IBM’s 127-qubit ‘Eagle’ spreads its wings.

A notion that existed only on blackboards in the 1990s has burgeoned into a multi-billion dollar competition between governments, tech giants and plucky start-ups: harnessing the counter-intuitive properties of quantum mechanics to build a new kind of computer.

2021 has welcomed a slew of announcements revolving around the quantum realm, from Google eyeing a practical quantum computer by 2030, to the discovery of new materials as a replacement for silicon, to supercomputing experts challenging “quantum supremacy”. An Australian-German start-up even suggested that portable, quantum-powered electronic devices will soon be within reach.

And the industry is not showing signs of slowing down. Recently, IBM unveiled the development of a 127-qubit quantum processor, the most powerful one so far at the time of the announcement. (and hopefully still accurate by the time this article is published)

What’s new in Eagle?

Breaking the 100-qubit processor barrier, IBM built the new quantum processor based on a new technique – one that has qubit control components placed in a multiple physical layer architecture, with the qubits held in a separate layer.

In Eagle’s predecessors such as the 27-qubit Falcon and the 65-qubit Hummingbird, the wiring scheme is comparatively rudimentary. One chip containing qubits is “bump bonded” with another containing the necessary wiring. This arrangement poses real-estate issues and makes it hard for scale-up to be done.

Eagle’s multi-layered architecture is designed to overcome this problem, allowing more efficient circuitry to address each qubit in the processor. As shown in the hero image, the four layers sandwiched between the silver plates are the qubit plane, the resonator plane for qubit readout, the wiring plane that routes signals to the qubit plane, and the interposer plane that delivers signals. Check out the interactive breakdown of the individual layers in this IBM blog post. Spreading out the circuitry on several layers while isolating the qubits on their own layer greatly reduces the number of elements that can interfere with quantum coherence.

The new design also drew inspiration from the Falcon quantum processor. Eagle’s qubits are connected to either two or three neighbours via a heavy-hexagonal lattice – akin to a honeycomb – and are linked together using a circuit element called the quantum bus. The ability to pack qubits on one graph while limiting errors caused by interactions between neighbouring qubits preserves the qubits’ coherence time and yields functional processors. 

Realising a long coherence time, where qubits remain in their superposition state, is crucial in the design of a practical quantum computer. IBM is still working to benchmark the performance of Eagle, and one metric of particular interest is the circuit layer operations per second, or CLOPS for short. (Ah, yet another acronym added to the never-ending list of IT acronyms!)

The CLOPS metric is significant as a quantum computer does not generate a single, accurate result for a calculation like a classical computer. Instead, answers vary with every calculation. So to reach an accurate solution, the same calculation is performed by the quantum processor a dizzying number of times, eventually converging to an accurate solution. To illustrate, out of 1000 calculations, if A is found to be the solution 920 times, then A is concluded to be the accurate answer, even if the machine gave B 80 times.

More majestic birds to spread their wings.

Scaling its quantum technology is also on IBM’s radar. Plans to develop the 433-qubit Osprey by 2022 and the 1,121-qubit Condor by 2023 are already given the green light, supported by the IBM Quantum System Two infrastructure.

Instead of the single chandelier cryostat design housed by IBM’s Quantum System One, System Two takes a modular approach that provides flexibility in scaling the quantum chips. Working with Bluefors, a company specialising in ultra-cold chillers, IBM will introduce a new generation of scalable qubit control electronics together with higher-density cryogenic components and cabling. The platform offers a larger shared cryogenic workspace, linking more quantum processors together via novel interconnects – a glimpse of a quantum datacentre of the future.

It’s exciting to witness the rapid development within the quantum world, where companies race toward realising the quantum advantage that will ultimately accelerate science and research and solve relevant, real-world issues in fields from renewable energy, materials, finance and cybersecurity.

Stay tuned (often) for the latest news and information on everything quantum!


Main picture credits: IBM

By Mitchell Lim

Mitchell Lim is the latest addition to DUG’s science communication department. With a PhD in Chemical Engineering, Mitch is an expert in the fields of catalysis and ultrasonics. Full-time science geek, part-time fitness junkie, Mitch is living proof that brains and brawn are not mutually exclusive. His mission is to make science more accessible to you, while dreaming of having access to Martian land someday.