In the auto industry, car makers will often make minor improvements in the cars without completely introducing a new model. Oftentimes they do this when changing to a model year, but sometimes they will do this in the middle of model year. We’ve just seen something similar in the quantum industry with Honeywell Quantum Solutions (HQS) improving their Model H1 quantum processing unit (QPU) to achieve a Quantum Volume level of 1024 from the previous 512 and Google has also improved their Sycamore processor to significantly improve readout time, implement a Controlled-Z gate, provide a multi-level reset capability and implement mid-circuit measurement capabilities. It appears that all of these changes were implemented with improved qubit control firmware and algorithms and did not require any significant changes in the underlying hardware.
The Honeywell Model H1 was introduced in September 2020 with a Quantum Volume measure of 128. In March, they improved the Quantum Volume to 512 and this month they have reached 1024. Since the improvements were made by improving various parameters of qubit quality and it was not previously limited by the 10 qubits currently available in the H1. However, further improvements in Quantum Volume will require more qubits because a Quantum Volume of 1024 uses all 10 qubits, so we may have to wait until Honeywell introduces their Model H2 before that happens. In related research, Honeywell has used the H1 to provide an error correction demonstration that takes 7 physical qubits to create one logical qubit using a [[7,1,3]] color code. You can view a blog post from Honeywell about reaching Quantum Volume 1024 on their website here and you can read another blog post about their error correction work here.
In this week’s Google Quantum Summer Symposium, they described some work they have been doing to improve their Sycamore processor. Much of this has been driven by the research to demonstrate a viable error correction architecture for their end goal of achieving a million physical qubit processor (which will provide a thousand logical qubits) by the end of the decade. We reported on some of their recent error correction research in an article earlier this month. Perhaps the most significant feature added to Sycamore is the capability to perform mid-circuit measurement and qubit reuse, a very important feature needed for error correction. To do this, they were able to make a significant reduction of qubit readout time from the original 3000 nanoseconds to 600 nanoseconds. They have also implemented multi-level qubit resets for much improved qubit reset accuracy and created a high fidelity Controlled-Z (CZ) gate which is used in many quantum algorithms. You can view videos from their recent Quantum Summer Symposium (Day1 and Day2) which includes several talks that describe these improvement and how they will help Google advance to its goals.
July 23, 2021