Link should be http://www.yougamers.com/hardware/stats/3dmark06/priceandperformance/?mainnavi=true

Az, I dont know if u intended to write that or not the way you did, but the reason they are different by 10 FPS or so is because different cards from different manufactures are better then others, that is what this program does, lets you know which cards are more powerfull then others

*all examples please dont flame me fanbois

say your ati 3870 will be 20 FPS faster then a Nvidia 8500GT

the nvidia 8800GTS will be 20 FPS faster then the 3870 

this program measures quality of the cards, and also performance, the reason such a program varies by the amounts you stated are because of the diff capabilities of certain cards, its not biased to 1 manufacturer

this could work and work well, there is always problems when you are about to purchase a new video card, you do some reaserch, so you warm up the google fingers, and start looking for some benchmarks, the search returns some "trusted" benchmarks

but are they? trusted?
I notice every time they compare other cards to the one in the spotlight for that test, say they are testing the new 8800GT, they run all their benchmarks with their uber tastic pc that they have there. The problem is they compare other cards that were tested 3 months ago on different hardware. So how accurate are the results? not very I have been emailing benchmarking sites for years to get 1 test system and test so you can have verifyable results, and consistant.
this new 3dmark thing will add to that, being the same program, and I hope to god they use just 1 test bed to do it, so all the results are comparable.
This could work if they do it right

Its a lousy benachmark that does not relate to realworld performance. The whole 8600series has no point for gamers. If their budget is low 2600xt is cheaper and gives similar performance and if they look for realworld performance, 3850 is twice as fast and only slightly more expensive than 8600GTS.
Nvidia knows it that 8600 Series is obsolescent, thats why G94 is coming to the rescue aka 9600GT.
That shows us once again how badly Futuremark fails in predicting performanc of graphics cards in future games. 3dmark is a nice tool to test Stability and Overclocking gains on graphics cards, but thats about all volks.

3dMark is fine for comparison, since most games use similar GPU instruction sets used to benchmark the card. The difference you see is because ATi and nVidia don't really have similar performing cards AND they obviously use different arcitecture to make up their cores meaning that 3dMark will provide an average performance mark, where nVidia may do better in some instantces, ATi will do better in others.

I hope it measures bang for buck for the specific machine being tested - else once we've seen the cards compared, it's done... i.e. everybody gets rhe same result assuming the prices are 'equal'? Hey' isnt' that amazing, my best bang for the buck is also a 3850 512MB, we must have a lot in common :-)

A basically good idea and potentially useful.

I mean, it's obvious and well known to those likely to know these things that 3DMark scores aren't the be-all, end-all. Not by far. Indeed, there are quite a few exceptions to get-the-highest-3DMark/local-currency out there. But when someone who would otherwise buy based on how much video memory a card has calls and wants to know *right*now* which to buy of cards x and y, costing z and t Zimbabwean Dollars, respectively (z, t being expressed in scientific notation, naturally), then having a 3DMark result database handy is very helpful in making a decent educated guess.

I have a modest similar database myself, indeed. Since the days of 3DMark 2001 SE.

But did they have to use 3DMark06?

Well, I guess they did, it being their latest and all. But the problem, see, is that CPU results are an integral part of '06 scores (the white-paper on Futuremark's site has a handy equation, if you'd like a look). So even assuming they'll be averaging the scores they get and so on, higher-end cards (price-wise) will be getting their scores slightly-inflated while lower-end ones will feel the opposite effect since the former will be in better-spec'ed rigs than the latter.

Less useful than it might have been otherwise, so.

3dMark isn't such a good indicator though because in some games the difference between ATI and Nvidia is over 10fps or so.