Introduction

NVIDIA's GeForce GTX 480 (codenamed Fermi) has been one of the hottest forum topics of the last months. There was drama from all sides, NVIDIA users speculating about the nearing end of ATI, ATI users comforting themselves in words like yield, heat and power. The public release of GeForce GTX 480 last month shed light on many things: yes Fermi is fast, yes it also needs a lot of juice which makes it hotter and noisier compared to AMD's offerings. When the cards were first announced there was no immediate availability, and shelf dates were expected to be mid-April. In the recent days we have seen the first merchants listing availability of the GeForce GTX 480, and shortly thereafter "out of stock". We were lucky enough to receive a GTX 480 production card from ZOTAC for an SLI review.



Both retail and press samples are mostly identical, the only exception being a little piece of foam on the back of the retail GTX 480 boards that helps keeping multiple cards apart in SLI configurations. Also there is a newer VGA BIOS that slightly changes the fan settings. According to NVIDIA the fan will ramp up more quickly when the card is cool (~70°C) so that the maximum temperature is reduced. In terms of performance nothing has changed, no changes in the the new 197.41 driver either.



In order to run a GeForce GTX 480 SLI setup you need a motherboard that supports SLI- most Intel based motherboards today support it. Depending on the number of cards you plan to run you also need the proper number of PCI-Express slots available with enough spacing between them. Most motherboards have no problems accommodating two cards. In order to run three or four GTX 480 cards you usually have to buy a motherboard specifically designed for such setups. Four cards? Yes, you read right. The GeForce GTX 480 is the first NVIDIA card to support actual 4-way configurations across all vendors.

In this review we will take a look at a dual-card GeForce GTX 480 setup which should provide enough power for all games at any resolution and detail level.