Quote Originally Posted by Katzenschleuder View Post
That you think that the benchmarks shown do not represent the reality about how realistically 3DMark mimics the GPU workload of games is just your baseless opinion. You can pick ratings with many games like here and will see the same obvious discrepancy when comparing to 3DMark results.
And again, considering ratings of games in higher resolution even favours a correlation to 3DMark Vantage because they are more limited by the GPU.
What are those performance charts based on? DX9, DX10, DX11 or a combination of all?

You're undermining your own argument by using either too little or too much data to prove that 3DMark Vantage, an indication tool for DX10 performance, is inherenlty producing false results. If you want to compare DX10 performance accuracy, you need multiple (more than two) DX10 benchmarks using the same resolution as the preset you're comparing your results to. Preferably also the same IQ settings. You have to exclude all data that does not fit the comparison, which includes alternative resolutions, IQ settings and game-types.

Then, to prove 3DMark is inherently incorrect, you will need to do correlative research to find out if one of your OWN games is biased specificly towards one type of graphics card as it might bias your overall figures.

If you've done all that research and still come to the conclusion that 3DMark is wrong 50% of the time, you can write an article which will end with the conclusion: "although 3DMark gives you a good view on what hardware is capable, when considering purchasing a new graphics card it's better to check out more than just the 3DMark number". Also, you might want to give readers the advise to also check out other hardware websites to see if different platforms give different performance.

So, the conclusion of all this: if you rely on one graph to decide what product is best for you, you're a retard.

Quote Originally Posted by Katzenschleuder View Post
Why do you bother when you evidently don't care about the realism of the 3DMark-workload anyway considering that you argue to have the feature to pull 3DMark apart into even more meaningless sub-benchmarks!
Sadly enough, you again incorrectly analyze data.

The ability to run sub-tests separatly has nothing to do with the benchmark's relevance as a performance indicator. When comparing products, you should always use the same methodology (you know this, right?). For comparison purposes, it's quite obvious that any overclocker will also use the full benchmark.

Being able to run sub-tests separatly has nothing to do with performance comparison between hardware products. It's for overclockers to compete against each other. Those two are very different segments of the IT community.