Quantinuum Scientist Says Benchmarking Data on H2 System Model Shows It’s a Quantum Computer to Beat
- The researchers report the H2 Quantinuum System Model surpasses H1 in complexity and qubit capacity while retaining all the capabilities and fidelity of previous generations.
- The newest entry in the H-Series starts with 32 qubits while the H1 starts with 10.
- The data also shows that none of these hardware changes detract from the high level of performance achieved by the System Model H.
Quantinuum’s new H2-1 quantum computer proves that the trap-ion architecture, renowned for achieving exceptional qubit quality and gate precision, is also built to scale – and the Quantinuum benchmarking team has the data to prove it.
Bottom line: The new H2 System model outperforms the H1 in complexity and qubit capacity while retaining all the capabilities and fidelity of previous generations – a tremendous achievement when developing successive generations of quantum systems.
The newest entry in the H-Series starts with 32 qubits while the H1 starts with 10. The H1 underwent several upgrades, eventually hitting a capacity of 20 qubits, and the H2 is ready to take the torch and run with it. Staying true to its ultimate goal of increasing performance, H2 has not only increased the number of qubits but has achieved a higher Quantum Volume than any other quantum computer ever built:216 or 65,536.
Most importantly for the growing number of industries and academic research institutions using the H-Series, benchmarking data shows that none of these hardware changes have detracted from the high level of performance achieved by the H1 System Model. That is the main challenge in scaling a quantum computer – maintaining performance while adding qubits. The error rate in the connected circuit is fully comparable to H1, even with a significant increase in qubits. Indeed, H2 outperforms H1 in several performance metrics: single qubit gate error, two qubit gate error, measurement cross-talk, and SPAM.
The key to the engineering advances made in the second generation of H-Series quantum computers is the reduction in the physical resources required per qubit. To get the most out of a device coupled with a quantum charge (QCCD) that formed the basis of the H-Series, the hardware team at Quantinuum introduced a series of component innovations, to remove some of the performance limitations of the first generation in areas such as ion loading, voltage sourcing, and sending high-precision radio signals to control and manipulate ions.
Research paper, “Race Track Trapped Ion Quantum Processor,” details all these engineering advances, and what impact they have on the machine’s computing performance. This paper includes results from component and system level benchmarking tests that document the capabilities of the new machine at launch. This benchmarking metric, aggregated by company advances in qubit topologyrepresents a new phase of quantum computing.
In addition to expanded capabilities, the new design provides operational efficiencies and a clear growth path.
At launch, the H2’s operations are still classically emulated. However, Quantinuum releases H2 at a small percentage of its full capacity. This new engine has the ability to upgrade to more qubits and gate zones, pushing it past levels that classical computers could hope to keep up with.
Improved Efficiency in New Trap Designs
This new generation of quantum processors represents the first major trap upgrade in the H-Series. One of the most significant changes is the new oval (or race track) shape of the ion trap itself, which allows more efficient use of space and electrical control signals.
One of the main engineering challenges presented by the new design was the ability to route the signal beneath the top metal layer of the trap. The hardware team resolved this by using radio frequency (RF) tunnels. This tunnel allows the inner and outer voltage electrodes to be implemented without connecting directly on the top surface of the trap, which is key to creating the two-dimensional trap that will really increase the computing speed of these machines.
The new traps also feature voltage “broadcasting”, which conserves a control signal by tying multiple DC electrodes inside the trap to the same external signal. This is achieved in the “conveyor belt” area on each side of the trap where the ions are deposited, increasing the efficiency of electrode control by requiring only three voltage signals for the 20 wells on each side of the trap.
Another important component of the H2 is the Magneto Optical Trap (MOT) which replaces the effusive atomic oven used by the H1. MOT reduces startup time for H2 by cooling neutral atoms before shooting them into the trap, which will be critical for very large machines that use large numbers of qubits.
Industry Leading Results from 15 Benchmark Tests
Quantinuum has always valued transparency and backed up its performance claims with publicly available data.
To measure the impact of these hardware and design improvements, Quantinuum ran 15 tests that measure component operation, overall system performance, and application performance. Full results from the tests are included in a new research paper.
The hardware team ran four system-level benchmark tests that included more complex multi-qubit circuits to provide a broader picture of overall performance. These tests are:
- Mirror benchmarking: A scalable way to measure arbitrary quantum circuits.
- Quantum volume: A popular system-level test with comparable established constructs across gate-based quantum computers.
- Random circuit sampling: A computational task to sample the output distribution of a random quantum circuit.
- Entanglement certification at Greenberger-Horne-Zeilinger (GHZ) certifies: A demanding qubit coherence test that is measured and widely reported on a variety of quantum hardware.
The H2 demonstrated state-of-the-art performance on each of these system-level tests, but the GHZ test results were impressive. Verification of the globally entangled state of the GHZ requires relatively high fidelity, which is increasingly difficult to achieve with larger numbers of qubits.
With 32 qubits of H2 and precise control of the environment in the ion trap, Quantinuum researchers were able to achieve an entangled state of 32 qubits with an accuracy of 82.0(7)%, setting a new world record.
In addition to system level testing, the Quantinuum hardware team runs the following component benchmark tests:
- SPAM experiment
- Single qubit gate random benchmark
- Two-qubit gate random benchmark
- The two-qubit SU gate scrambles RB benchmarking
- Random comparison of two qubit parameterized gates
- Crosstalk benchmarking/resetting
- Inserted transport random comparison
This paper includes the results from that test as well as the results from benchmarking this app:
- Hamiltonian simulation
- Quantum Approximation Optimization Algorithm
- Error correction: repeat code
- Holographic quantum dynamics simulation
The H2 datasheet is available on GitHub here: https://github.com/CQCL/quantinuum-hardware-specifications