HPCQS Pioneers Federated Quantum Supercomputing in Europe
The new “High-Performance Computer and Quantum Simulator hybrid” (HPCQS) project aims to develop, deploy, and coordinate a European federated infrastructure, tightly integrating two quantum simulators that each control about 100+ qubits (quantum bits) in the Tier-0 HPC systems Joliot Curie of GENCI, operated at CEA/TGCC, and the JUWELS modular supercomputer at JSC. The seamless integration of quantum hardware with classical computing resources, thus creating a hybrid system, is an essential step towards utilizing the power of quantum computers to handle the first practical applications. At JSC, HPCQS is a logical continuation of the efforts driven by JUNIQ (the Jülich UNified Infrastructure for Quantum computing) to establish tight hybrid quantum HPC simulations.
The European High-Performance Computing Joint Undertaking (EuroHPC JU) has approved the infrastructure project within the framework of its Research & Innovation (R&I) action “Advanced pilots towards the European exascale supercomputers”. HPCQS will receive a total budget of € 12 million over the next four years, which is provided by the EuroHPC JU and the six participating member states in equal measure. HPCQS will carry out a public procurement of innovative solutions (PPI) to implement the two 100+ qubit quantum simulators and launch an ambitious research programme on 1 December 2021. Prof. Kristel Michielsen from JSC, one of the five participating European HPC centres, is coordinating the project.
HPCQS will develop the programming platform for the quantum simulator. It is based on two European software developments: the Atos Quantum Learning Machine (QLM)™ and ParTec’s ParaStation Modulo™. Together, these technologies allow the deep, low-latency integration of quantum simulators into classical modular HPC systems as a first-of-its-kind realization. In addition, HPCQS will work on the deployment of a full hybrid software stack. It will include cloud access, resource management of hybrid workloads, tools and libraries, as well as benchmarking and certification/performance analysis. Engaging users in the co-design process will lead to prototype applications in machine learning and scientific simulations. For further information, please visit https://www.hpcqs.eu.
Contact: Prof. Kristel Michielsen