Xtreme Compute Technologies this week said that it worked with informatics consulting firm Scalable Informatics to deploy a graphical processing unit-based cluster at Stanford University for initiatives including computational biology, molecular dynamics, and structural biology and chemical biology simulation.
XCT said that Stanford used its a-BriX servers, built with NVIDIA Tesla GPUs, as part of a new 50-node heterogeneous cluster called Numbercruncher. The system includes 100 Tesla C2050 GPUs, XCT said.
The company said that it worked with Vijay Pande, associate professor of chemistry, structural biology, and computer science at Stanford, to define the functional requirements and objectives for the cluster.
Pande, perhaps best known for leading the [email protected] protein-folding project, has long been a proponent of GPU-based biological computing. In 2009, he and his colleagues released OpenMM, a GPU-enabled implementation of the GROMACS molecular-dynamics package (BI 2/20/2009).
In addition to the a-BriX Tesla GPU servers, the cluster uses host servers based on AMD Magny-Cours multicore processors, Delta-V storage, and Bright Computing's Bright Cluster Manager software and Accelereyes' Jacket and GPU Engine software.