This week in a blog post written by Rigetti's VP of quantum device architecture, Rigetti details the release of their new Ankka-1 system. This release marks the fourth generation of architecture Rigetti has released, following the three generations of their Aspen-class systems.
What Makes this Chip Different?
The Ankka-1 System is representative of a 84 qubit chip, an upgrade from Aspen-M-3's 80 qubit chip. Four additional qubits may not seem like a lot, but the chips have entirely different architectures leading to stronger qubit connectivity and more efficient algorithms. One of the struggles with creating a qubit chip is regulating qubit connectivity. When placed next to each other, superconducting qubits can couple strongly and quickly. This is desirable when utilizing a chip in computing, but this process also needs to be controlled. Without control, the coupling can be too strong. The design of the Aspen-M-3 system used frequency (wavelength/energy) separation to weaken qubit interaction when the chip was not in use. This process "parks" one of the neighboring qubits at a higher frequency than the other, meaning the interactions between the two qubits weakens. The frequency of the qubits can then be manipulated again to bring them closer together when computing is active. However, this process is time consuming, and weakens overall coupling strength.
Why is coupling so important? Rigetti uses superconducting qubits. This type of qubit requires superconducting circuit architecture and materials to design the chip that holds them. Once the qubits are in the chip, they can be coupled by local interactions (with their neighboring qubits) and undergo a series of quantum gates (similar to classical computer gates). Strong coupling, as described above in the unregulated qubits, can lead to quicker decoherence time and increased error. However, superconducting qubits are also easily controlled by microwave signals (frequency manipulation) which can help offset these issues. When designing a superconducting qubit chip, it is important to find a balance between these properties to give both strong coherence to be used in rapid quantum gate operations, and slow decoherence time.
With their new chip design, Rigetti seems to have struck this balance. The Ankka-1 chip has a lattice-like structure and uses tunable couples to moderate qubit interaction. The tunable couplers allow for neighboring qubits to remain close in energy (in comparison to the "parking" technique of the Aspen-M-3 system), but prevent them from interacting. With just a small change to the tunable coupler, the system can be activated with a strong quantum gate. Think of a tunable coupler as a mediator between the two qubits; by tuning the frequency of the coupler, the effective coupling between the two qubits can be turned on and off. With this technique, the qubit's own frequencies are not being manipulated, only that of the coupler, which can lead to a controlled, but strong coupling.
The use of tunable couplers also allow for each qubit to have more neighbors. This changes the structure of the chip from that of an octagon lattice in the Aspen-M-3 system (3 qubit neighbors) to a square lattice in the Ankka-1 system (4 qubit neighbors). The tunable couplers prevent unwanted interaction between neighbors, meaning qubits can be placed in a more densely occupied chip, enabling more controlled qubit coupling and efficient algorithms.
Riggeti's Scalability Advantage
The new lattice structure of the chip with the tunable couplers leads to increased coupling and computing power. However, the use of the tunable couples adds to the logistics of chip operation. Since each coupler is technically a qubit (just a different form from the superconducting qubits mentioned before) the chip hosts about 3x as many qubits than the Aspen-M-3 chip. Thus, as each coupler is controlled independently, control signals routed to the chip have doubled. Readout lines have also increased from 20 to 28.
One of the big questions surrounding quantum computing is scalability. In order for quantum to have the applications predicted, in optimization problems, sensing, chemistry, etc, there needs to be a machine built that connects hundreds of these chips. This is really hard to do as a singular chip is already hard to regulate in an efficient manner. However, something that sets Rigetti apart from competitors is their use of lateral signal delivery. In the Ankka-1 system, signals are delivered to the qubits from directly above, meaning there is less signal fan out or routing needed from the chip's perimeter. Thus, when scaling their system, Rigetti only needs to expand vertically to connect the chips, helping with both spacing issues and connectivity between chips.
Steps Forward
Although the new system was only announced earlier this week, Rigetti is already working to update Annka-1's capabilities to hopefully roll out Annka-2 later this year. Rigetti has also partnered with Riverlane, a startup that works in error correction for quantum devices. Error correction is another important factor in terms of scalability. CEO and Founder of Riverlane Steve Brierley states, “This project enables us to target real-time error correction decoding with our algorithms on Rigetti’s FPGA hardware, which we hope will help improve performance on future systems.” The Ankka-1 system and partnership with Riverlane are important steps for Rigetti and their growth in the up and coming quantum industry. Check out the full article from Rigetti here and read more about superconducting qubits and their potential here!
Comments