3d Mark score help

[adamsh]

Hi Guys,

I recently ran a 3d Mark 2001 test and have a question.

After comparing my results to others with the same set-up I found that I am near the bottom of the pile, does anyone know how this person managed to get 3000 points more than me on a stock set-up..


My Details: http://service.futuremark.com/compare?2k1=6753212

his Details: http://service.futuremark.com/compare?2k1=6804870


cheers
Adam

[fondie]

links don't work

but first guesses would be :-

1. running on a completely clean windows install
2. not having other programs running in the background
3. using faster graphics drivers
4. turning down graphics quality settings
5. luck

[fondie]

ok, you've sorted links, i'd still suggest all the things i previously mentioned and also :-

6. faster memory timings
7. overclocked graphics card

[Austin]

Basically most of the results posted are from o/c'ed systems, very often very high FSB speeds and gfx cards right on the brink of crashing. It's always best to look for average results rather than the top ones, look for where scores begin to converge and that is realistic for most users with the same setup ... again still with some o/c'ing.

If we take alook at your links we see...

* He is using AGP2x, probably also disabled Sideband and Fastwrites in order to o/c as much as possible.

* You are using 1GB but diffs are tiny in almost eveything so he isn't losing at all in having 512MB.

* He is using DX8.1, seems strange as DX9 should be even faster esp with a DX9 card and optimised drivers.

* You have P4C800 while he has the P4P800, is that 865 vs 875?

* His driver version is older (not sure of CAT number) but this may not render certain scenes correctly, a fix in the newer drivers could render things correctly at the expense of a little speed.

* He is about 20% faster, you could have a 9700nonPRO with a modded BIOS to show it as a 9700PRO, are your clock speeds 275/270 or 325/310?

* He could score 20% higher by...

1. o/c'ing the gfx card. 9700+20%=350/310 which is common. 9700PRO+20%=390/370 which is a little high but that could come under luck. All he'd have to do is complete the 4 main bms, the fact he didn't bother with the additional tests suggests his system was NOT truly stable.

2. Alterring detail settings like LOD bias. It makes things render much faster but looks absolutely disgustingly awful.

3. Cheating the system in some way. Editing the file/data. Flashing a 9800PRO with a 9700PRO BIOS. Etc.

4. Somehow running his RAM faster than FSB though that would not make a 5% diff let alone 20%. Are you running your RAM in Dual Channel in sync with FSB (ie 2x400mhz PC3200)? According to one source 865/875 work best when using all 4 slots and esp with Double-sided DDR sticks (not necessarily chips on both sides). Again this could make up SOME of the diff.

EDIT: Comparing my PC's results with yours you are about where you should be. With my prev XP1700+ @ 2.2ghz 400FSB (XP3000+ish) and o/c'ed 9500PRO @ 350/310 (roughly even to a 9700nonPRO) we see you are 17%ish faster. That's what you'd expect a 9700PRO to do. Bottomline your system seems fine!

http://service.futuremark.com/compare?2k1=6463724

[adamsh]

cheers for the replys...

P4P is 865 I think, with my P4C being 765.

I have a standard stock 9700 PRO from crucial,
My System memory is dual channel Corsair 3700 (gives BF1942 a little bit of power).

Thanks for putting my mind at ease, all I wanted was a realistic comparison from a computer that is usable with every day settings, i'm not sure why people feel the need to run their computer at stupid unusable settings just to get the Highest score they can.

[fondie]

the most likely reason is that he's got his graphics card overclocked and is running with quality settings set at minimum, prehaps with a mix of some of the other things.

people overclocks higher than normal in benchmarks because they want to see how fast their system is against others and with no standard on 'usable' settings, the max that will continue to run the benchmark is the only one that can be followed with any regularity, it's regarded as a competition and people will try and gain any advantage they can.

[Byatt]

I think I remember something about ASUS's "hyper path" on the P4P800 actually makes it as fast and sometimes faster than the P4C800 in some benchmarks...