There seems to be a bit of controversy on the web these days about benchmarks that favor a certain card or design philosophy. Admittedly, it's Voodoo fans that complain about 3DMark 2000 and it's composite score which always favors NVIDIA cards. Must be a bias, right?

No. 3D Mark is based on a game engine. I seem to remeber a time when 3dfx used "synthetic" benchmarks to promote their cards. NVIDIA has taken the speed lead. If 3dfx wants to focus on the composite score bias, fine. Instead, look at the individual numbers the benchmark produces. The NVIDIA cards beat the Voodoo cards. Sorry, it's fact.

So why am I ranting? It seems that if you use benchamrks like Quake3 and 3D Mark you are tagged as biased or not thorough, and that may be true in my case (the bias part), but the fact is that the numbers do not lie and Q3A and 3D Mark are the two most demanding benchmarks. If you have a feature like T&L and the other guy does not, too bad. Now, here comes where I am going to change things up a bit.

The GeForce2 Ultra has afforded the gamer a new level of power, pushing games to the next plateau of 1600x1200x32bpp. Unfortunately for the indecisive, it has also opened up FSAA as a viable high performance gaming option. What is a gamer to do?

Most of the time benchmarks are run through with the resolution changed, 16bpp first and then 32bpp, etc. I found that I would jump straight to certain scores that I used as performance landmarks. What I have done is to collect a series of benchmarks landing at those performance points. I will give a screenshot of each and you will get to see what the card is capable of if you are a an eye candy nut, or just want good clean high framerate gaming.

I have set gaming targets. These targets are different for different games. One target is to keep the FPS right at 20 at the slowest moment in demo001 of Quake3 Arena when the gibs are really flying, usually around 45-50fps average. Another is to maintain an average well above the 60fps mark, preferably 70fps.

I am also going to run ALL benchmarks with a conservative but worthwhile overclock. All benchmarks are using a 285MHz core clock and 486MHz memory clock. Why? How many times have you skipped straight to the overclocked benchmarks when reading a review? If you have an Ultra, you are probably not going to be afraid to try and get a little more out of your system.

Here is a quick rundown of the test system from Alienware.

Alienware Area 51:Aurora

I have to take a second to praise Alienware for building an amazing system. Name brand gear in a sweet modified Antec case with 300W powersupply. They were extremely friendly, and responded to questions quickly. An Alienware system will cost a bit more, but with things like the KoolMaxx video card cooling sytem, using a Thermaltake Silver Orb CPU cooler and outstanding color schemes, it's worth it.

You now have info on the system used for benchmarking, so let's take a look at the Fast Trax driver suite.

Next Page: Fast Trax