Optimizing resource efficiencies for scalable full-stack quantum computers
The preprint “Optimizing resource efficiencies for scalable full-stack quantum computers” submitted on September 12th, 2022, by Marco Fellous-Asiani, Jing Hao Chai, Yvain Thonnart, Hui Khoon Ng, Robert Whitney and Alexia Auffèves implements the Metric-Noise-Resource (MNR) methodology that helps quantify and optimize various figures of merit of a quantum computer with a full-stack view, including their energy consumption.
The paper showcases what kind of research efforts are promoted by the Quantum Energy Initiative. It consolidates a broad number of disciplines and technologies: quantum physics, quantum thermodynamics, control electronics, wiring, filtering, cryogeny, quantum error correction codes, system design, all in a systemic approach.
One key goal here is to ensure that the energy consumption of a quantum computer is constrained, reasonable and not a showstopper for scalability. The paper describes various findings, some of which may be counter-intuitive, with the example of large scale superconducting qubits based quantum computers, using a detailed full-stack analysis.
- It shows that design choices are twofold: either use room temperature control electronics and find solutions for cabling multiplexing, or, use cryogenic electronics and preferably so-called SFQ, superconducting electronics.
- Control electronics will be the main source of power consumption in scalable superconducting qubits quantum computers, before cryogeny.
- Many significant research and industry efforts will be necessary to enable large scale QC: better qubits, lower energy consuming control electronics, simpler cabling, higher cooling power, better yield cryostats, and efficient error correcting codes
- With NISQ quantum computers (noisy qubits quantum computers), an energy advantage may show up before a computing advantage. This adds to another potential benefit of NISQ systems which could bring better quality results than traditional computing like with quantum machine learning. #QEI
The NMR methodology is applicable to other types of qubits and even other programming paradigms (quantum annealing and quantum simulation).