Futuremark releases 3DMark03 patch - alleges NVIDIA cheats

[gordon151]

Originally posted by c4c
Why? Most likely there will be plenty of cheating in the timedemos...Heh, Nvidia already tried to pass of their own homemade demo to iD.

The only thing that won't invalidate those benchmarks is if web reviewers make their own demos. Often..

That's such a good idea since most benchmarks are done using timedemos that are included in games. It could be easy for them to detect that it is being run and then apply whatever optimisations they have to skew the results. Reviewers need to start using FRAPS and in game benchmarking backed up with screenshots to make sure of these things now it seems.

[Hanners]

Originally posted by Nv40
Mmmm.. how can this be ? ATI is cheating too
now who is inflating scores here ?

Indeed, both major IHVs appear to be cheating in drivers.

Although having said that, inflating your score by 25% is somewhat overkill on the cheating side! It's pretty shameful to be honest.

[Behemoth]

Originally posted by Moose
I think the exact opposite.

Name one benchmark that has a way of detecting cheating on the part of the video card companies.


yep... 3dmark2003


I think this vindicates Futuremark even more. Not only have they caught cheating, they have also implemented code to prevent it.

No other benchmark can say the same.

NONE!

hmm, 3dmark03 didnt have a way of detecting cheating.

How Were These Driver Cheats Found?
Members of Futuremark’s BETA program3 first noticed how parts of the tests in 3DMark03 were
rendered differently on different hardware. When testing NVIDIA hardware on 3DMark03 with socalled
developer’s version’s free camera enabled, they noticed how some parts of tests were
rendered strangely, and informed Futuremark of their findings. Futuremark investigated further
and our findings show that certain NVIDIA drivers seem to detect when 3DMark03 is running and
then replace the 3DMark03’s rendering requests with manually implemented alternative rendering
operations. These alternative rendering operations reduce the amount of rendering work and
thereby increase the obtained benchmark result.

Jesus Christ, why are we supposed to "trust" a company? A company is designed for one thing--to get YOU to give YOUR MONEY to THEM. I don't trust ATI, I don't trust NVIDIA--I look at performance, cost, and image quality (not necessarily in that order), and then I'll make a decision. I have no brand loyalty--I just buy what's best at the time.

I do, however, have a deep-seated hatred for Futuremark--if Futuremark didn't exist, then we'd be getting benchmarks from games. So if we were getting benchmarks from games, then the companies would optimize for games. And even if they did cheat a la the NVIDIA 3DMark cheats, then we (AS A COMMUNITY) would be able to tell--not some guy with a developer edition.

Hell, release an edition to everyone that allows you to change the camera, Futuremark. Then I'll believe that you don't suck.

[zakelwe]

Originally posted by Hanners
Scores submitted by users and those shown in reviews are two very different things.

I don't suppose OEMs choose who to make their contracts with by looking throught the Orb.

Besides which, none of those rendering errors will occur any more with the latest patch, as this has effectively 'disabled' all cheats.

Well they excluded them in the first place, so it must have mattered to them ? No ?

Also, this was the set that was sent out to reviewers specially to use, so it is no different :-

http://www.hwextreme.com/reviews/misc/nvidia_4268/

"We received these drivers directly from NVIDIA, and it's available at their Press FTP. The Press FTP is a private FTP site which requires a login and password and it's not open to the public."


I bet you those errors are still in the new patch when using 42.68

Regards

Andy

[Moose]

Originally posted by The Baron
If NVIDIA cheats in games, and I can't see that it's a cheat EVER, then it's not a cheat--it's an optimization.

How will you know that you aren't being cheated?????

Will you know from looking at your monitor that that was how the game was coded to look???

Wow, the logic of some "obviously fanboy blinded" people here amazes me.

[Hanners]

Originally posted by The Baron
Jesus Christ, why are we supposed to "trust" a company? A company is designed for one thing--to get YOU to give YOUR MONEY to THEM. I don't trust ATI, I don't trust NVIDIA--I look at performance, cost, and image quality (not necessarily in that order), and then I'll make a decision. I have no brand loyalty--I just buy what's best at the time.

So, you look at performance? What makes you think the performance numbers you read in the press and on the Internet are correct?

This is why you expect to be able to have some trust in a company - When you see any benchmark scores in a review, you expect that to be what you will get when you play games, not an artificially inflated score that makes product x look better than product y.

[Moose]

Originally posted by Behemoth
hmm, 3dmark03 didnt have a way of detecting cheating.

Then how did they find the cheating???

Obviously they DID have a way...


DUH!!!!

Originally posted by Moose
How will you know that you aren't being cheated?????

Will you know from looking at your monitor that that was how the game was coded to look???

Wow, the logic of some "obviously fanboy blinded" people here amazes me.

*sigh* That's why games have SDKs--you can put your own stuff in, and you will know if it's working correctly or not.

[Smokey]

Originally posted by Gar
WEh00000 I"m on the rollercoaster of denial.....aaahhhhaoooo..

It's obvious that NVidia has finally been exposed and now we all know (nvidiots included) what the company will do to remain competitive. Not releasing good hardware, but releasing SUp0r H4x3d drivers to maintain that competition. Its a sad set of circumstances.

I dont give two hoots about 3DMark, when buying a videocard I just want the card to excel in games, not 3DMark. As far as I know, ATI cards still have problems in a few games, one bieng RavenShield which had been tooted as an optimised game for ATI cards!

All that really matters, is that videocards run games fast and without errors at high quality, and I think in most cases both ATi and Nvidia cards do exactly that?

Then how did they find the cheating???

Obviously they DID have a way...

ARGH! There is no ACTIVE cheat detection--it got by the ATI and NVIDIA cheats by going one step beyond simply changing the filename; they changed the CRC on the executable very VERY slightly, and as a result, the drivers did not detect 3DMark.

[Behemoth]

Originally posted by Moose
Then how did they find the cheating???

Obviously they DID have a way...


DUH!!!!

duh

How Were These Driver Cheats Found?
Members of Futuremark’s BETA program3 first noticed how parts of the tests in 3DMark03 were
rendered differently on different hardware. When testing NVIDIA hardware on 3DMark03 with socalled
developer’s version’s free camera enabled, they noticed how some parts of tests were
rendered strangely, and informed Futuremark of their findings. Futuremark investigated further
and our findings show that certain NVIDIA drivers seem to detect when 3DMark03 is running and
then replace the 3DMark03’s rendering requests with manually implemented alternative rendering
operations. These alternative rendering operations reduce the amount of rendering work and
thereby increase the obtained benchmark result.