Imperial College London and the University of Cambridge have pooled their computing resources to offer a high-performance computing (HPC) cloud available on-demand to both academic and industry customers, as well as offering expertise on developing and operating HPC infrastructure.
CORE is part of the UK Government's Department for Business, Innovation and Skills (BIS) infrastructure expansion programme, and utilises over £8m worth of installed equipment. This includes over 22,000 Intel processor cores and is capable of 300 teraflops of sustained performance, plus over 3PB of storage between the two institutions.
The goal of CORE is to provide a competitive advantage for UK academics and industry, taking the expertise of the two world-leading computing facilities and making them available to partners of any size.
It is also hoped that it will open up public-sector infrastructure, generating growth in the UK which will in turn lead to greater investment in infrastructure, forming a "virtuous circle", said Paul Calleja, CORE director at the University of Cambridge.
In particular, CORE will remove the barrier to entry into HPC for smaller organisations by offering a pay-per-use model, whereas such firms often find HPC facilities beyond the reach of their budgets.
"CORE is more than just about being good academically, it's about working with industry, and providing a service for industry," said Lyn Gladden, pro vice chancellor for research at the University of Cambridge.
The infrastructure is already being used by some customers for modelling and large data analysis, including one customer, post-production firm Atomic Arts, which has used the facilities for digital imaging processing.
However, CORE is not a cloud like that provided by operators such as Amazon, which are largely based on spinning up virtual machine instances, and instead provides remote resource provisioning based on a high level of technical support and assistance from the CORE team.
In fact, the systems that make up CORE are not fully interconnected to make a single giant compute resource, as is the case with a typical cloud service.
The CORE cloud comprises several systems that are already among the most impressive in the country, including the UK's highest performance Intel-based cluster; one of the largest Nvidia GPU clusters; plus the largest single-image shared memory computer system in the UK, consisting of 2,000 Intel cores with a single 16TB memory space.
As well as providing the compute resources, CORE is also offering a set of consultancy packages to companies wanting to build and operate their own HPC infrastructure.
These services include HPC design, procurement and development help, profiling of HPC requirements, troubleshooting, and even scientific support in the form of post-graduates to help with scientific expertise.
CORE access is available on a trial basis for new customers, with a free starter pack providing 10,000 core hours and one day of free support.
More information on the service is available at the CORE website.
Politicians in attendance complained that Zuckerberg skipped all the tough questions
Are you paying attention?
Private equity firm Permira only acquired Magento from eBay for $200m three years ago
Before robots can take over from humans, we need more humans