Eviden explores energy advantages in near-term quantum computing systems through research partnerships

Paris, France – March 13, 2024 – A Franco-Singaporean collaboration has been announced to benchmark and optimize the energy efficiency of quantum computing. The partnership includes Eviden, the Atos Group business leading in advanced computing, A*STAR’s Institute of High Performance Computing (IHPC) in Singapore, and MajuLab, an international research laboratory in quantum physics. Majulab is a joint laboratory of the Centre National de la Recherche Scientifique (CNRS), Université Côte d’Azur (UCA), Sorbonne University (SU), National University of Singapore (NUS) and Nanyang Technological University (NTU).

In the context of conventional high-performance computing (HPC) environments, energy optimizations are generally achieved through improved hardware architectures and better cooling systems. However new approaches must be set up to optimize energetic consumption in future quantum computers.

Quantum computing leverages fundamental science to solve complex problems much more efficiently than classical methods: this is known as the quantum computational advantage. While the age of large-scale, fault-tolerant quantum computing seems years away, Noisy-Intermediate Scale Quantum (NISQ) devices are already a reality. In such devices, the energy cost to solve a problem on quantum devices could be much less than solving the same problem on a classical HPC system. This offers the possibility that the energy advantage of quantum algorithms may be established before quantum the computational advantage itself.

This research collaboration aims to build a user-friendly framework for accurate benchmarking of energy efficiency in NISQ/near-term quantum computing systems. This framework is based on a novel holistic methodology recently proposed by one of the partners¹, to estimate and optimize energy consumption for the full stack of the quantum computer.

The three partners will conduct research on various options to estimate the performance and energy consumption of various algorithms, supported by Eviden’s quantum emulator Qaptiva 800, which can emulate over 100 qubits depending on the algorithm and emulator used. The collaboration will rely on three main work groups: control parameters and energy benchmarking metric; implementation of resource monitoring within Eviden’s quantum emulation environment; and application-based benchmarking (VQE).