Based on Nvidia's G8 series chips and its Cuda software development kit, it will be for workstations and server clusters. Such computing jobs are common in the oil and gas industry, finance, medicine, and scientific research.
The company sells its graphics processing units, or GPUs, for use in consumer electronics, especially gaming computers and portable entertainment devices, and in professional workstations, such as those used in computer-assisted design. In branching out, Nvidia is using the same parallel-processor design it offers in its other product lines. Its chips work alongside a primary CPU and off-load the graphics processing to provide faster rendering and more efficient CPU use.
In building applications for Nvidia's design, a developer writing C or C++ code for the CPU would extract the piece related to the applications' mathematical computations and use Cuda's compiler, so that the component would run on Nvidia's chip, says Andy Keane, general manager of Nvidia's GPU computing.
Nvidia is looking to move into a new market at a time when its graphics processing business faces a threat from Intel and Advanced Micro Devices. Both are incorporating more graphics functionality into their chips. AMD bought Nvidia rival ATI Technologies last fall and is furthest along in the effort. AMD plans to release in 2009 a line of processors that would merge x86 and ATI graphics cores onto one chip. The processors initially would be available for notebooks, then later desktops and workstations.
For now, Nvidia is trying to drum up support for its upcoming product in meetings with circuit-board manufacturers, computer makers, and companies that use high-performance computing systems for petroleum exploration models, financial research, and other computationally intensive areas. In pushing the new use for Nvidia's GPUs, Keane says, "We're the evangelists."