More proof 3DMock'03 is useless*as if we needed more* ;)

[DAPUNISHER]

link

 

[modedepe]

Hah, that's outrageous. Maybe ati's new drivers are "optimized?"

 

[Lonyo]

They have a thread in the Futuremark forums for people to post scores which don't seem very attainable.
It's not the first 10k+ 3D Mark 03 score to have been discovered.

 

[kylebisme]

just becasue bencmark numbers can be fudged doesn't mean that the benchmark is useless, just that the fudged results are.

 

[Jeff7181]

It's not completely useless... just inaccurate and biased (Futuremark software in genreal I mean, not specifically 3DMark2003). Users need to know HOW to use it, and what purpose it serves.

 

[FreshPrince]

the numbers were not fudged! everything was stock, the program just sucks

 

[Jeff7181]

I wonder if FRAPS would report 200 FPS in Mother Nature???

 

[Regs]

Originally posted by: FreshPrince
the numbers were not fudged! everything was stock, the program just sucks

I think the term "fudged" includes program error.

 

[nemesismk2]

It sounds to me like he was using 3danalyzer to get a score like that, I've used it myself to get 3dmark03 to run with an old kyro 2 which shouldn't even run any of the tests.

BTW with my old kyro 2 emulating things like cube mapping etc I got a huge score of 174 which is impressive isn't it!

I could of got an even more impressive score by disabling useless things like textures but that would be cheating wouldn't it?

 

[DAPUNISHER]

Originally posted by: Lonyo
They have a thread in the Futuremark forums for people to post scores which don't seem very attainable.
It's not the first 10k+ 3D Mark 03 score to have been discovered.

Why is it the only one in the database, do they remove it when discovered?

 

[jim1976]

Originally posted by: TheSnowman
just becasue bencmark numbers can be fudged doesn't mean that the benchmark is useless, just that the fudged results are.

 

[JonnyBlaze]

the only test out of wack is: GT4 - Mother Nature if it were a program causing that wouldnt all the tests be screwed?

i only get 39.4fps on that test. he gets 199.4

thats crazy.

JB

 

[RussianSensation]

No guys you are all wrong, the score is legit

Some dude on the assembly line accidentally inserted R420 into his card, oh yeah....also he has a Nitrous button for added performance on the side of his case.

On a lighter note, I think that basing performance of your system even in isolation to gaming environment on one single benchmark implies putting too much weight on one benchmark which could not alone reflect in reality the true performance of your system. It is unfair to simply blame 3dmark 2003 as a single benchmark that has produced unrealistic scores for the hardware at the time, because the same problems have occurred with 3d01 and pcmark02.
However, nothing is to say that you could not reasonably compare vertex and shader performance of your card towards other cards and your improvements after and before overclocking of the videocard using the Futuremark software. The results might not be indicative of reality since most game engines do not utilize the programming techniques implemented by Futuremark, but you can compare the raw performance of the card in general and say to yourself just how well your card will run graphically intensive applications which are not optimized like game software is. Yes, of course, this isnt real comparison, but if the company's did not cheat, why can you not simply look at the FSP in 3dmark 03 as the raw, unoptimized performance? Also, when videocards becomes fast enough and the score is able to reflect that difference somewhat as opposed to now where all the top cards are bottlenecked around 5500-6000 points. Maybe 03 will gain wider acceptance as a videocard benchmark, but I seriously believe Futuremark undermined the sole purpose of this benchmark but representing it as a whole system benchmark as this aspect of it is misleading.

 

[FreshPrince]

Originally posted by: RussianSensation
No guys you are all wrong, the score is legit

Some dude on the assembly line accidentally inserted R420 into his card, oh yeah....also he has a Nitrous button for added performance on the side of his case.

On a lighter note, I think that basing performance of your system even in isolation to gaming environment on one single benchmark implies putting too much weight on one benchmark which could not alone reflect in reality the true performance of your system. It is unfair to simply blame 3dmark 2003 as a single benchmark that has produced unrealistic scores for the hardware at the time, because the same problems have occurred with 3d01 and pcmark02.
However, nothing is to say that you could not reasonably compare vertex and shader performance of your card towards other cards and your improvements after and before overclocking of the videocard using the Futuremark software. The results might not be indicative of reality since most game engines do not utilize the programming techniques implemented by Futuremark, but you can compare the raw performance of the card in general and say to yourself just how well your card will run graphically intensive applications which are not optimized like game software is. Yes, of course, this isnt real comparison, but if the company's did not cheat, why can you not simply look at the FSP in 3dmark 03 as the raw, unoptimized performance? Also, when videocards becomes fast enough and the score is able to reflect that difference somewhat as opposed to now where all the top cards are bottlenecked around 5500-6000 points. Maybe 03 will gain wider acceptance as a videocard benchmark, but I seriously believe Futuremark undermined the sole purpose of this benchmark but representing it as a whole system benchmark as this aspect of it is misleading.

Yes, the fact that nature score can skew the score so much is silly. I mean, all other scores represent what this system should perform, but because of one renegade score, the entire point system is skewed out of whack. This then can be said as BS, kinda like BCS point system

 

[Jeff7181]

Originally posted by: RussianSensation
No guys you are all wrong, the score is legit

Some dude on the assembly line accidentally inserted R420 into his card, oh yeah....also he has a Nitrous button for added performance on the side of his case.

On a lighter note, I think that basing performance of your system even in isolation to gaming environment on one single benchmark implies putting too much weight on one benchmark which could not alone reflect in reality the true performance of your system. It is unfair to simply blame 3dmark 2003 as a single benchmark that has produced unrealistic scores for the hardware at the time, because the same problems have occurred with 3d01 and pcmark02.
However, nothing is to say that you could not reasonably compare vertex and shader performance of your card towards other cards and your improvements after and before overclocking of the videocard using the Futuremark software. The results might not be indicative of reality since most game engines do not utilize the programming techniques implemented by Futuremark, but you can compare the raw performance of the card in general and say to yourself just how well your card will run graphically intensive applications which are not optimized like game software is. Yes, of course, this isnt real comparison, but if the company's did not cheat, why can you not simply look at the FSP in 3dmark 03 as the raw, unoptimized performance? Also, when videocards becomes fast enough and the score is able to reflect that difference somewhat as opposed to now where all the top cards are bottlenecked around 5500-6000 points. Maybe 03 will gain wider acceptance as a videocard benchmark, but I seriously believe Futuremark undermined the sole purpose of this benchmark but representing it as a whole system benchmark as this aspect of it is misleading.

This, along with a ballpark idea of what your card "should" score if it's operating properly are the only two things I find useful about 3DMark2003. Obviously if your 9800 Pro scores 2000 while everyone else's scores 6000+ there's something wrong.
Oh... one more use... if you like staring at pretty graphics but suck at games, 3DMark2003 is a great "game" for you

 

[Acanthus]

I stopped using 3dmark over a year ago.

 

[Pete]

What the devil ... a buggy program?! ALERT THE PRESSES!

 

[BFG10K]

nVidia went to an awful lot of trouble to cheat in a "useless" benchmark.

 

[THUGSROOK]

its is useless.
use a real game for benchmarking for gads sake!

i personally think its great that ATI was able to fudge 3dmock so easily ~ trying to prove a point are they?