3DMarks/3DWangs

[Guest Timmay]

Where is your stand point on 3DMark 05/06?

I know it is used for the SSDB and the OCDB but I'm talking about consistent benching and nothing but a numbers tug test.

 

For example a new piece of hardware comes out (8800GTX) and the only thing people can do with it is run 4 programs, 3DMark 01/03/05 and 06.

 

I for one get very tired of all the epeen contests and want more then just 20,000pts on 3D0X to be run in the way of tests... I want REAL games to be tested, REAL applications not synthetic piles of code that will lead to an unrealistic idea on performance.

 

3DMarks do have their place but what I'm trying to say is it's not the ONLY way to bench a new GPU/CPU on the market, why not give us some variety.

 

Please, discuss.

[MACarter02]

You mean like Toms Hardware type tests? LIke multiple applications (Video Encoding while burning a CD) and stuff like that?

 

Sure 3DMark might be a rather superficial way of testing your computer, but the results you get are anything but trivial. These type of benchies show you an objective perspective as to where your current hardware/ settings stand compared to others.

 

Your only other option is to use a given computer for a few days, and see how you like it. Sure that seems to be a more practical way of assessing your hardware. It would be rediculous for me to say: 'I wont buy that video card unless it raises my 3DMark06 Score 1000 pts.' However, if it does raise the score 1000 pts, you can be SURE that this hardware is better than your previous.

[Guest Timmay]

However, if it does raise the score 1000 pts, you can be SURE that this hardware is better than your previous.

 

This is the mentality, lets say a 7900GTX + C2D @ 4Ghz gets 10,000pts 3DMark06 and along comes Kentfield (sp?) and has a 7900GTX + Quad-Core @ 2.8GHz and gets 12,000pts... I can easily say that the C2D will do FAR FAR better in games then the Quad-Core will yet the Quad-Core gets a higher score how is that a true representation of games today?

[Thraxz]

Not really. They really only allow you to compare to the same make and model.

 

Also, though the XTX's are considered to be the superior card of their generation in real gaming somehow 06 always seems to rate them that much lower than their competition. Seriously, a single XTX at 760/1720 is outright faster than a single 7900GTX in real games short of active cooling and yet, we're getting the same 3dmarks? Pfft.

 

05 and 06 are nothing more than e-penis contests and excellent testers of stability(though not foolproof).

 

Anandtech recently had a review on the 8800 on games and quite frankly it sucked. Their numbers were skewed ridiculously low (my 1900XTX at STOCK blasted their numbers at the settings they gave).... their testing methods weren't given besides 4xAA/16xANISO... Ugly ugly thing. They really didn't even try. I suppose it was preview but at least try to explain your . numbers. I took this and this alone away from the article, "Yes, the 8800 is better than the last generation's card(1950XTX)."

[ollietwinam]

Most sites do have game results, normally Quake 3, Doom 3, FEAR, CSS and all that...

But yes, I agree with Timmay... while 3d Mark utilises all 4 cores in a way that no game will for a while (on a Kentfield), it doesn't show true life performance.

But then, if you know that, then you know to account for that in the score... so doesn't it all balance out?

[Thraxz]

But then, if you know that, then you know to account for that in the score... so doesn't it all balance out?

 

No, because they are nothing more than e-penis contests. How often, when you got SMOKED in 3dmark do you honestly say "well I know mine's just as good in games as theirs?" I know I still want to pull out my penis and wave it. Even in that act, it really trivializes exactly what you're trying to say.... it's now a penis contest, not a measure of hardware speed.

[Angry_Games]

i mostly use them as stability tests and a bit of comparing system to system

 

I believe that they have no real bearing on how a system will truly perform under true stress of a real 3d game

[Guest Timmay]

Agreeing with happy games, Hell in 3D05 I could use 256MB of PC2700 Ram and still get a decent score as long as I had SLI of 7900GTX's... Wheres the point in that.

[OsamaBinAthlon+]

For example a new piece of hardware comes out (8800GTX) and the only thing people can do with it is run 4 programs, 3DMark 01/03/05 and 06.

 

No they run stress tests with diff games at diff resolutions with diff features enabled showing diff FPS scores.

[Guest Timmay]

No they run stress tests with diff games at diff resolutions with diff features enabled showing diff FPS scores.

 

No, How many threads do you want to wager (on XS) that have NOTHING but 3DMark 0X screen shots in them for the new 8800GTX's?

[OsamaBinAthlon+]

No, How many threads do you want to wager (on XS) that have NOTHING but 3DMark 0X screen shots in them for the new 8800GTX's?

 

sorry I'm talking about real benchmarkers such as the people from CPU magazine.

[Thraxz]

Once again, the thing is... the benchmark numbers aer worthless. Why do them? They are e-penis only... not real comparisons besides against your own rig pre/post OC. Even then, that is pointless because games will STILL tell you if it helped.