The known universe for microprocessors has revolved comfortably around AMD and Intel for many years, at least as far as the enterprise is concerned. But that may be about to change.
It seems that more and more enterprises are turning on to the raw number-crunching capabilities of high-performance Nvidia's graphics processors. This profile of New York software developer Hanweck Associates offers a view of how the Tesla processor can be used to sort mortgage and financial data with a single PC, a process that normally takes a rack full of servers. Applications like these are becoming easier to create thanks to new technologies like CUDA, which allows graphics processors to understand C language and other developer-friendly environments.
Rumors are also flying that Nvidia is looking to join the x86 fray through the acquisition of all or part of a California company called Via. While both firms deny talk of a possible merger, the combination of AMD and ATI at least establishes a precedent of how graphics and raw processing can come together. In fact, embedded graphics processing is one of the key ways that both AMD and Intel plan to broaden their processor lines.
Branching into the x86 market also makes sense for Nvidia considering that AMD and Intel are aggressively pursuing both the graphics and video market. AMD is nearing the release of the Radeon HD 3830 video card to compete directly with the 8600GTS, and is also prepping a new HD 3800 series to take on the GeForce 9600 GT.
Meanwhile Intel already sells more graphics chips for PCs than Nvidia, although they don't quite measure up performance-wise, which is something Intel would like to change. The company is working on an advanced standalone graphics chip called Larrabee that not only will support DirecX and OpenGL, but will be based on the same Intel Architecture that developers are already familiar with.
Clearly, the Nvidia brass has some decisions to make. If they decide wrongly, or fail to act altogether, they just may find themselves as the acquisition target before too long.