Big graphics hardware releases can often fall foul of two stumbling blocks – either the card doesn’t live up to the pre-release hype, or the hardware is super fast and exceeds expectations, but there’s no stock available for anyone to buy. Thankfully nVidia made sure that the GeForce 7800 GTX avoided both these pitfalls, and not only is it the fastest graphics card on the planet, it was also available in the retail channel from day one.
So, less than two weeks after the initial launch of the GeForce 7800 GTX, I’m writing this review of the XFX retail card. In fact I got the boards (XFX sent me two for SLi testing) just over a week after the launch, but I wanted to have a good play with them before writing up this review, so I made you wait a little while.
Unless you’ve been hiding in a cave for the past couple of weeks, you’re probably pretty familiar with the 7800 GTX specs, but in case you’re a bear and have just come out of hibernation I’ll give you a bit of background.
The GeForce 7800 GTX is nVidia’s new high-end graphics chipset, replacing the previous king of the hill, the GeForce 6800 Ultra. Now, the 6800 Ultra was a massive step forward when it was released just over a year ago, raising the transistor count to a heady 222 million, but the 7800 GTX smashes that landmark with an unbelievable 302 million transistors! The extra transistor count has helped nVidia raise the pixel pipelines from 16 to 24, while the vertex pipes have also increased from six to eight. nVidia has also opted to stick with the tried and tested 0.11 micron manufacturing process, rather than dropping down to a 90nm model, which could be the reason that yield is high straight off the bat.
The standard core speed of the 7800 GTX is 430MHz, which is an improvement over the standard core speed of the 6800 Ultra, which was 400MHz. However, many board partners clocked their retail 6800 Ultra cards faster than 400MHz, and it looks like the same is going to be true with the 7800 GTX. XFX clearly thinks that the standard specs just aren’t fast enough and has clocked the core of its card at 450MHz. Likewise XFX has bumped up the memory speed slightly, with the 256MB of GDDR3 memory running at 625MHz (1.25GHz effective) instead of the standard 600MHz (1.2GHz effective). Coupled with the 256-bit memory interface, the XFX card should produce memory bandwidth of 40GB/sec compared to the 38.4GB/sec offered by the stock nVidia settings.