I am a gamer, its how I got involved with Quakenet. I suspect the vast majority of quakenet users are also gamers, some more hardcore than others. Now, games require hardware to run, decent hardware to run well. A major hate of mine is all the bullshit benchmarks and hardware review sites. The proliferation of these sites has got to ridiculous levels, the quality of content is in the main poor and potentially misleading.
Benchmarks are the basis for most hardware review sites. Now its a 'cool' thing to say that most benchmarks are useless, benchmarks have their value provided you understand they are not perfect interpretations of everyday situations. Benchmarks are inherently biased. This in itself is not a bad thing, however the interpretations of these benchmarks leaves a lot to be desired.
Benchmark graphs that have absolute values (ie have origins at zero) or the values normalised such that one of the items being tested have a score of 1; give a much clearer representation of how the cards really perform against each other. People that randomly choose scales, are either sales people, or stupid. In reality I am not interested in how one GF4 Ti4400 card performs against another. For the most part card to card comparisons have less than 1-2% deviation across the entire range. I am more concerned with reliability/price and how each generation of cards compares with each other.
The point here is for the reader to be able to recreate the best gaming expierence possible. Key issues here are speed, consistency and visibility. For that to happen you have to create 'worst case' benchmarks of the likely situation you will find in the game. That means action, lots of action. One such demo is the OCAU 'slayer' demo for quake3. 12 bots and silly amounts of gunfire movement etc. Now when creating the 'ultimate gaming expierence' the first thing you have do is work out what the limiting factors are. 9 times out of 10 this is your monitor. I am fortunate that I can run q3 at 1024x768@120hz, however for most people this is 85hz. Buying a decent monitor may improve your game more than say a 20% increase in cpu speed. At this point you have to decide on the trade off resolutions vs frame rate. In multiplayer frame rate wins the majority of the time.
A plea to all games developers. Please Please Please rewrite your benchmark calculations to include the minimum, maximum, average and standard deviation (like the unix ping function). This will give a much better feel for the consistency of the gameplay and how it runs on a given system. Here is where the 'limiting factor' comes into play. The maximum that can be displayed (in my case) is 120fps. So I might as well set com_maxfps to be 120 (ignore the q3 quirky physics for the time being) and vsync on. This means that to create 'the utimate gaming experience' the minimum and average fps should be ideally 120fps also. This perhaps isnt possible so the standard deviation gives a clear picture on how much variation there is from the average value.
I have a gf3 ti500 and 1ghz tbird, using the 'worst case' benchmark (ocau_slayer demo)
Q3 Normal Q = 73.6 fps
Q3 High Q = 73.2 fps
Q3 High Q (1024x768) = 71.3 fps
The demo is conducted on one of the more graphically intensive maps tourney4. Watching the timedemo the fps was dropping as low as ~40 and peaked at the 120fps limit. Upgrading my gfx card would undoubtably increase my average fps, if and only if I uncapped the framerate. This is how most of the benchmarks are conducted. The current boom in the graphics card market and decline in cpu sales is partially due to the benchmarks giving the wrong impression of what is actually needed.
However there is one leading light when it comes to hardware reviews. Dansdata. www.dansdata.com go check it out. Informed unbiased reviews about anything technical.