Athlon XP 2400+ (2.0GHz)
x2 512MB DDR PC2100 RAM
ATI x1650 Pro AGP
3dMark05 Score: 3750
I found it odd that Computer 2 had such a higher 3dmark score, but I figured that the x1650 pro must be significantly better than the 6800xt. It wasn't just a fluke either. The test was repeatable, and the FPS were often 10 higher with computer 2 than computer 1. So I pop in Battlefield 2 and try to find the best settings for each computer (trying to keep the FPS above 30).
I start at 800x600 resolution for both PC's with every option on "Medium" except for Textures on "High." What I see in game is that Computer 1 is now whooping Computer 2! On Computer 1, I can look around in all directions and my FPS never drops below say 30. It is usually 40-50. On Computer 2, the FPS is usually about 30, but it often drops to 18-25.
Even when I lowered the settings to everything "Low" and textures on "Medium," it seemed to make very little difference. I was still seeing drops to about 23 FPS. I ultimately had to set everything to Low, and turn off dynamic lighting on Computer 2 to get the same FPS as Computer 1.
So my question is, why is this so, even though Computer 2 whoops Computer 1 in 3dMark05 at the default resolution? (3750 points vs. 2050 points)
More about :weird 3dmark real world performance x1650pro 6800xt
If memory serves me well, at lower resolutions, the CPU becomes the bottleneck in any game, and as system 1 has a faster CPU than system 2, that's most likely the reason why it's performing better.
I myself have found Battlefield 2 to be fairly CPU sensitive, even when the resolution is turned up a bit.
It must just be cpu bottlenecked. I don't buy that it's because the x1650 pro is rendering more features than the 6800xt. I know this because even when the x1650 is rendering everything on "Low", and the 6800xt is rendering everything on "Medium" (w/Texture on "High), the 6800xt still gets much better FPS in high activity areas.
I know that 3dMark scores just give you a general idea of how well a card performs, but I'm pretty surprised by how off it was in this case.
Even though the computer with the x1650pro trounces the one with the 6800xt in 3dMark05 by like 75%, it does extremely poorly in BF2 while the other does relatively well. Why does 3dmark05 not show this CPU limitation at low resolutions (the default test is 1024x768 or something like that)?