Extract from ABC News
Australian researchers have unveiled an invention being hailed as "the first decent pick and shovel" in the "gold rush" to develop large quantum computers.
Key points:
- A new kind of chip that works in ultra-low temperatures removes a major barrier to scaling up quantum computers
- Experts say it could lead to much faster quantum computers within the next few years
- The invention was the product of an ongoing collaboration between Microsoft and University of Sydney
The device potentially opens the door for quantum computers with thousands of qubits — the building blocks of quantum computing — while the most advanced current models have a few dozen.
This, in turn, could eventually lead to scientific advances that translate into better medicines or more accurate climate and financial system models.
The research, published in the journal Nature Electronics, describes a new way of approaching one of the great limits to scaling up quantum computers: heat.
Most quantum computers operate close to absolute zero (minus 273 degrees Celsius, or 30 times colder than interstellar space), but also need to be wired to a standard "control chip" that runs at room temperature.
These wires transfer data, but also heat. As more qubits are installed, more wires are needed to control them, and the task of keeping the qubits cool becomes harder.
A team of engineers and scientists from the University of Sydney and Microsoft solved this problem by inventing a control chip that can operate at the same temperature as the qubits.
In essence, they put the chip inside the quantum computer deep-freeze.
To achieve this, they designed, from scratch, an integrated circuit that could withstand extreme cold and produce negligible heat.
The result means only two input wires could, in theory, control thousands of qubits.
"We've lifted the barrier that was limiting qubit count to tens of qubits," said David Reilly, a co-author of the paper who is employed by the University of Sydney and Microsoft.
"Over the next few years, these types of chips will be the reason machines will be able to scale into the many thousands."
'This is going to be transformational'
In a field where advances are incremental, and each new machine has only a handful more qubits than the last, this is a bold claim.
Other experts in the field, however, are eager to trumpet the advance.
University of Queensland's Andrew White, director of the Centre of Engineered Quantum Systems, was not involved with the research, but described the new kind of chip as "transformational".
The chip replaces racks of electronics with an integrated circuit the size of a thumbnail — a feat of miniaturisation that Professor White said was like leaping straight from a "1970s mainframe" to the tiny chip in a modern smartphone.
"Given there's a worldwide gold rush for quantum technology, with big players like Google, Amazon and IBM, what David and his team have done is built the first decent pick and shovel," he said.
Quantum computing harnesses some of the mind-boggling properties of quantum mechanics to deliver huge leaps forward in processing power.
In traditional computing, bits are either 1 or 0, while in a quantum computer, qubits can be both numbers at the same time.
This ability to perform multiple calculations at once means that — in theory — quantum computers can perform calculations many times faster than a traditional computer.
But despite billions of investment dollars in recent years, the technology has not yet lived up to its potential.
In late 2019, Google announced that its 53-qubit quantum computers had been able to solve a problem in 200 seconds that they estimated the fastest supercomputer in the world would take 10,000 years to complete.
IBM later cast doubt on this, saying one of its own computers would take 2.5 days.
In any case, though, the problem it solved was conceptually interesting, it had no real application beyond quantum research itself.
It's estimated quantum computers will need to be composed of thousands — perhaps millions — of qubits before they can prove truly useful, and that remains at least several years away.
UNSW quantum computing expert Andrew Dzurak, who was not involved in the work, said the announcement by Professor Reilly's team was "a very important next step.
"The thing that's impressive and a real breakthrough is they have integrated a very significant amount of electronics onto a single chip.
"And that chip only operates at 100 millikelvin [a 10th of a degree above the lowest possible temperature] and dissipates such a small amount of heat, it doesn't heat up the fridge or the quantum chip next to it."
Approaching the heat problem
Up to now, classical computer chips have been unable to work at the temperatures required to operate most quantum computers.
And even if they could, they produce too much heat to maintain those chilly temperatures.
In response, laboratories wanting to build quantum computers with more qubits have simply built larger deep freezers called "dilution fridges".
IBM, for example, plans to build a 3-metre-tall "super-fridge" codenamed Goldeneye that is larger than any commercially available today.
Another approach has been to build qubits that operate at higher temperatures. Last year, two teams of physicists — one in Australia, the other in the Netherlands — demonstrated silicon-based quantum processors that function at 1.5K.
Though that's still very cold, it's 15 times hotter than the quantum computers being developed by the likes of Google and IBM.
Four years ago, Professor Reilly and his team set out to find another way around this heat problem.
Though this was a daunting challenge, they had a few advantages.
That same year, in 2017, Microsoft chose University of Sydney as a site for its global network of quantum research laboratories.
The partnership gave the university access to the best computer chip foundries in the world — multi-billion-dollar plants capable of producing new chip designs.
For the researchers, the lightbulb moment came when they realised that the control chip in conventional quantum arrays used most of its power forcing signals to travel several metres down wires to the quantum chip in the fridge.
As it did so, the chip generated heat, which was transferred down the wires to the quantum chip.
By placing the new control chip, which could operate at ultra-low temperatures, beside the quantum chip, the researchers effectively shrunk that distance to a few millimetres.
This meant the chip saved huge amounts of power — and produced almost no heat, Professor Reilly said.
"That was the key moment where we realised there is a way of doing this," he said.
"We built lots of models and design libraries to capture the behaviour of transistors at deep cryogenic temperatures.
"Then we had to build devices, get them verified, characterised and finally connect them to qubits to see them work in practice."
Boost to quantum research reputation
The new control chip design has solved a hardware problem, Professor White said, but there are other limits to scaling up quantum computers: namely, designing software that works on quantum computers and making better, more reliable qubits.
Qubits remain crippled by errors in the form of noise and faults.
Despite these challenges, Professor White said the world was undergoing a "second computer revolution".
Government has been pouring money into the quantum computing research, with Australian Research Council funding for quantum research since 2011.
Post-GFC federal government stimulus also partly funded construction of the University of Sydney's nanofabrication facility, opened in 2013, where Professor Reilly and his team worked on the new chip.
"We missed out on our opportunities in the first computer revolution," Professor White said.
"But this research has got us a seat at the table of international semiconductor development and that's a really hard table to get a seat at."
No comments:
Post a Comment