Quantum Computing Firm D-Wave Leaps To Higher Level With New Lab
The world’s only commercial quantum computing company has moved to new premises as it builds bigger chips
D-Wave, the only commercial quantum computer company, has opened new labs in British Columbia. The firm has sold quantum computers, which promise to radically speed up complex problem solving, to both Google and NASA.
The 40,000 square foot facility has been upgraded and retrofitted over the last year to support the work of the maverick venture-funded startup. This year, the company created the second generation of its quantum computer, D-Wave 2, and set up a manufacturing deal with a silicon fab which ensures the firm can develop even faster computers as well as building up a possible market for them.
Commercial versus academic quantum computers?
Quantum computers offer a massive speed-up over conventional computers because they make use of the fact that quantum systems exist in multiple states at the same time, an effect made famous by Schrodinger’s cat paradox, in which a cat sealed in a box away from all interactions, can be simultaneously alive and dead.
Quantum computers can process multiple inputs at the same time – effectively in parallel universes – and determine which is the right answer to problems which have huge numbers of possible solutions, as long as their quantum bits (qubits) can be kept in a coherent state, isolated from the outside world.
So far, quantum computers are mostly the domain of scientific research, with centres like Bristol producing systems that have a few quantum bits (qubits) made using photons. Mostly these are used for scientific research around the concepts of quantum computing.
D-Wave has taken a different approach. Backed by around $100 million of venture capital, it has built systems using qubits created through spin in superconducting magnets. While academics struggle to make handfuls of qubits, D-Wave has put together the D-Wave 2 machine, which has 512 qubits, and claims that it solves real-world problems faster than any other option available.
NASA Ames and Google have both bought D-Wave machines, though neither have said how much they cost, and D-Wave wants to sell more.
“We can make 120 quantum chips at a time, on an eight-inch wafer,” Colin Williams director of partnerships at D-Wave told TechWeekEurope. Williams joined D-Wave from NASA’s Jet Propulsion Laboratory (JPL) which built earlier chips for D-Wave. The newer generations are now built by a silicon fab in Minnesota – which presumably means it is Cypress Semiconductor.
“While academics get about two designs of quantum computer a year, we can manage six to eight,” he said. “That is what differentiates us from the academic approach.”
D-Wave also uses a distinctive approach to quantum computing – quantum “annealing” – which has caused some controversy. The machine is designed to solve optimisation problems where the user wants to find the best solution out of many millions of possibilities. The D-Wave machine solves a particular one of these “NP-hard” problems, in the following way. It is designed so that the solution to the problem will have the lowest energy state: it starts in a coherent state in which the possible solutions exist simultaneously and then settles until only one state exists – the solution.
It’s called annealing because of the similarity with metallic solids which can gradually settle to a lower energy state as their crystalline structure re-aligns. “If you can solve one NP-hard problem well, you can transform others onto it,” said Williams. This means that quantum computers could potentially be used for jobs like financial modelling and analysing the structures of proteins.
The systems are expensive because they require serious cooling. It needs “dilution refrigerators”, which take the quantum system down to a temperature of 20 millikelvin (mK), far colder than the superconducting magnets of the Large Hadron Collider, and about 150 times colder than interstellar space. The system also has to be shielded from radio signals and the earth’s magnetic field.
As well as designing its chip, D-Wave has been working to make dilution refrigerators more cost-effective, and also has patented ways to pass signals from electronics at room temperature to the parts of the system at close to absolute zero.
But is it really quantum?
From the start, D-Wave has faced controversy. At first, academics expressed doubts that its systems were actually exhibiting quantum computing at all – but a USC research team led by Sergio Boxio found in July that it appears to be real quantum computing (paper published here). However, sceptics still argue over whether the system actually produces real benefits over classical computer systems that could be run on ordinary laptops.
Critics worry that there is no real point in building a quantum computer if a simulated quantum computer will do the job just as well. According to MIT professor Scott Aaronson: “D-Wave founder Geordie Rose claims that D-Wave has now accomplished its goal of building a quantum computer that, in his words, is ‘better at something than any other option available”. This claim has been widely and uncritically repeated in the press, so that much of the nerd world now accepts it as fact. However, the claim is not supported by the evidence currently available.”
Williams disputed this, saying that NASA found the system to be 10,000 times faster than a state-of-the-art algorithm running on an off-the-shelf 2.4GHz Intel chip. “The system is also tunable,” he said, “so we have got up to 35,000 times faster.”
He also made two other claims. Firstly, that quantum computers can scale up faster (exponentially) than conventional computers, with each qubit added: “As the number of qubits goes up, the search space gets exponentially bigger”. And secondly, he claimed that while the classical approach brings back correct solutions, the quantum computer brings back the best correct solution. “You can use it in conjunction with a high performance computer to get better results,” he said.
There is some overhead in mapping other NP-hard problems to the D-Wave system, but D-Wave is working on other possible architectures that solve problems with easier mappings to real-world problems, said Williams.
D-Wave’s record is particularly impressive compared with conventional systems, he said: “While Nvidia spent $2 billion building a conventional GPU, D-Wave has produced a fundamentally new architecture using $100 million of venture capital. ”
He expects to have 2000 qubits by 2015 and this represents a substantial improvement, since “each time you double the number of qubits you get 2 to the power of n faster”, he said.
Interested in the Internet of Things? Try our quiz!