Originally posted by: Yield
Originally posted by: AdamK47 - 3DS
I saw the title of this thread with the word 3DMark in it and had to check the originators registration date. The results I found are not suprising.
Even though the benchmark doesn't really mean much... it can at least tell you if your system is running good enough or not. It matters more if you get good FPS in the games you play... but 3dmark is a good comparison.
That's the problem. I don't believe that 3dmark is a good comparison. Computers which are identical (as reported in ORB) can have HUGE varience in score because of driver tweaking (well, driver mucking). This causes people to expect higher scores than what they measure with default settings, and they never realize that with extra 3Dfarks comes a trade-off in image quality / stability.
For example, this guy thinks he should be scoring 8000-8500 with a 2.4G p4 and a Ti200. I'm guessing that if he checks some basic things (to which I provided a link above) his system would score about 7500 with default settings. Overclock the card and mess with the drivers and 8000-9000 is easy. So, which score is "right"? 7500? But when he hits the compare button all he sees are PC's with the same "stuff" as he has, but they are scoring 9000. That's when the confusion sets in and they post to sites asking why their 3Dfarks are so low.
In addition, there have been numerous cases on these forums where people ask about low 3Dmark scores, without ever running a real-world application to benchmark. When they do so, the scores end up right where they should be. I'm not sure why, but it happens... and I wouldn't put something past the fact that all of the cards are so heavily optimized towards 3Dmark that if something is a bit "off" and the optimizations aren't, well, "optimal" anymore you'll see a performance drop-off.
Just my $0.02 why I belive that 3Dmark has become a poor comparison tool, unless comparing changes is same system.