Nvidia introduced us to DirectX 10 and Unified Shaders with its superb GeForce 8800 graphics chip, which has been used in a range of GeForce 8800 GTS, GTX and Ultra graphics cards. The original G80 chip uses a 90nm fabrication process but Nvidia's latest G92 silicon has moved to a 65nm process. This is great for Nvidia as it gets more chips from each wafer of silicon, but what's in it for the customer?

Reducing the size of each transistor means that Nvidia has been able to add features while making the chip smaller and cheaper. The transistor count has increased from 681 million to 754 million, thanks to the addition of a VP2 video engine and bitstream processor to handle HD movie playback without putting a strain on your CPU.

There are two more new features, with a move from PCI Express 1.1 to the faster PCI Express 2.0 bus, plus the adoption of HDCP over dual link DVI, which supports displays that have a resolution as high as 2,560 x 1,600 pixels.

Both GT and GTS use a six-pin PCIe power connector and have a single SLI connector on the top edge of the card.

The curious thing about the new G92 chips is that they fit around the existing G80 products, so the new GeForce 8800 GT is positioned between the current GTS and GTX models. The GT is a new model but there's also a new GeForce 8800 GTS which is very similar to the existing GTS, but there's an easy way to work out whether the chip you're getting is old or new, as we'll explain.

Click the 'NEXT' link below to find out more.