NEWS
Tags: 

New CFMS supercomputer a sign of things to come at Cray

The Centre for Modelling and Simulation (CFMS) has announced the availability of its new supercomputer at an opening ceremony, held at the end of October.

The new system, which is based on Cray’s cluster supercomputers, will replace the previous CFMS system, increasing performance by a factor of five. However, this system also represents par Cray’s strategy to develop relationships with HPC users in the UK – this activity gained pace in 2015 when Cray opened a UK office in Bristol, close to the CFMS centre.

Announced in June 2016, CFMS awarded the contract for the refresh of its high-performance computing infrastructure to Cray, a global leader in supercomputing. The new system consists of a Cray CS400 cluster supercomputer, utilising the Intel Xeon E5-2600 v4 product family formerly codenamed ‘Broadwell’ with a parallel storage system provided by ArcaStream.

Also included are NVIDIA Tesla Graphical Processing Unit (GPU) accelerators used to accelerate applications, which will provide customers with faster response times and enhanced performance levels for demanding computational tasks.

Dominik Ulmer, VP business operations EMEA at Cray explained the node architecture of the new CFMS system: ‘At CFMS we have four node types, the majority is 128 scale-out nodes you can run a single application across all of these 128 nodes, each node has 2 Broadwell processors from Intel, and each of these nodes has 128 gigabytes of memory.’

In addition to the 128 ‘basic’ nodes, there are also 20 high memory nodes, 32 Ivy-Bridge nodes used for higher frequency processors that can cope with COU intensive tasks and a final eight nodes that include Nvidia K80 GPUs.

Up to five times more powerful than the previous system, industry organisations including Rolls-Royce and Airbus use the CFMS high-performance computing (HPC) cluster for research and methods development, which underpins simulation-based design activities. Undertaking a collaborative role, CFMS acts as an interface between industry, academia and technology organisations, offering access to advanced simulation expertise and technologies that enable companies to accelerate product development and engineering design capability.

Sam Paice, chief operating officer at CFMS: ‘With an evolution in engineering design taking place, we are welcoming a new era, one where advanced simulation takes centre stage, accelerating product development and differentiation while enabling technological advantages and business growth. We are pleased to announce the availability of the new supercomputing cluster at CFMS.’

The Centre for Modelling & Simulation (CFMS) is an independent, not-for-profit organisation that specialises in high-value design capability. The centre promotes advanced modelling and simulation, underpinned by HPC, enabling users to get access to the latest computing technologies.

Ulmer explained that Cray is focused on developing relationships with HPC users in the UK as it is important to participate in the HPC ‘ecosystem’. ‘In supercomputing you cannot just sell a piece of hardware and walk away. It is always about embedding yourself in an ecosystem and building relationships, not just between CFMS and ourselves; we want to expand this further to include members of CFMS, and we can help CFMS to establish ties to our customers in the UK.’

‘One of the reasons that we picked Bristol as a location is because there is a very lively HPC community. We already established a strong collaboration with a local university, and we want to expand this to further. CFMS is one institution that we want to work with closely in the future.’

‘We want to be part of the overall community’ concluded Ulmer.

Twitter icon
Google icon
Del.icio.us icon
Digg icon
LinkedIn icon
Reddit icon
e-mail icon
Analysis and opinion
Feature

Robert Roe looks at research from the University of Alaska that is using HPC to change the way we look at the movement of ice sheets

Feature

Robert Roe talks to cooling experts to find out what innovation lies ahead for HPC users

Analysis and opinion