Originally posted by saturnotaku
But the numbers can be misleading. I had a Radeon 8500 and a GeForce3. The 8500 would score consistently 700-1000 points higher in 3dmark than my GF3. But in games, that ATI card would choke with low framerates and/or graphical errors whereas the NVIDIA card would deliver smooth performance and no visual errors to speak of.
I agree with your findings to a point. I agree with your concuslion to a point. The issue I have with your reply is 3dmark was buit and advertised as a forward looking benchmark. Thus the scores you saw and the games you played then did not really have any correlation. Its with the new games that really use those features of DX8 that will tell us if there that bench was accruate. For example we know that Doom3 will use most of the DX8 feature set just in an OpenGL version. We know that the 8500 has its own path (dubed the R200 path). We know that JC has also built other paths for different cards. In JC's plan late last year he said that there are cases were the GF4 beats the 8500 and cases were the 8500 beats the GF4. We all know that the GF4 > GF3 so can we conclude that the 8500 in most cases will be faster than the GF3 just like 3dmarks2001 "said" it would? Dont know unilt the game ships and we see benchs.
The point is that its very easy to not understand what the numbers are telling us. They are not telling us how current games will work. I agree that 3dmark is/was/always will be poor at current games. Its track record for furture game was not all that....but only time will tell how well/bad 3dmark2001 ...
Again I am not saying your point is not valid for current games. It is and I agree with it. Just to early to tell on the future stuff....