YAVCDT: Article about cheating in drivers

[bandana163]

I've just found a Hungarian article about nVidia and ATI drivers. They've tested the 3.4 catalyst and the 44.03 detonator using default settings and Rivatuner's anti benchmark-detector script. The differences are extreme.



Explanation: For those who don't speak Hungarian, the pics order on each page is the following for the 3DMark 2001 and UT2003 tests:
1st block. Test with default settigs. (with cheating in drivers)
2nd block. Test with anti benchmark-detect script (normal real-life, as it should be)
3rd block. Loss of performance% due to loss of benchmark-optimizations

What happened if they tested 3DM 2003, too?The test page 9

 

[Robor]

So if I read that correctly ATI "cheats" a lot more than Nvidia. The question is, what is the "rivatuner anti-benchmark detector script"? Who made it and how do we know the script isn't biased? Please keep in mind I'm not biased toward/against either maker. I own a 9800Pro and TI-4200.

 

[bandana163]

Nope. It says the performance loss of nVidia cards are 7-57% when the benchmark optimizations are diabled, and Ati cards barely lose 1-2-3%. Trying the new Rivatuner addon personally should help, though. The test itself is going to be continued with image quality tests soon, because it is suspected that nVidia cards use different filtering methods thanthose set.

 

[Schadenfroh]

do we really need to bring this back up again?

 

[nemesismk2]

Where is the proof that these anti-cheating programs actually work? I know with anti-cheating enabled it results in lower scores for nvidia video cards but that can be caused by alot of different things.

 

[McArra]

Originally posted by: Schadenfroh
do we really need to bring this back up again?

 

[Blastman]

Originally posted by: Schadenfroh
do we really need to bring this back up again?

What is with these NV fanboys trying to censor the discussions on Anandtech because they don?t want to hear about how much NV has been cheating anymore? If you don?t want to discuss it just don?t enter the thread --- it will disappear pretty fast. It?s not like there?s been 10 threads on the subject in the past few days. And until the cheating is stopped it?s still an issue --- right?

My understanding is that some of the ? ?optimizations? ? ie shader optimizations which
both Nvidia and ATI used on game 4 of 3Dmark2001 are really a legitimate optimizations because this type of optimization can/will actually work in a game. Now doing it on a benchmark is another story. On the other hand ithas been shown that Nvidia uses cheats on games 1-3 whereas the ATI scores on games 1-3 stay the same with the antidetector. The antidetector has yet to find a cheat on any game with the ATI drivers whereas just about every popular game that comes with a benchmark has been found to have a Nvidia cheat/optimization.

Since Nvidia was cheating with their GF4 series like the 4200, when the 5600 turned out to be a somewhat poor performer, in order to make it look good against the GF4 series they had to cheat even more. That?s a slippery slope. ?

5600U ? 21%
4200 ? ?14%

 

[Genx87]

I think you missed the point blastman. What does the anti-cheat actually do? I mean p00p I could revert Nvidia drivers back to Det 3s and you wouldnt even know it. Then claim Nvidia cheats......

The question is "What does the anti-cheat program actually do?"

Just because somebody makes an application and labels it "Anti-cheat" it doesnt make it valid for anything.
Also in case you didnt know in Game 4 ATIs optimization actually changes the output of said picture.

Personally I dont think it is as big of a deal as people are making it out to be. If the hardware vendor is willing to make optimizations for games and the image quality stays the same what do I care? It is like optimizations for the P4. Without optimizations the P4 is a dog but with them it flies. DO people cry and whine about the P4?!?!?!?!?!?!?!?!?

 

[Jeff7181]

Die horsey!!! <wack> <wack> DIIIIEEEEEEEE!!! <wack> <wack> <wack>

 

[Blastman]

Originally posted by: Genx87

Also in case you didnt know in Game 4 ATIs optimization actually changes the output of said picture.

I think I read that the Nature scene (game 4) calls for Pixel shader 1.0. Now suppose ATI uses PS 1.4 or 2.0 because it is more efficient or faster. But also, the newer card is rendering in FP24 rather than the original FP16 as intended so there are less rounding errors which means a difference from the original but even better than the original. Is that a cheat or a legitimate optimization? It seems to me the ATI card maintained exact frame rates in all the other game tests which the Nvidia card didn?t do, so it looks to me like Nvidia is cheating for sure on some of the game benches. On the Game4 Nature scene the evidence is not conclusive just because there are minor differences in the pixels --- and NV rendered the scene differently too..

 

[DaveSimmons]

Originally posted by: Jeff7181
Die horsey!!! <wack> <wack> DIIIIEEEEEEEE!!! <wack> <wack> <wack>

LOL, too late it's been dead for weeks now but considering the Zombie Movies thread in off-topic perhaps it's Undead.

 

[Amorphus]

It's not dead! I saw it twitch! AIEEEEEE!!!!!!

also, I'm sort of skeptical as to all these allegations of "cheating". is there an actual performance increase with the "optimizations" turned on? as in, there are image quality/fps increases with the optimizations? well then thats not cheating. thats called "tweaking", and we should thank nVidia for doing it for us, eh?

just playing Devil's advocate, because thats what may be happening.

 

[DaveSimmons]

I shouldn't even reply, but if you read any of the earlier threads, you'll see that nvidia has been caught with cheats in real games that affect image quality. But this horse really has been beaten to death, so if you are interested just read the earlier threads for many, many statements pro and con.

 

[Jeff7181]

<listens to the low pitched thud of the baseball bat beating against the rotting corpse of the dead horse> <wack> <wack> <eyes widen as the corpse splits open and swarms of maggets crawl out> <wack> <wack>