Overview

What is GPGPU?

GPGPU technology bundles multiple GPUs within an integrated system, either on board or through deskside/rackmount add-on modules, and uses the massively parallel architecture of the graphics processor for general-purpose computing tasks. Because of that parallelism, single-precision floating-point operations can speed up dramatically.

Applications such as molecular modeling, weather modeling, and geophysical research can immensely benefit from GPGPU technology. At Supercomputing 2007 in Reno, NV Western Scientific demonstrated a desktop workstation with three nVidia 8800GTX GPUs running NAMD molecular dynamics simulation [http://www.ks.uiuc.edu/Research/gpu]. This version of NAMD was ported to utilize two of the nVidia 8800GTX cards as GPGPU resources while the third 8800GTX was used for high-resolution, accelerated VMD visualization. Running a simulation that moves a glycerol molecule through the cell membrane of an E-coli virus the system achieved sustained performance of 691 GFLOPs.

GPGPU System Options

Option 1: FusionGP™

Western Scientific offers a fully integrated GPGPU system in the FusionGP™. With up to 3 nVIDIA Tesla GPUs, the FusionGP™ can achieve computational speeds of up to 1.55 TFLOPs. It's essentially a supercomputing machine at a fraction of the cost and size. The FusionGP™ is available as a desktop workstation or a 4U rackmount server.

FusionGP product page »

Option 2: Deskside or Rackmount Add-on

Already have an existing system or workstation? No worries. Western Scientific can enhance your system by linking it to GPGPU add-on modules. Available as a deskside box or series of 1U rackmount modules, these self-powered systems can easily be integrated into your system.

GPGPU add-on product page »