The high performance computing cloud designed for research is now available to SMEs, on-demand
The University of Cambridge and Imperial College London have announced a partnership that will see their High Performance Computing (HPC) infrastructures combined to create CORE, a world-class cloud supercomputer, available for hire.
CORE is the 93rd most powerful HPC system in the world, and the fastest Intel-based HPC system in the UK, with 22,000 cores, amounting to over 300 teraflops of sustained computational throughout.
It was launched yesterday and is now being offered to SMEs, industry and academia as a pay-per-use service.
CORE also provides complete technical support for its supercomputing cloud, by the same experienced team that manages the universities’ resources.
Getting to the CORE
HPC resources like CORE are really hard to find and the problem isn’t money, but access. Corporations and research facilities usually don’t like the idea of opening up their precious systems to the outside world.
CORE is different. By opening access to its infrastructure, the CORE consortium plans to get more funding to fuel the growth and developement, which in turn will increase system capacity. The HPC service director at Cambridge, Dr. Paul Calleja, calls this “the virtuous circle”.
The service is a key component of the UK e-infrastructure initiative led by the Department of Business, Innovation and Skills.
“Cambridge has been running a pay-per-use model for the last six years. If we combine both HPC services, we are actually serving one of the largest scientific, technological and medical research communities in Europe. That’s a huge number of users, and it gives us an awful lot of experience in doing HPC,” said Calleja.
CORE consists of two components. The first, HPC and Data Cloud, is available for rent, complete with a support team that has 10 years of experience using the systems they are now selling.
The second part is the CORE Consultancy, which sells complete packages for companies which want their own HPC capability. It helps with design, deployment and management of the systems.
CORE even offers starter packages with 10,000 free hours of computing and one day of support, for organisations which aren’t sure they need an HPC system.
And it is environmentally friendly too. “Legacy machine rooms are running with something like PUE [Power Utilisation Efficiency] of about 1.5. That means it costs you 50 percent of the power to cool the system as it does to run it,” Calleja told TechWeekEurope.
“We have machines like that in Cambridge and it costs us approximately £340,000 a year. The new data centre that we are building with the investment from the University will have a PUE of 1.15. It is more efficient to use the new, centralised HPC systems.”
So, who is CORE for? Here’s five examples of companies using it to benefit their business:
- CORE was able to provide CAD workstations to the Caterham F1 team in a very short space of time, while their own infrastructure was being assembled.
- CORE offers a scalable rendering platform for the post-production agency AtomicArt, whenever extra capacity is needed.
- CORE allows Audio Analytic, which sells sound recognition software to the security industry, to process centuries’ worth of sound in hours.
- UK-based high performance storage manufacturer Xyratex is both a supplier and a client, using CORE to develop and test its equipment while the system uses its Clusterstor solution for best-in-class performance.
- OASIS (Open Acess Catastrophe Modelling for the Insurance Sector) uses CORE to “explore uncertainty” and run complex numerical simulations.
“The scalability, low cost and flexibility of the CORE resource gives a start-up like us a real competitive advantage,” said Christopher Mitchell, CEO and founder of Audio Analytic.
Do you know all about Green IT? Take our quiz!