No sooner did the news break that Atos had accepted a £24 million settlement with the UK government over a major supercomputing contract award (HPE’s weather supercomputer for the UK Met Office) than the UK announced it has purchased a second quantum computing simulator from Atos.
The Quantum Learning Machine (QLM) is a quantum simulator – a computer system for simulating the behavior of quantum computers, according to Bidstats, a UK site with information on government contract awards. The system being procured allows simulation of quantum computers up to 38 qubits, and comes with three-year support and maintenance. The contract was awarded by UK Research & Innovation.
The Hartree Centre, a facility near Liverpool that enables UK businesses and the public sector to explore new technologies such as AI and quantum computing, already has one unit, “and is purchasing a second to add to the address resourcing issues with large quantities of users wanting access,” according to a notice on Bidstats.
The Atos QLM uses Atos’ BullSequana X800 servers and generates simulations of quantum computers based on three different quantum programming modes – the gate model, the annealing model, and the analog model. It was launched in 2017.
A focus of the Hartree Centre is its support for industry supercomputing applications, “meaning that businesses can access specialist expertise and HPC that isn’t normally accessible outside of large-scale industries and academia,” according to a story in Data Centre Dynamics. “The center houses the 4 petaflops Scafell Pike Bull Sequana X1000 and an existing Atos Quantum Learning Machine, as well as the Bull Atos Nvidia DGX1 Max-Q system known as Jade 2 for the University of Oxford. It formerly hosted 1.4 teraflops IBM Blue Gene/Q Blue Joule supercomputer, before it was moved to the DiRAC facility at the Institute for Computational Cosmology at Durham University in 2016.”
Atos has released announcements of deals to deliver QLMs to the Galician Supercomputing Center, the Leibniz Supercomputing Centre, and OVHcloud.