SCIENCE

Customers in the Private Sector, Engineering, Government and Research Institutes Adopt vSMP Foundation as the Affordable, Software Alternative to Proprietary SMP Systems

ScaleMP has announced record revenues and sales of its vSMP Foundation aggregation platform with both more than tripling for the fourth quarter of 2010, compared to the same quarter last year.  In addition, the company announced the delivery of product to its 200th customer. ScaleMP made particularly strong gains in the healthcare and life sciences, manufacturing and education sectors. Additionally, the company announced that over 20 percent of its 2010 sales were generated by repeat customers, including Massachusetts General Hospital, San Diego Supercomputer Center and the University of Nice Sophia Antipolis.

“2010 has been a year of breaking records for ScaleMP,” said Shai Fultheim, founder and CEO of ScaleMP. “We released the latest generation of vSMP Foundation which provides the highest shared-memory capacity (up to 64TB in one system), the highest compute capacity (up to 16,384 CPUs in one system) and the best performance of any product on the market, with more than double the performance for 2-socket systems and four times better performance for 4-socket systems. We anticipate even greater adoption as customers see the cost and performance advantages of gaining a virtual SMP at cluster prices.”

Repeat customers expanding their deployments include RWTH Aachen University in Germany, a ScaleMP customer since 2007, that just announced the deployment of a 512-core virtual shared-memory system using 64 sockets of Intel Nehalem-EX processors (8-cores each).  This new system will be used for general purpose research and will have 4TB of shared memory.  The system is expected to go into production in the second quarter of 2011. 

“Our research requires a shared-memory programming model, such as OpenMP, and a computer system that can scale to the highest degree.  vSMP Foundation allows us to achieve a high level of scaling, coupled with the flexibilities and cost advantages of a software platform,” said Dieter an Mey, head of the HPC Team at RWTH Aachen's Center for Computing and Communication. “We want to increase the productivity of our scientists by providing a platform for shared-memory programming for applications which cannot easily be mapped onto a standard cluster configuration. Particularly for adaptive and irregular algorithms, the virtual-SMP approach offers a highly interesting solution. One focus of our research will be the analysis of simulation output data and pre-processing for displaying simulation results in a virtual reality environment.”

ScaleMP has also recently received significant industry recognition including, being named a Network Products Guide New Technologies and Solutions finalist.