Nvidia has finally lifted the NDA on the G80 and allowed us to announce its fantastic specifications. The new G80 GPUs mark the arrival of the world’s first DirectX 10 card along with a boatload of new features that puts these new cards heads and shoulders above anything that ATI/AMD offer right now. Before we embark on our exploration of the new core let us quickly run down through some of the basic features that have changed or been introduced.
Clock speed and Physical Dimensions
The GeForce 8800 GTX is clocked at 575 Mhz-core clock and 900 MHz -memory speed, which runs on a 384 Bit Bus. The new 'Stream Processors' (we will just talk about that in a minute) run at a speed of 1350 MHz. The RAM being used is still GDDR3. Overall this may seem a tad slow but this is not where all the optimization has taken place. The card’s new unified shader design now pushes nearly 86.4 GB of memory, which is seriously a lot of memory bandwidth.
Physically the card is a monster. It is extremely long and quite easily dwarfs its younger brother—the GeForce 7950 GX2. It is so big now that the power connectors for the card have been moved to the top of the card from its traditional side position. It's PCB is a beautiful black, which is something you do not see everyday. The card’s HSF (Heatsink Fan) is huge but surprisingly runs quite silently. The card has two DVI connectors and surprisingly two SLI connectors, which we believe maybe a move for future readiness. In terms of power requirements, the card with its heavier power requirements will now need two PCIe power connectors and Nvidia recommends that this card be run with nothing less than a 450W PSU.

