At today’s Quantum Summit, IBM announced the successful development of its 433-qubit 2022 roadmap target, the “Osprey” QPU (quantum processing unit). The new QPU significantly increases the number of qubits operating within a single QPU. The previous generation “Eagle” QPU had only 127 of them.
The new launch is another part of IBM’s aggressive quantum computing roadmap, which aims to deliver QPUs with tens of thousands (and potentially hundreds of thousands) of qubits by 2030. A sure step.
Dr. Dario Gill, senior vice president and research director at IBM, said:
“We are working with partners and clients around the world to continuously scale and advance quantum technology across hardware, software, and traditional integration to meet the greatest challenges of our time. This work will lay the foundation for the next era. Quantum-centric supercomputing.“
The launch of Osprey is significant for IBM. It’s in the middle of IBM’s roadmap, offering the biggest increase in the number of qubits in a single chip. Compared to Eagle, Osprey increases the number of qubits by 3.4 times. This is an even greater qubit count increase than the company, which plans to introduce his Kookaburra QPU of 4,158 qubits, expects he will achieve in three years. It’s also higher than any other qubit jump since the introduction of Falcon and its 27 qubits in 2019.
Osprey’s position within IBM’s roadmap will see the company deploy multiple QPUs to Heron and its p Coupling — Increasing the number of qubits without compromising quality is very important. But perhaps more impressive is the fact that the qubit count explosion was designed at the same time that IBM laid the foundation for most of its future modular offerings.
The company aims to deploy a 133-qubit, scalable Heron QPU by 2023. It utilizes p-coupling to interconnect multiple Heron chips. The idea is that it’s easier to scale the qubits within a given package and link the separate packages together than to create a monolithic QPU.
That poses a challenge for workload distribution — there are several ways to partition a large quantum problem and fit it onto the available chip (or chips) to run the quantum circuit. severely impacts performance. However, multichip scaling is considered essential, and adopting this approach would mean redesigning the entire control electronic subsystem, providing a bridge between classical and quantum computing. .
According to Dr. Oliver Dial, chief hardware architect at IBM Quantum, qubit control within the company’s dilution refrigerator, hardware for cooling superconducting qubits to near absolute zero (-273.15 °C). A change in the mechanics brought significant improvements.
Before Osprey, IBM used coaxial cables to transmit microwave control information to working qubits. Coaxial cables have now been superseded by flexible ribbon cables (the same type of cables used in electronics and hinges, such as laptops). These ribbon cables themselves occupy significantly less space, offering much higher throughput than previous solutions while reducing deployment time and resources. Dr. Oliver Dial says he was able to improve control density by 70% while IBM reduced costs by a factor of five.
Another key element of this new quantum generation from IBM was the increased performance of Field-Programmable Gate Arrays (FPGAs) within the control subsystem.
IBM’s future of qubit control passes through quantum-specific ASICs (application-specific integrated circuits), but FPGAs have so far handled the hustle and bustle of the task because of their flexibility. IBM can prototype various control schemes within FPGA programmable designs. This allows for rapid experimentation and iteration until the company is confident enough to go the full ASIC route. Dr. Oliver Dial says this change reduces the wattage required to control a single qubit from about 100 W to just 10 milliwatts, resulting in significant power efficiency improvements.
image 1 of Four
Importantly, Osprey’s superconducting qubits match the company’s best coherence times (despite a significant increase in qubit count), said Dr. Oliver Dial. It should increase with the number of qubits.
According to IBM, the number and quality of Osprey’s qubits are such that conventional systems trying to describe the computational state of that qubit would require more available bits than atoms in the universe. is. It seems that we have already entered the quantum dominance phase of the equation.
Of course, qubits can improve both in number and quality, but pure quantum hardware solutions have little to offer the average user. Dr. Oliver Dial was quick to point out that anyone, really anyone, can now spin up his IBM quantum technology via IBM’s cloud offerings.
Quantum computing requires a rigorous abstraction effort to allow non-quantum experts to interact with these systems, and IBM The Quiskit Runtime software has also doubled Just to make it easier for users to do this. Sacrificing runtime for correctness is as easy as changing a software setting.
With driver improvements, Quiskit runtime, and parameterized circuit improvements throughout 2022, IBM shaved the 1,400 CLOPS score down to a maximum of around 15,000 CLOPS. What metaphor is appropriate for this nearly 11x performance improvement?
Hardware improvements are only half of what it takes to develop a new technical system. The other half is actually using that hardware. To that end, IBM also announced his 100×100 challenge at the Quantum Summit. This is an initiative that aims to put a computational depth of 100 qubits x 100 gates into the hands of users by 2024. By leveraging Heron, the company’s next-generation modular quantum architecture, IBM aims to challenge users with the question, “What fits here?” Question — What kinds of quantum computing problems can be handled within these constraints?
As far as we know about quantum computing, something really special can come out of this challenge. It doesn’t matter if humanity will benefit greatly when quantum computing starts solving problems. It’s simply a question of when and how to do it, and this challenge gives IBM a sure way to keep that conversation going. After all, this is one of the company’s concerns for Quantum Summit 2022.
“IBM Quantum Summit 2022 marks a pivotal moment in the evolution of the global quantum computing sector as we move forward along our quantum roadmap. We will continue to look to make them more usable,” said Jay Gambetta, IBM Fellow and vice president of IBM Quantum. “Defining the next wave of quantum called supercomputing, modularity, communications, and middleware will contribute to enhanced scaling computational power and the integration of quantum and classical workflows.”
Against the backdrop of these statements and the speed of IBM’s quantum development (and flawless execution according to its roadmap), the company’s bet on superconducting qubits seems to be paying off. As IBM has shown, apart from increasing qubit densities and coherence times that enabled the leap from 127-qubit Eagle to 433-qubit Osprey, there is still a lot of time left to optimize their quantum system. is spent.
Osprey took flight, but IBM never stopped looking to the quantum future. That will continue with his 1,121-qubit Condor in 2023.