Towards an application-oriented measurement of quantum hardware performances
par Xavier Geoffret – Quantum Computing pre-sale leader, Atos
Research on Quantum Computing hardware is developing very fast over numerous technological approaches. Over the past few months, there have been frequent announcements stating major performance breakthroughs for superconducting qubits and trapped-ion-based platforms, but many other paths are being explored such as cold atoms, photonics or even Majorana fermions.
New milestones achieved, but no clear winner
Players are communicating on technological progress with regards to stability (resilience to quantum noise), scalability (number of qubits) and connectivity (ability to entangle qubits in dense topologies). Improvements of these technical characteristics can be measured using the Quantum Volume, a benchmark method on which most quantum hardware providers have been relying so far. In August 2020, IBM announced they achieved a 64 Quantum Volume with 27 superconducting qubits, closely followed by Honeywell’s announcements of a Quantum Volume of 128 using 7 trapped ions in September 2020. Only one month later, IonQ achieved a 4-million Quantum Volume on a 32 trapped-ion chip, showing the limits of this metrics in terms of practical interpretation.
All of these technological paths are of interest and as of today, there is no clear winner in the race to quantum supremacy, which will consist in solving a real-life problem that is intractable to the most powerful supercomputers using quantum computing.
Atos – A hardware-agnostic approach
In 2016, Atos launched its Atos Quantum initiative to foster the adoption quantum computing without betting upfront on a specific technological approach. Atos’ strategy consists in five pillars, which range from providing a quantum programming and simulation platform – the Atos Quantum Learning Machine – and associated expert services to investing in research on quantum-safe cryptography. This hardware-agnostic positioning provides Atos with a unique observation spot for the evolution of quantum hardware technologies.
Measuring the quality of quantum hardware using meaningful and objective metrics has been a major concern for Atos from the start. In order to keep the link between programming and actual hardware constraints, several optimizers were added to the Atos QLM stack as well as the possibility to include noise models in quantum simulations. We also developed a tomography module which allows hardware providers to infer noise models from their quantum chip. These tools are aimed at accelerating research on both hardware platforms and algorithms.
As quantum computing’s TRL (technical readiness level) increases, hardware characteristics such as stability, scalability and connectivity are becoming insufficient to answer industrial end-users’ growing concern: what are the performances of these technologies in solving real-life problems? They are now looking for an efficient way to compare quantum technological approaches.
This is why Atos has decided to make the tools and methodology, initially developed for its internal needs, freely available to potential users of quantum technologies. These tools and methodology, known as Q-Score, allow any user of quantum technology to evaluate the performance of that technology on their own.
One of the main interests of Q-Score is to provide a performance index that does not focus on hardware characteristics, but on the performance of an application that is representative of real-world issues. By design, Q-Score does not favor any quantum circuit technology and focus on NISQ processors.
 Noisy Intermediate-Scale Quantum refers to the current generation of quantum processors with limited stability and scalability.