, the researchers claim it does not require very low temperatures to operate.

The team purposely used some of the advanced algorithms that are in use today to demonstrate the advantages of using quantum computing. The research has demonstrated that even early-stage "noisy" quantum computers offer a distinct advantage over classical computers.

The research team said that the computations achieved by Jiuzhang could also help researchers apply the technology in areas such as data mining, biological information, network analysis, and chemical modeling research, the research team said.

The research findings were published in the peer-reviewed journal Physical Review Letters last month.