Nvidia Corp has unveiled its new desktop graphics chip, which it claims will enable the PC platform to compete with the new Sony Playstation. The GeForce 256 chip – which Nvidia is calling a graphics processing unit (GPU) – essentially moves all of the pixel rendering, color, transformation and lighting functions that are currently handled by the CPU onto the new processing unit. Rival company S3 Inc has also done a similar thing with its new chip – the Savage2000 – moving the transformation and light functions onto the graphics engine, although this chip will be targeted at the lower end of the market.

Nvidia claims that the GeForce 256 is the first graphics processor that is more complicated than a main processor. The 23 million transistor chip runs at least 120MHz and features a 256- bit four pipe pixel rendering engine. The company says that the chip will render 480 million pixels per second and draw 15 million polygons per second. Basically, this means that objects and landscapes look more detailed and move more realistically onscreen. The company says that the GPU will offer better performance than the Playstation 2 by the time the console is launched in America, sometime next year.

However, despite the impressive performance statistics, which were backed up by spectacular demonstrations at the Intel Developers Forum in Palm Springs, California yesterday, there are some concerns about the new chip. The chip will be introduced in a variety of formats, however, it will cost more than Nvidia’s current Riva TNT2 family. In addition, the chip runs extremely hot, consuming more power than any other graphics engine so Nvidia won’t be launching a mobile version any time soon.

Nvidia looks likely to use the technology as the basis of a forthcoming workstation product – ‘chip X’. The company has said that it will link with a major workstation vendor to launch the product. Nvidia isn’t giving any names yet but it seems that SGI would be likely candidate, especially as the companies have already signed a technology agreement. The company expects to have the chip on the market by the end of September and in OEM systems by mid-October. The chip is being made on a 0.25 micron design rule but it seems likely that the company will go to 0.18 micron with the next generation of chips, which are due in six months time.