Published 15 May 2012
Posted by Greg Corke
It may be a workstation by name, but SuperMicro’s specialist high-end SuperWorkstation 7047GR-TRF is really a GPU compute server with high-end professional graphics.
Two Xeon E5-2600 processors, up to 512GB of DDR3 1600MHz ECC registered memory and a high-end Nvidia Quadro GPU is pretty much standard fare for any high-end Sandy Bridge Xeon workstation, but the SuperWorkstation 7047GR-TRF takes things up a notch when it comes to GPU compute.
SuperMicro’s colossus of a machine has space for up to four Nvidia Tesla C2075 GPUs - an incredible amount of dedicated processing power for simultaneous compute intensive tasks such as simulation and rendering.
By comparison, most high-end workstations only support a single Tesla, with Dell’s new Precision T7600 standing out from the Tier One manufacturers for being able to host two Teslas.
The SuperWorkstation 7047GR-TRF is Nvidia Maximus-enabled, which means it is specifically designed to free up CPU resources by offloading the heavy computational work to the Tesla GPUs.
So who would want to buy such a machine? You’d have to be pretty serious about rendering to invest in one of these beasts. Four Tesla cards should do wonderful things to your rendering times in GPU-enabled applications such as Bunkspeed Shot, CATIA Live rendering and 3ds Max Design.
But don’t get too excited if you’re heavily into Finite Element Analysis (FEA) or Computational Fluid Dynamics (CFD). Most GPU-enabled simulation software, including Abaqus and Ansys, can currently only make use of a single Tesla card, though this is likely to change in the future.
Prices start at £4,400, though you’ll need to spend well over £10,000 to get the 7047GR-TRF kitted out with four Teslas. And to be honest, there’s probably not much point in buying this machine if you don’t intend to do that.