Using the latest version of EA Games Battlefield 1942 we used Fraps to run a demo. The 6800 Ultra's performance is off just a hair and in fact is most likely limited by the CPU. The game plays very smooth on either Card.

Counter Strike Source has a built in benchmark. Again we pushed the Cards to 1600x1200 4x/8x and recorded the results. The Nvidia Card needed to have AA and AF applied in the driver Control Panel, the in-game settings do not apply for some reason.

Half Life 2 required us to move from the Forceware 71.20's to the 71.80's while using the eVGA 6800 Ultra to fix a bug with the water and some missing texture problems we saw in certain areas of the game. Using the in game AA feature it would not work with one of the Video Cards, we had to force it in the Control Panel with the 6800 Ultra. We used the HOC Half-Life 2 Benchmark to run these tests. We had no problems with the Catalyst drivers. Even though Half Life 2 was designed for the ATI cards, the 6800 Ultra performed within a few frames - keep in mind the X800XT PE is running at 520 Mhz core speed while the 6800 Ultra is running at 425 Mhz. The performance difference was negligible and the game played smooth on both Video Cards.

As we all know Doom III was written to run on the Nvidia hardware and I think this test proves that. The eVGA 6800 Ultra just slaughtered the X800XT PE at 1280x1024 with 4xAA enabled. The game runs fine on both Cards with the nod going to the 6800 Ultra. I find that using Anti-Aliasing in Doom III just isn't necessary since the game is so dark - without AA the X800XT PE scores within a few frames of the 6800 Ultra. We used the built in demo feature (timedemo demo1 from the console) to run these tests.

Using the FarCry Benchmarking Utility we again pushed both Cards and again we had issues with the 6800 Ultra. Using the in game AA setting there was no Anti-Aliasing. We had to force AA and AF in the 71.80 driver control panel for it to work. The game ran great on both pieces of hardware, however for smoother gaming I'd suggest dropping to 1280x1024 with 4xAA/8xAF.

For GTA VC testing we again used Fraps for scoring. The eVGA 6800 Ultra played very well however the framerates were never quite as high as the X800XT PE was able to achieve. Again, both Cards delivered a more than satisfying gaming experience.

Halo 1.05 uses a series of scripted events as a timedemo (add -timedemo to desktop shortcut). This is not entirely indicative of real world testing. While using the eVGA 6800 Ultra we had to drop to 71.20 drivers, the 71.80s have an issue and there are many missing textures. This benchmark was run without Anti-Aliasing since it doesn't work properly in Halo. Both Cards delivered a more than satisfying gaming experience with the 6800 Ultra slightly pulling ahead.

For our Madden 2005 testing we again used Fraps to benchmark. Madden 2005 has an framerate cap of 60 frames, so the scores basically rely on the dips in performance. Both Cards played Madden 2005 with ease, again the X800XT PE maintained a higher average by having less dips in performance.

Quake 3 1.29h once again shows NVidia's superiority in OpenGL. ATI has made strides in this area claiming a full OpenGL driver rewrite is in the works - however it still fails to deliver performance on par with Nvidia. Both Cards had no problems running the game, however the eVGA 6800 Ultra pulls an extra29 frames over the X800XT PE Card.

MS Rallisport Challenege is still a popular game and still commonly played. We can see that the performance in this game is just not up to par with the Nvidia drivers. Being that this is an older game I wouldn't expect a performance boost anytime soon, however this is one of the "Way it's meant to be played" games - I would think Nvidia would want to push performance in games with this logo on it. The ATI X800XT PE delivered a much better gaming experience here. The eVGA 6800 Ultra was not smooth, was choppy and overall performance was terrible for a racer. Dropping to 1280x1024 the 6800 Ultra was much more playable.

Using Tiger Woods 2005 and Fraps we took a reading at the teebox with the graphics on the maximum settings. The graphics are phenomenal and with swaying grass and trees it's like being on the links without having to actually carry your bag. Both cards were very playable, using the older Forceware 6x.xx drivers the performance was half of what the 7x.xx series Forceware drivers delivered.

Our last benchmark is Unreal Tournament 2003. Again, using the built in benchmark feature we pushed the cards as far as the benchmark would allow us.

Next Page - Conclusion and Final Words