The December 2022 issue of IEEE Spectrum is here!

Close bar

IBM Unveils 433-Qubit Osprey Chip

Next year entanglement hits the kilo-scale with Big Blue’s 1,121-qubit Condor

5 min read
A blow out view of the three internal layers of a silver rectangle labelled IBM Quantum Osprey

IBM's 433-qubit Osprey quantum processor paves the way for the company's 1,121-qubit Condor processor in 2023.

Connie Zhou/IBM

In 2021, IBM unveiled Eagle, the first quantum processor with more than 100 qubits. Now the company has debuted Osprey, which possesses more than three times as many qubits. The advances IBM made to triple the number of qubits on a chip in just one year suggest Big Blue is on track to deliver Condor, the world’s first universal quantum computer with more than 1,000 qubits, in 2023, the company says.

Quantum computers can theoretically find answers to problems that classical computers would take eons to solve. The more components known as qubits are quantum-mechanically linked or entangled together in a quantum computer, the more computations it can perform, in an exponential fashion.

The qubit numbers of IBM’s quantum computers have steadily grown over time. In 2016, the company put the first quantum computer on the cloud for anyone to experiment with—a device with 5 qubits, each a superconducting circuit cooled to near-absolute-zero temperatures of roughly 20 milliKelvin (-273 degrees C). In 2019, the company debuted the 27-qubit Falcon; in 2020, the 65-qubit Hummingbird; and in 2021, the 127-qubit Eagle.

“Each one of the improvements we used led to a little improvement in speed, but once we lost all these bottlenecks, we saw a major improvement in speed.”
—Oliver Dial, IBM

Next year, IBM aims to launch its 1,121-qubit Condor processor, which stands poised to become (barring any surprises in the meantime) the world’s largest general-purpose quantum processor. The new Osprey chip, unveiled at IBM’s Quantum Computing Summit on 9 November, reveals key steps the company is taking in order to scale up to this ambitious goal.

One strategy that began with Eagle and continues with Osprey is to separate the wires and other components needed for readout and control onto their own layers. This multi-level wiring helps protect infamously fragile qubits from disruption, helping the processor incorporate larger numbers of them.

“We probably didn’t need all that technology to deploy a 100-qubit device, but doing all that helped set up Osprey and Condor,” says Oliver Dial, IBM Quantum’s chief hardware architect. “We now have the technology in hand to go way beyond 100 qubits.”

Osprey possesses two major advantages over Eagle outside the chip, Dial notes. One is replacing the “quantum chandelier” of microwave cables IBM used with its previous quantum processors with flexible ribbon cables, the kind you might find that carries signals between, for instance, the motherboard and screen if you open up a cellphone or a laptop, he says.

Chart with images and labels shows IBM Falcon (27 qubits), IBM Hummingbird (65 qubits), IBM Eagle (127 qubits), IBM Osprey (433 quibits).IBM’s 433-qubits Osprey quantum processor more than triples the 127 qubits on the IBM Eagle processor unveiled in 2021.Connie Zhou/IBM

“All these microwave cables, which get microwave signals in and out of the refrigerator where the qubits are stored, are not very scalable,” Dial says.

Osprey’s flexible ribbon cables are adapted to cryogenic environments. The electrical and thermal resistance of the cables are tailored to help microwave signals flow while not conducting too much heat that might interfere with the qubits. This led to a 77 percent increase in the number of connections leading to the chip—“basically, almost twice as many wires”—which will help IBM scale up its quantum computers, Dial says.

The other major advantage seen with Osprey is a new generation of the control electronics that send and receive microwave signals to and from the quantum processor. Whereas Dial says IBM’s first phase of control electronics (2019-’21) enjoyed a greater flexibility, Osprey’s control electronics “are more specialized, more tailored to quantum devices, to produce the exact signals we need, the frequencies we need, the power we need,” Dial says.

“We play Osprey to get on the cloud the middle of next year.”
—Oliver Dial, IBM

These improvements “have reduced cost, which is an important consideration as we scale up,” Dial says. “With our first generation of five and 20 qubit devices, we needed an entire rack of control electronics, and with Eagle we saw 40 qubits per rack. Now we can control more than 400 qubits with one rack of equipment.” He adds that qubit density has also increased with Osprey.

