Benchmark of nine cloud quantum computers finds entanglement covers few qubits

Started by Dragon, Today at 08:57 AM

Previous topic - Next topic

Dragon and 1 Guest are viewing this topic.

Topic: Benchmark of nine cloud quantum computers finds entanglement covers few qubits   Views(Read 17 times)
Active members in this topic:
Dragon(1)

Dragon

A new benchmark puts quantum computers' qubit counts in perspective. Quantum Computing Report says a Spanish consortium led by Gradiant, UPV/EHU and BCAM tested nine processors from five vendors through IBM Quantum Platform, Amazon Braket and CESGA

The test measured how large a GHZ state, where every qubit is entangled, each machine could certify. It used default compilation and no error mitigation, so it reflects what an ordinary cloud user would get

IBM Phoenix reached 27 of its 120 qubits, IBM Basquecountry 24 of 156, IBM Pittsburgh 21 of 156 and IQM Emerald 20 of 54. Smaller systems did proportionally better, with AQT IBEX Q1 reaching 12 of 12 qubits and IQM Garnet 13 of 20, while Rigetti Cepheus-1 managed 4 of 107 and CESGA Qmio from OQC managed 3 of 32

Under the strictest certification test, only IBM Phoenix, IBM Basquecountry, AQT IBEX Q1 and IQM Garnet passed. The work is an arXiv preprint and has not been peer reviewed

Headline qubit counts can be misleading, and benchmarks like this show what users can rely on in practice. Small, high quality machines entangling all their qubits is a reminder that quality matters as much as quantity. With error mitigation and better compilation, the numbers would likely improve

Should vendors publish entanglement benchmarks alongside qubit counts? And how much do default cloud settings hold machines back?