Page 6

Benchmarking

For this test I decided to drop the synthetic benchmarks. They mean absolutely nothing. 3DMark 06 for example can't apply AA with HDR and so on. This is all about gaming, so well use some pretty new titles.

A game's frames per second (FPS) is a measured average of a series of tests. That test often is a timedemo, a recorded part of the game that is a 1:1 representation of the actual game(play). After forcing the same image quality settings this timedemo is then used for all graphics cards so that the actual measuring is as objective as can be. If a card can only manage less than 30 FPS then the game is barely playable.

With 30 FPS up to roughly 40 FPS you'll be very able to play the game with perhaps a tiny stutter at certain graphically intensive parts.

When a graphics card is doing 60 FPS on average or higher then you can rest assured that the game will likely play extremely smoothly at every point in the game.

You are always aiming for the highest possible FPS, versus the highest resolution, versus the highest image quality.

What you also need to understand is that with each game tested we enabled every possible setting that could improve image quality, so we made sure that Quad SLI is brought down to its knees as best as we can.

Battlefield 2 - EA

The award-winning Battlefield franchise invades the high tech frontlines of modern warfare in the new sequel, Battlefield 2. The game brings the intensity and excitement of Battlefield 1942 into the modern era with enhanced team play and the latest, most technologically advanced vehicles and weapons systems available to man.
 

This is probably one of my favorite games ever. I love what it can do in terms of gameplay and image quality. It's also jackpot on the first run. We had to use FRAPS to measure the performance and by command prompt remove a nasty framerate cap. The the results do not lie.

BTW the game by itself does not support this kind of resolution, but simply make a little .bat file and insert the following:

bf2 +menu 1 +szx 2560 +szy 1600 +fullscreen 1

The charts: red colored bars equal the Quad SLI configuration with several AA settings, blue colored bars are a single GeForce 7950 GX2 at either 4xAA or 8xAA.

Now your resolution will be forced to what you filled in. When you look at 2560x1600 you'll notice that 8xSLI-AA is slightly slower than a single GX2 card in the same resolution at 4xAA. If we compare directly and set both cards at 8xAA then the Quad SLI setup is very playable with 50 FPS where the single GX2 would do 30 FPS.

That's really impressive.

Splinter Cell: Chaos Theory

Sam Fisher returns for his third installment. Splinter Cell: Chaos Theory, the third game in the acclaimed Splinter Cell series, manages to improve the games visuals, make the gameplay a bit more nonlinear and adds some new gameplay modes to the already exhaustive Splinter Cell brand. Anyone who has seen Chaos Theory in action can attest to its visual masterpiece. Dynamic lighting is back in a big way. No longer are shadows blobby, elongated representations of the characters. Now we have shadows that are detailed and exact.
Another of the biggest renovations of the graphics is the amazing use of bump and normal mapping. Now when you are sulking around in the shadows of espionage Sam actually has a recognizable face, with expressions and features that look real. Rather than the flat textured faces we have seen in the games previous.

The game is so darn good.  

We cannot apply AA in this game. This is one of the most hurtful limitations for Quad SLI. Therefore this Quad SLI rig versus one GX2 equals the same performance. This is something you need to remember, Quad SLI is all about high AA levels (at this point in time). I am deliberately including results like these as these are the facts. On a good number of games Quad SLI will do absolutely nothing as it's all about Antialiasing.

It's still extreme good fun at 45 FPS @ 2560x1600.