Country missing? Please select your nearest region...
2 July 2014 by Nick Booth - -
High performance computer (HPC) maker SGI has installed three SGI ICE X systems to the UK Atomic Weapons Establishment (AWE) which are now fully operational.
AWE, which makes the warheads for the UK's Trident nuclear deterrent, is using the computers to make testing safer through research and theoretical modeling of situations.
The systems use Intel Xeon E5-2680v2 10C processors with 2.8GHz capacity and run on Suse Linux Enterprise Server 11. With 139,000 GB of memory it is capable of 997.248 Terra Flops (floating point operations) per second which consumes 855.10 kW in power.
The configuration of the data center and amount of information stored and processed is classified information, according to a spokesperson.
The data center is cooled using a closed-loop airflow system with integrated hot aisle containment.
It also supports warm water cooling and can accept a broad range of temperatures, which helps to cut costs.
Other features include both air-to-water and liquid-to-water heat exchangers.
The system is used to make the entire life cycle of Trident’s nuclear warheads safer, using high quality simulations and 3D modeling. Under the Comprehensive Nuclear Test Ban Treaty any test that produces any nuclear yield underground, underwater, in the atmosphere or in space, is prohibited.
The supercomputers are used to help model probabilities and create simulations that make all steps of the life cycle safer.
The three systems will be number crunching simulations on every stage from initial concept, assessment and design through component manufacture and assembly, in-service support, decommissioning and then disposal.
The HPC is also used to study forensic seismology and contribute to nuclear intelligence.
The data center underpins the safety and effectiveness of the Trident warhead in the comprehensive test ban treaty era, said Paul Tomlinson, head of high performance computing at AWE.
“Everything has to be tested, studied and monitored without physically trialling anything which produces a nuclear yield,” a spokesperson said.
“Working in this way generates incredible volumes of data and an HPC system was required to enable teams to efficiently manage and use it.”