Called "Maximus", the technology combines Quadro workstation graphics cards with Tesla C2075 companion cards to provide hardware-accelerated computing environments. Nvidia said that applications that support Maximus are offered by Adobe, ANSYS, Autodesk, Bunkspeed, Dassault Systèmes and MathWorks. Prices start at about $200 for an entry-level Quadro 600 card and $2500 for the Tesla board.
According to Nvidia, the C2075 comes with 448 processing cores and 6 GB of GDDR5 memory, which are good for 515 double-precision GFlops. A Quadro 600 card leverages 96 cores, while the high-end $3500 6000 model, which closely matches the specs of the C2075, runs 448.
Maximus appears to be the most aggressive attempt by Nvidia to drive Tesla deeper into workstation computing since the launch of the original Tesla products and the D870 deskside accelerator in 2007. Back then, Nvidia offered two Tesla GPUs that offered 1 TFlops of single precision floating point performance. Prices then started at $7500.