CRN Network, December 22, 2011, 0930 hrs
Boston IT Solutions, the global distribution partner of Super Micro Computer, has unveiled its latest range of GPU and CPU based solutions optimised for high performance computing (HPC).
The new range consists of Nvidia Tesla MD SimCluster, an off-the-shelf, ready-to-deploy solution for HPC that enhances scientific research and discovery.
It is available in a variety of configurations that deliver up to 42 teraflops of HPC performance; and with the SimCluster suite, it delivers an easy way to purchase, deploy and maintain a hybrid GPU and CPU based HPC environment.
Available in three pre-configured hybrid compute configurations, the MD SimCluster claims to cut down simulation time from days to hours. The entry-level SimCluster configuration consists of 8Tesla M2075 GPUs and 4Intel CPUs, providing computational performance of up to 8 teraflops.
A mid-range SimCluster combines 16 Tesla M2090 GPUs and eight Intel CPUs to give a compute performance of up to 20 teraflops. The high-end configuration hosts 32 Tesla M2090 GPUs and 16 Intel CPUs in a 42U rack to provide up to 42 teraflops of processing power.
In collaboration with Calxeda, Boston also unveiled the latest generation of ultra-low power computing platforms based on Calxeda ARM EnergyCore SOC processors. With 48 nodes available in a 2U enclosure, the Boston Viridis Project can provide for upto 900 servers per industry standard 42U rack and can deliver up to 10x the performance per watt over existing processor technologies. |