NVIDIA has been talking up their new Tesla GPU Computing solutions for HPC this week. NVIDIA has taken their 8800 series graphics cards and positioned them for use by scientists.

NVIDIA has taken three different directions with the Tesla products, the C870 GPU computing processor that can provide over 500 gigaflops of floating point performance. The Tesla S870 GPU computing server designed to fit into enterprise rack mount environments.

The final device is the Tesla Deskside Supercomputer, which connects to a PC via PCI express adapter for an expandable system. Programs for the Tesla systems can be developed in NVIDIA’s CUDA programming language. via NVIDIA