Quantinuum's 56-qubit H2-1 quantum computer has broken Google Sycamore's all-time quantum supremacy record by delivering 100 times higher performance in test measurements.
Quantum computers are computers in which quantum states replace bits, and due to their special properties, they can perform much more complex calculations. A quantum computer can also solve puzzles that a classical computer cannot solve in a reasonably expected time – this is called quantum supremacy. An example illustrates it all: the quantum supremacy record set by Google Sycamore in 2019 was a calculation performed in 200 seconds, which would have taken a classical supercomputer ten thousand years (true, IBM said at the time that it wasn’t, because instead of ten thousand years, their supercomputer would have taken two and a half days). This is what happened to H2-1.
Quantinuum was formed in 2021 from the merger of two quantum computing teams. Former members of Honeywell Quantum Solutions provided hardware expertise, and Cambridge Quantum Computing provided quantum software expertise. The new company is working in collaboration with Microsoft on development. Our paper also reports its results, which were achieved in the area of one of the key bug fixes needed to implement working quantum computers.
Switch over
The new result is also linked to this success, because at Quantinuum they have achieved fault tolerance that was years ahead of its time. Quantum bits carrying quantum information are technically solved using superconductors, photonics or ion traps, but quantum states are extremely fragile even with cryogenic cooling. The Colorado company’s ion trapping technology has made it possible to continuously improve the noise that creeps into qubits.
For the cross-entropy (or XEB) scale, which measures quantum entropy, 1 is flawless and 0 is completely flawed. Google’s Sycamore achieved quantum supremacy with 53 superconducting qubits at a very noisy 0.002 XEB. Working with support from Caltech, Agronne Lab, and JPMorgan, Quantinuum achieved a value of 0.35, meaning that 35% of the qubits were flaw-free.
Programs running on quantum computers are not intended to shoot Cacodemons, but rather demonstrate their true capabilities in a narrower set of algorithms. As Nobel laureate Richard Feynman pointed out, it is extremely difficult for classical computers to simulate many-body quantum systems, but for quantum computers this is their home turf,
They “solve themselves”.
According to the report uploaded to the arXiv server, the H2-1 was not only 100 times faster at calculating the Random Circuit Sampling algorithm, but also consumed 30,000 times the energy of a classical supercomputer.
According to the thesis, the era when traditional supercomputers will die out is imminent, but this requires better physical error correction, even more logical qubits, higher clock rates and network operation.
(Internet News, Life Science, TechSpot)
Follow the index on Facebook too!
