IBM readies System Two modular quantum architecture

IBM has unveiled the next milestone in its quantum computing roadmap. The company’s latest machine, Eagle, is a 127-qubit quantum computer and is being positioned as a step in a technological revolution in the history of computation. It is significantly larger than IBM’s existing 65-qubit Hummingbird processor.

In a blog post about Eagle, Jerry Chow, a manager at IBM’s experimental quantum computing group, and IBM fellows Oliver Dial and Jay Gambetta described how the new quantum processor enables greater problem-solving. This complexity makes it harder to simulate and test quantum algorithms on classical computing architectures.

“As quantum processors scale up, each additional qubit doubles the amount of space complexity – the amount of memory space required to execute algorithms – for a classical computer to reliably simulate quantum circuits,” they said in the blog post.

Doug Finke, industry analyst and publisher of Quantum Computing Report, said: “A key characteristic of computers of this size is that their operation can no longer be simulated with a classical computer because the computational resources required double every time another qubit is added. So classical simulation of these sized machines becomes an intractable problem.”

Testing algorithms on a quantum computer requires a different approach to the way a programmer would traditionally debug code on a classical computer. Bob Sutor, vice-president, IBM Q strategy & ecosystem, said: “One of the things about classical computing is that it can run in debug mode, where you look at a program [running] step by step and see if something goes wrong.” This enables a programmer to see the path of logic the program code takes while it is running. 

However, in quantum computing, Sutor said: “While you do operations on a single qubit,you can’t stop and look at its value as a qubit relies on quantum mechanics, so you can’t debug in the same way as a classical computer.”

Instead, said Sutor, it is necessary to test algorithms designed to run on a quantum computer at a smaller scale. In classical computing terms, this is equivalent to an app developer using a subset of data to test a financial application for the iPhone. If the results from the applications are what the programmer expects, then the algorithm is sound.

Eagle represents the last of IBM’s first-generation System One design for quantum computing. The next generation, System Two, is being designed as a more modular architecture. Sutor said: “System One is 3m³. Among the challenges in scaling the number of qubits is how to wire them altogether. Eagle is the last generation of IBM quantum computer that will fit in our existing configuration. We need to fix the wiring problem.”

Instead of using one wire per qubit, which is what IBM has done on the System One architecture, “we now multiplex”, said Sutor.

Chow, Dial and Gambetta said that System Two also introduces a new generation of scalable qubit control electronics, together with higher-density cryogenic components and cabling. Quantum processors are designed to run close to absolute zero, and IBM said it has collaborated with Bluefors on a new modular cryogenic platform.

“This platform brings the possibility of providing a larger shared cryogenic workspace, opening the door to potential linking of quantum processors through novel interconnects,” Chow, Dial and Gambetta wrote in a blog post describing System Two. “We think that System Two represents a glimpse into what the future of quantum computing looks like – a true quantum datacentre.”