It was December 2019 when the Intel released the first details of the cryogenic control chip Horse Ridge. The project is expected to contribute to the development of the quantum computing. Like? This week, the company revealed that it had been able to prove that the chip fulfills its promise to mitigate one of the limitations of quantum computers: the interconnection bottleneck.
In the current phase, quantum chips need to operate at very low temperatures, often around 20 millikelvin, a measure that is equivalent to -273 degrees Celsius (absolute zero). To achieve such low temperature levels, these chips are subjected to cryogenic refrigeration chambers.
On the other hand, the complex electronic components that allow the manipulation or reading of the chip’s qubits status by the researchers are operated at room temperature.
In addition, the connection of these components to the qubits must be made through a single wire for each one of them, otherwise, there will be a loss of performance and other problems.
In time, qubit is a simplification of quantum bit. Traditional computing is based on the bit, which represents the value 0 or 1 in the symbolic model created to facilitate our understanding. A qubit, on the other hand, can assume 0, 1 or an overlap of both values.
This logic should allow quantum computers to solve problems in areas such as artificial intelligence in minutes that, because they are so complex, would require even years to be solved by traditional computing.
The temperature difference between the chip and the other electronic components is an important bottleneck because it makes it difficult to manipulate the qubits. To make matters worse, this limitation tends to increase as more qubits are added to the chip due to the amount of wires needed to connect each one.
Not by chance, the problem became known as “interconnection bottleneck” or “wiring bottleneck”.
The Horse Ridge Chip
Developed by Intel in conjunction with QuTech (an organization stemming from the partnership between Delft Technical University and the Dutch Organization for Applied Scientific Research), Horse Ridge was presented at the end of 2019 as an attempt to resolve this bottleneck.
The chip was built on Intel’s own 22-nanometer FinFET technology. The second generation of the chip was introduced in 2020. Tests have since been conducted to assess whether Horse Ridge meets expectations. The results are encouraging:
The results of our research, conducted in partnership with QuTech, prove quantitatively that our cryogenic controller, Horse Ridge, can achieve the same high fidelity results as room temperature electronics while controlling various silicon qubits.
Stefano Pellerano, an engineer who leads Intel Labs, explains the relevance of this advance:
Making electronic controls operate in high fidelity at cryogenic temperatures is the key to overcoming what is known as an “interconnection or wiring bottleneck”.
Stefano Pellerano, lead engineer at Intel Labs
Edoardo Charbon, QuTech researcher who works on the project, adds:
“As electronic components operate in very different ways at cryogenic temperatures, we use special techniques in the design of the chip to ensure its correct functioning and to handle qubits with high precision.”
Edoardo Charbon, QuTech researcher
In the same work, Intel was able to successfully demonstrate frequency multiplexing with a single wire to control two qubits.
Why is it important? Today, each qubit is controlled with its own thread, as has already become clear. The problem is that, with the expected progressive increase in the number of qubits on a quantum chip, this approach will become more difficult.
Today, quantum chips have only a few dozen qubits; but in the future, imagine having millions of wires on a chip with millions of qubits. Controlling more than one qubit with a single wire is therefore a necessity.
Intel hopes that the research will enable, in the next phases, the cryogenic control chip to be combined with qubits in a single matrix (in the same way that we can have GPU and CPU in the same module, for example).
Such a solution can pave the way for “quantum scalability”, that is, for quantum chips with increasing numbers of qubits.
The results of the work were published in the scientific journal Nature.