So, if youâ€™ve read this far, youâ€™re probably wondering when Iâ€™m going to talk about performance. Well, as far as performance goes there is one thing thatâ€™s true for both the Radeon X800 XT and the GeForce 6800 Ultra â€“ youâ€™re going to have a very tough time finding a CPU fast enough to make the most of these cards.
To say that ATI and nVidia have taken graphics to the next level with their latest chipsets is an understatement of epic proportions. ATI claims that the Radeon X800 XT is twice as fast as the Radeon 9800XT, which is a pretty amazing performance jump by anyoneâ€™s standards. The only problem is that running a Radeon X800 XT in almost any PC available today, means that your graphics performance is limited by the speed of the CPU and not by the ability of the graphics card.
We tested the X800 XT using a 2.6GHz Pentium 4 Northwood HT chip and using a 3.2GHz Pentium 4 Prescott processor. The former represents the average home system, where someone might want to upgrade their graphics card to get a bit more games performance. The latter represents a reasonably high-end system, but not a super-powerful setup where money is no object.
The 2.6GHz machine is exactly the same system that we used to test the nVidia GeForce 6800 Ultra, so some of the scores are directly comparable. I say some of the scores because we only had the 6800 Ultra for a short while, so we could only run a limited number of tests.
Looking at 3DMark 2001 SE the X800 XT beats the 6800 Ultra at 1,024 x 768 with no FSAA or AF. Pushing things up to 1,280 x 1,024 with 4x FSAA and 4x AF saw the 6800 Ultra edge ahead, while at 1,600 x 1,200 with the same features on, the X800 XT came out on top.
However, running AquaMark3 the 6800 Ultra managed to stay ahead of the X800 XT until we pushed things up to 1,600 x 1,200 with 4x FSAA and 4x AF, where the ATI card nosed ahead by a couple of frames per second.
Using Tomb Raider: Angel of Darkness, at 1,024 x 768, the 6800 Ultra again edged ahead, but only by 2fps. While running X2 The Threat the numbers are equally close.
Ultimately though, all these figures are not really showing the true performance of each card since the CPU just canâ€™t cope with the speed of these graphics chips.
We ran both an X800 XT and an X800 Pro on the 3.2GHz Prescott system, and the results showed that the faster CPU does make a significant difference. If you look down the graphs on the following pages you can see differences in the scores that equate reasonably to the differences in specification between the XT and Pro cards.
However, if you look at the graphs for the Far Cry benchmark, youâ€™ll see that even running a 3.2GHz CPU is severely limiting the performance of the XT card. Looking at Far Cry on the X800 Pro shows a steady drop in performance as features are turned on and resolutions are raised. But looking at the performance for the XT card shows that all the tests at both 1,024 x 768 and 1,280 x 1,024 produced almost identical results. This means that the card is completely CPU limited, and itâ€™s only when the resolution is pushed up to 1,600 x 1,200 that you can see any deviance in the numbers.
Strangely, when we first ran Far Cry, there were some severe texturing problems, where dry land transformed magically into water at random moments. However, while I was waiting for ATI to get back to me with a fix for this issue, it cured itself with no intervention necessary. Iâ€™m not sure what caused this issue, but I am assuming that it wonâ€™t be rearing its head when production boards hit the street.