Nvidia is a bunch of cheaters?

[evh347]

Is the Geforce FX 5900 Ultra really faster than the ATI 9800? Maybe not and maybe it isn't the fastest card out there. Just maybe, the ATI 9800 is still the best of the best?

Nvidia accused of cheating in 3DMark 03
http://www.gamespot.com/pc/news/news_6028894.html

http://www.flightsim.com/dcforum/Use...150436a827.jpg

[junky354]

[font size="1" color="#FF0000"]LAST EDITED ON May-25-03 AT 09:47PM (EDT)[/font][p]This should answer all of your questions.

http://www6.tomshardware.com/graphic...512/index.html

It includes a 3dMark 03 test, but it also has other game benchmarks. If you look the FS5900 comes out on top in almost every test!

no more pretty sig....

[tjkotula]

What's to stop ATI from doing the same? Funny how no one ever investigates the underdog. ;-)

Every company has done it at some point. Intel has done it, no doubt AMD has done it and I wouldn't be surprised if ATI has done it. It's just another form of 'cooking the books' so to speak.

But I don't put loads of faith into 3DMark 03' since it is a synthetic benchmark. It's nice to have for bragging rights and some cool benchmark tests which stress your graphics hardware's capabilities to their max, but it's not the 'end all, cure all.'

I put more faith into actual games and last time I checked, the FX 5900 still has the lead.

If you want to accuse nVidia of tweaking their drivers to deliver better performance in consumer games, be my guest. You have no case however since better performance is typically what we download the latest drivers for. ;-)

Enough said.

http://www.flightsim.com/dcforum/Use...31461d33c5.jpg

Tim Kotula
KDTW/KPTK/KPLN

XP 1800+ o/ced to XP2200+ @ 1.8 Ghz
MSI KT3 Ultra Motherboard
512 MB Crucial PC2100 DDR
OC Systems Enhanced GeForce 4 Ti4200 64 MB DDR o/ced to 350/700 Mhz.
Sound Blaster Live! 5.1
11217 3D Marks

Pentium 4 - n. A fast, but often overpriced central processing unit capable of emptying the wallets of the common people and causing much financial woe. ;-) :-lol

[Bill_Browning]

[font size="1" color="#FF0000"]LAST EDITED ON May-25-03 AT 09:54PM (EDT)[/font][p]Actually, there is an article here that says they are BOTH less than honest about optimizing for specific benchmarks.

"http://www.extremetech.com/article2/0,3973,1105259,00.asp"
(You'll have to cut and past the URL because the forum doesn't like commas.)



Best,
Bill

[Bigshot]

[font size="1" color="#FF0000"]LAST EDITED ON May-25-03 AT 10:43PM (EDT)[/font][p]Here's the re-run scores after Nvidia's hack was found out:
Say what you want about 3dMark benchmarks; but they usually tell you what the cards will do in regular games.

http://www.x-plane.org/users/mgdbottled/Hack.gif

Summary of what occurred from a Post at ATI forums:
____________________________________________________________
Nvidia - introduced multiple "cheats" for each part of 3DMark that contributed to the final score, and at least one part that doesn't.
ATI - introduced one "cheat" that contributes to the final score(might have been two, but I'm pretty sure it was just one, according to Futuremark's official report)

Nvidia - used tricks like static clip planes and no back-buffer clearing, tricks that cannot be applied in real-world gaming situations where the path of motion is not predetermined.
ATI - rewrote a shader to be more efficient on its architecture, something which can also be done in real-world gaming situations.

Nvidia - introduced cheats that altered the final image output, thereby not rendering what Futuremark wanted them to render
ATI - replaced one shader with a functionally identical piece of code that produces the exact same output as the original under all circumstances, "on the rail" or not.

Nvidia - still won't even admit that they did what they did, even though there's pretty much enough evidence to get a conviction in court. Went to far as to accuse Futuremark of just trying to make Nvidia look bad because it won't pay them money to be in their Beta program. All this over a benchmark they claim is useless and irrelavent, even though it's pretty obvious they spent a lot of time developing for it, something they claimed -on the record- that they don't do.
ATI - came clean with specifically what they were doing, and even though it produced the exact same output as the original code, acknowledged the appearance of impropriety and promised to remove the replacement code ASAP.
____________________________________________________________

[junky354]

Looks like the FX5900 is less in the benchmark, but matches up in the actual games. As seen in the tomshardware review.

no more pretty sig....

[Holliday]

Actually ATI got caught cheating not long ago. The hacked their own drivers so the Quake III test would run faster. And this wasn’t an just an allegation; they got caught.

[Aloha717200]

If they're going to cheat by tweaking the drivers to make a card run faster, why don't they just use these tweaks in all of their cards and make the faster option available to the public?

It just seems strange that a special code developed to make a game run faster, wouldn't be implemented into the card for all of us to enjoy.


http://www.flightsim.com/dcforum/Use...151b67ba1a.jpg
Brandon Williams, Aircraft developer, flightsim.com
Forum moderator: http://www.genesis-music.com

[GHD]

[font size="1" color="#FF0000"]LAST EDITED ON May-26-03 AT 06:25AM (EDT)[/font][p]But so is ATI !
They claim that the 9800 pro supports two screens, but they fail to tell you that an undocked window on the second screen does not have acceleration. An undocked VC on the second screen gives 1FPS !

[DemolitionMan]

Three days ago,Futuremark issued a patch for 3D Mark that is supposed to "compensate" for the cheats in the drivers from ATI and Nvidia.Before/after results by me: 80 Marks less!(from 9800)Go figure...
Best
Grahame EDHL

XP Pro+SP1
DirectX 9.0a
Epox 8K9A KT400
AMD XP 2700(2250MHz)
512mb Infinion DDR 333
MSI GF3 Ti200 128mb
SB Live! 5.1
WD 60GB 7200rpm
Chieftech case+Enermax PSU
Titan+Pabst cooling