3DMark questions

[BunnyStroker]

Sup guys. I have a couple of questions:

1. Does anybody actually know the scoring system that 3dmark uses? I'm not talking about "faster=higher!", but actually curious as to the specific mathematical formula used to compute a score.

Also, is score "adjusted" for quality? I.e. if you run 4x AA your frame rate is going to drop, for sure, with the benefit of a better looking image. Does 3dmark "count" this, or is it merely based on raw performance?

I'm sure if I spent enough time on the 3dmark site I could find this all out, but frankly the site is poorly designed and annoys the hell out of me. The forums there aren't great either.


2. In your opinion, how valuable do you think the 3dmark scoring system is? I.E. do you find 3dmark to be unfairly biased to a certain processor or chipset? Or do you think that score and real performance are correctly related?

Please, keep any flames to a minimum, and if you are going to claim some bias you better have a valid point.



<b>ßunn¥§troker 0wn§ j00!</b>

 

[HonestJhon]

well, this is something that i have been told by people at mad onion. 3dmark 2001 (and i think 2000, but not as much) is greatly dependant on processor speed and capabilities. sure, it relys a great part on your graphics card to get a higher score, but people with overlcocked fsb, and overclocked cpu's will get higher scores than those who are running at stock speeds. as for the equation that you are asking about, i have no idea. i dont even know if they would release something like that. but they might. and i do take the scoring of 3dmark seriosly to some extent. if you look on the box for max payne, it says that the required computer is one that scores at least 1000 on 3dmark 2001. my computer in the delapitated state it is in right now, it only can score like 750. but it is a k6-2 500 with a geforce2 pro. not overclocked. i dont know how much i would gain by overclocking it, but maybe a few points. also, i think that if i ran the test on my dad's celeron 500, he might score slightly higher than me because intel chips are usually better number crunchers. (I'm talking in technology that is pre-athlon!!!). but i have yet to do so, because he has no time to run it. i will do it someday. gaming magazines like "PCGamer" use 3dmark2001 as a benchmark to see how good a computer is. at least as part of the decision. they also run quake3, MDK2, and some other software for benchmarking (i think). overall, i think that 3dmark 2001 is a good benchmark tool, but it's limit is direct3d. also, to test for image quality, like running with fsaa on, you would have to purchase the full version of 3dmark2001. it will then have the image quality test. other than that, you are better off turning fsaa off when you run the test.


-DAvid

-Vacuums suck...really...-

 

[Matisaro]

Your cpu is whats holding your videocard back, agf2 pro should get at least 2000 in 2001.

~Matisaro~
"The Cash Left In My Pocket,The BEST Benchmark"
~Tbird1.3@1.5~