SGI has unveiled its new Altix ICE high-performance computing (HPC) platform that leverages a tightly integrated, cool-running blade solution. The company achieved density improvements by implementing a new, highly integrated version of the Atoka board, which SGI co-designed with Intel. The next-generation board, designed specifically for the HPC market, allows a single SGI Altix ICE 8200 blade to be powered by two Dual- or Quad-Core Intel Xeon processors, with up to 32GB of memory.
Altix ICE supports large scale-out environments while minimizing the demands on the data center’s space and power. Its ultra-dense rack architecture delivers up to 40 percent more compute performance per floor tile than competing blades, according to SGI. A single rack can be powered by as many as 512 Intel Xeon processors and deliver 6 TFLOPS of performance.
SGI Altix ICE systems have already been installed at several sites, including the University of Exeter in the UK and General Atomics in San Diego.
The new platform’s high compute density features an energy-smart design that can save organizations up to $53,000 in annual energy costs for a 10 TFLOP system. It is delivered pre-integrated for easy deployment.
The new platform features support for dual high-speed, low-latency Double Data Rate (DDR) InfiniBand backplanes, enabling fast communication between nodes and racks, even as deployments scale to thousands of processors. By dedicating a Gigabit Ethernet network for administrative tasks, SGI has freed up the 20Gb/s DDR InfiniBand connection for compute traffic. For optimal dynamic communication, the new system also uses a hierarchical network topology that requires no external switching.
For more information, visit www.sgi.com/products/servers/altix/ice.
The latest product and technology information
Future US's leading brands bring the most important, up-to-date information right to your inbox
Thank you for signing up to TV Tech. You will receive a verification email shortly.
There was a problem. Please refresh the page and try again.