IBM’s new control electronics include a cryo-CMOS prototype controller chip implemented using 14-nanometer FinFET technology that runs at roughly 4 Kelvin (-269.15 degrees C). (According to IBM, it is expected to be implemented into future generations of their quantum computer control electronics.) The prototype chip uses an application-specific integrated circuit (ASIC) design that is less bulky and power-hungry than previous field-programmable gate array (FPGA) approaches. “Instead of about 100 watts per qubit like we needed before, we only need about 10 milliwatts, so we can fit far more qubits onto a chip,” Dial says.

In addition, recent advances in microwave signal generation and reception from the telecommunications and defense industries “means direct digital synthesis is finally affordable,” Dial says. “Instead of generating signals at a few hundred megahertz and mixing to 5 gigahertz, you can now directly generate at 5 gigahertz, which reduces the number of components and increases simplicity.”

These hardware improvements—along with other factors, such as better methods to handle quantum computing workloads, and faster device drivers—have led to a major boost in speed. Based on an IBM quantum computing speed metric known as circuit layer operations per second (CLOPS), the company has gone from 1,400 to 15,000 CLOPS with its best systems, Dial says. (Quantum programmers run quantum algorithms on quantum computers that are made up of quantum circuits, which describe sequences of elementary operations, called quantum gates, that are applied on a set of qubits. CLOPS is a measure of the speed at which a quantum computer runs quantum circuits.)

“Quantum chandelier” no more: IBM says its Osprey processor introduces a high-density control signal delivery with flex wiring, pictured here.Connie Zhou/IBM

“Each one of the improvements we used led to a little improvement in speed, but once we lost all these bottlenecks, we saw a major improvement in speed,” Dial says.

Before Osprey becomes widely available for use, IBM is spending extra time setting up its control electronics and calibrating the system. “We plan Osprey to get on the cloud the middle of next year,” Dial says.

IBM is also preparing to include optional error mitigation techniques within the cloud software for its quantum computers that can essentially trade speed for more accurate results. “Instead of pushing complexity onto the user, we’re building these capabilities in the back end to take care of these details,” Dial says. “By the end of 2024, we expect that error mitigation with multiple Heron chips running in parallel in our ‘100 by 100 initiative’ can lead to systems of 100 qubits wide by 100 gates deep, enabling capabilities way past those of classical computers.”

IBM also announced that it was partnering with communications technology company Vodafone to develop post-quantum cryptography that can defend against future quantum computers that could rapidly break modern cryptography. “We’re working on crypto-agility, the ability to move between cryptographic schemes to acknowledge how cryptography constantly changes and advances,” Dial says.

Update 10 Nov. 2022: A previous version of this story incorrectly stated that the cryo-CMOS controller chip was being used with Osprey. The chip is instead a prototype.

The Conversation (1)
Anjan Saha10 Nov, 2022
M

D-Wave 2000Q system

has got 2048 Qubits & 6016 couplers & 128000

Josephson junction and

Commercially available in the market.

D-Wave 200Q system consumes 25KW whereas Traditional Supercomputer 2500KW.

Quantum computer is

in Quantum Chemistry ,

Cryptography,Healthcar& Medicine,Share Mkt.

Biological research, Genetics. IBM Osprey

Quantum processor may be compact but

less efficient than commercially available

D-Wave Quantum computer in Qubit wise.

D-Wave 2000Q system operate at less than 15millikelvin and use liquid helium refrigerant.

To learn more contact:

www.dwavesys.com/resources

Why Functional Programming Should Be the Future of Software Development

It’s hard to learn, but your code will produce fewer nasty surprises

11 min read
Vertical
A plate of spaghetti made from code
Shira Inbar
DarkBlue1

You’d expectthe longest and most costly phase in the lifecycle of a software product to be the initial development of the system, when all those great features are first imagined and then created. In fact, the hardest part comes later, during the maintenance phase. That’s when programmers pay the price for the shortcuts they took during development.

So why did they take shortcuts? Maybe they didn’t realize that they were cutting any corners. Only when their code was deployed and exercised by a lot of users did its hidden flaws come to light. And maybe the developers were rushed. Time-to-market pressures would almost guarantee that their software will contain more bugs than it would otherwise.

Keep Reading ↓Show less
{"imageShortcodeIds":["31996907"]}