ATI Radeon X1800 XT (512MB) review

There are also a number of deinterlacing enhancements for improving video image quality. You'll never see as refined an image when you convert an analogue video signal through a digital processor as you would with just a straight analogue signal (like from your set-top box to your television), but more and more people are combining their PCs with their home entertainment systems. Seeing as you will probably be watching at least some video on your PC, any aids to image quality are welcome, and ATI's Radeon and Avivo currently offer the most advanced set of features.

If you have video inclinations, you'll especially appreciate ATI's dual digital-video outputs. You can expect that ATI's various board partners will mix and match outputs as always, so you may find some cards using the same chip with one analogue CRT connector and one digital LCD output. But at least the raw capability is there. You'll also find an S-Video jack for outputting to a television.

Performance
Although we aren't thrilled with its design and wish that it brought more to the table than simply catching up with Nvidia's feature set, the ATI Radeon X1800 XT's performance is hard to knock, at least for certain types of games. The results remain consistent with last generation's conclusion that ATI fares better at Direct3D-based Half-Life 2 and Nvidia dominates at the OpenGL 3D API and Doom 3. The difference this time around isn't as dramatic as it was with last generation's cards, which makes it hard to declare one the outright performance winner.

Conventional wisdom was that with more games coded in Direct3D than OpenGL, you'd rather have superior Direct3D performance if forced to choose. That division has become muddied -- a number of upcoming high-profile titles will use the OpenGL-based Doom 3 engine (the forthcoming Quake 4 and Quake Wars, in particular). If those games live up to expectations, we expect both will get major play among hard-core gamers -- those most likely to spend upward of £350 on a 3D card. Thus, while fast Direct3D capability remains crucial, OpenGL prowess has gained in importance.

Our colleagues at GameSpot kindly provided us with test results this time around, and the scores are very near where we imagined they would be. Half-Life 2 results show that the ATI Radeon X1800 XT is the fastest single-card solution for this game, especially at high resolutions. At 1,600x1,200 pixels, the ATI Radeon X1800 XT was 22 per cent faster than the Nvidia GeForce 7800 GTX card, a remarkable difference. Don't let the disparity in memory between the two cards fool you, either. The textures in Half-Life 2 currently aren't large enough to take advantage of the Radeon's 512MB allotment. You can bet that in the near future you will want that much memory, so we're glad the Radeon has it. But for now, it's sheer horsepower that gives ATI's card the edge here.

As expected, the Doom 3 test reveals another story. If you thought the 22 per cent difference in Half-Life 2 was substantial, here, the ATI card loses to the GeForce 7800 GTX by 26 per cent at 1,600x1,200 pixels. Worse, the £350 Radeon trailed the £250 GeForce 7800 GT card on all of the Doom 3 tests. The disparity between the two cards at each task suggests that now, more than ever, it's important for you to pick your next graphics card based on the games you want to play, an unfortunate trend. Ultimately, though, we have to leave the single-card performance edge with ATI, for the sheer volume of Direct3D games compared to those coded in OpenGL.

From a value standpoint, we also thought it would also be interesting to compare two sets of dual-card setups to the ATI Radeon X1800 XT. The pair of 256MB PNY Verto GeForce 7800 GT cards in SLI mode dominated on the tests, as we expected them to, by a wide margin at every resolution. You'll pay £500 for those, though, not counting an SLI-compatible motherboard. The pair of 256MB Radeon X850 XT cards in CrossFire mode also defeated the Radeon X1800 XT but by a much slimmer margin (although the CrossFire test bed had a slightly slower CPU compared to the others).

We wouldn't expect a dramatic performance increase if we'd bumped that test bed to an Athlon 64 FX 57 from the Athlon 64 X2 4800+ we had installed on it, but for the £400 or so you'd pay for the CrossFire setup, you get relatively little performance gain over the high-end single cards.

Edited by Matthew Elliott
Additional editing by Nick Hide