3dmark vs real games

[kevineugenius]

The 2900xt review I was looking at over at ... OCW I think? Anyway, it showed the 2900hd outscoring the 8800gtx by a somewhat slim margin. However, in-game performance the 8800gtx spanked the 2900xt's bootie. So, what is the difference? Why doesn't 3dmark strive for more realistic 'gaming' benchmarks? I'm sure that's what they want, but an inaccurate benchmark program is worth what... nothing but the visual displays it shows (Matrix Lobby Scene was still the best).

 

[AltTabbins]

Why doesn't 3dmark strive for more realistic 'gaming' benchmarks?

Because all it takes is fraps and whatever game you want to benchmark in order to see results.

Generally, people who care a lot about 3dmarks couldn't care less about in game performance. They buy hardware for a higher score and the rights to brag to people who don't give a shit about 3dmarks.

 

[snowysnowcones]

If ATI could get all their drivers optimized the way they've done in 3dmark, they might have a winner on their hands. But that's a pretty big "if".

 

[Marvelous]

I think 3dmark 2006 emphasizes on shader performance more than anything else. Since 2900xt have 320 stream processors it performs better than 8800gtx with 128 stream processors.

This can be evident with Radeon 1650pro vs 850xt in 3dmark 2006. No way does the 1650pro even come close 850xt performance but you know it's neck and neck with x800 series in 3dmark 2006.

If you want to see a realistic synthetic benchmark. I think 3dmark 2005 does a better job.

 

[Brent_Justice]

3dmark uses an unrealistic scoring method of an unrealistic gaming engine of an application that is not a game.

 

[Weenis]

3dmark uses an unrealistic scoring method of an unrealistic gaming engine of an application that is not a game.

Couldnt have said it any better.


Games matter, benchmarking apps are a waste of your time, if you spend 15 hours reading reviews and comparing benchmarks, you wouldve been better off working for those 15 hours and buying a better card.