There's been something of a trend in the computer component industry over the last few years with Intel and nVidia ruling the roost in terms of CPU and graphics card performance, respectively. Meanwhile AMD/ATI has gone through a bit of a rough patch but has recently come back strong with some competitively priced products that, while perhaps not the fastest, have proved to be worthwhile investments nonetheless.
The one exception to this rule, however, was the ATI Radeon HD 4870 X2 graphics card that actually proved to be at least equal and generally faster than nVidia's then top-of-the-range card, the GTX 280. The obvious problem for ATI was that it had already needed to meld two graphics chips onto one board in order to get the performance needed to compete with the GTX 280's single chip, which begged the question, "If nVidia put two GTX 280 chips on one card, wouldn't that be faster?". Well, today that's precisely what we're going to find out.
Officially launched yesterday at CES, the GTX 295 is nVidia's latest top-end graphics card and, as that preamble will have suggested to you, it uses two GTX 280 chips working together in a single card (just like SLI but on one card) to theoretically hand the competition its own proverbials. In actual fact, the GTX 295 isn't quite two fully-fledged GTX 280s in so much as the clock speeds of the two chips have been lowered to the same as that of the GTX 260. Basically, the two cores run at 576MHz with the shader clock set to 1,242MHz. Memory speed has also been reduced to 1,998MHz.
As well as these speed reductions the other secret to enabling the GTX 295 to exist (and not end up cooking itself to death) is the shrink in manufacturing process. The normal GTX 280 had been produced using a 65nm process, but the chips used in the GTX 295 use a smaller 55nm process. This seemingly small reduction has a large cumulative affect on the overall power and heat output of the chip, enabling nVidia to fit the same computing power in a smaller thermal envelope than would otherwise be possible.
The card itself takes a radically different approach to that of ATI's 'two chips on one board'. Instead nVidia essentially creates two complete graphics cards in mirror image - with the main Graphics Processing Unit (GPU) chips facing each other - then sandwiches them together using a tasty HeatSink Fan (HSF) filling.
Well, actually, toasty might be a better word to describe it as, despite the manufacturing and clock speed changes nVidia has made, this is still one hot card. No hotter than most other high-end cards, mind, just very, very hot so watch your fingers!
Air is sucked in from holes on both sides of the cards and ducted through a series of angled fins that run up the middle of the card. Half of these lead to a gap that runs along the top of the card while the other half are directed towards a set of holes on the second expansion bay plate.