nVidia takes ATI's place as dirty rotten cheats

[meglamaniac]

Yet another reason to never put any faith in benchmarks.

I can't say I'm surprised, really...

[Leshy]

I believe the biggest performance gain was made by programming into the driver exactly which parts of the 3Dmark scene to render, eg. those that are actually shown during the standard camera flyby.

ExtremeTech discovered it by using a customized version of 3Dmark which allows for the camera to be rotated during scenes, only to find that the nVidia card suddenly screwed up all the graphics.

Not sure if this ranks above or below ATi's Quake 3 stunt.

[meglamaniac]

I'd say above in terms of pure underhandedness and, even sadder, the effort involved.

"All" ATI did was look for a process running called quake.exe or whatever and then force the filtering settings etc to low - they never actually tampered with the rendering itself.

ffs - how do either of them expect to get credibility when they pull these sorts of stunts?

[Leshy]

It's not about credibility. It's about framerates and 3Dmark score. :/

[Nodrog]

Originally posted by Leshy
Not sure if this ranks above or below ATi's Quake 3 stunt.

Its not as bad, because real world performance based on ingame benchmarks (such as Quake framerate) are both important and significant, whereas only dumb people who want to show off about their computer's penis size care about 3Dmark scores.

[Gayle29uk]

Originally posted by Nodrog
Its not as bad, because real world performance based on ingame benchmarks (such as Quake framerate) are both important and significant, whereas only dumb people who want to show off about their computer's penis size care about 3Dmark scores.

Say what? ATI and nVidia care about 3Dmark because that's the main selling factor for their cards (to most people).

[Nodrog]

Originally posted by Gayle29uk
Say what? ATI and nVidia care about 3Dmark because that's the main selling factor for their cards (to most people).

If the primary thing you look at when buying a graphics card is 3dmark scores, then you deserve what you get. If you have a clue about anything, youd look at performance in real world apps, which is why what ATi done is worse.

[Gayle29uk]

Originally posted by Nodrog
If the primary thing you look at when buying a graphics card is 3dmark scores, then you deserve what you get. If you have a clue about anything, youd look at performance in real world apps, which is why what ATi done is worse.

I have a Voodoo 3 3000, I can't even RUN 3Dmark

[Leshy]

Originally posted by Nodrog
Its not as bad, because real world performance based on ingame benchmarks (such as Quake framerate) are both important and significant, whereas only dumb people who want to show off about their computer's penis size care about 3Dmark scores.

I'm inclined to agree that cheating by affecting a game is worse than cheating by affecting a benchmark program.

However, 3Dmark is a utility used by many people to determinate a card's capabilities, and high 3Dmark ratings do affect sales. nVidia completely reprogramming pixel shaders and basically pre-rendering scenes in order to boost there score by a stunning 25% margin, thus indicating the card to have capabilities far beyond what it has, is very serious.

ATi on the other hand simply forced low quality rendering by overriding the game settings. The performance gained from this move was not nearly as big, and the lower quality was quite evident upon playing - in effect ATi painted a picture of trading image quality for speed.

As such, I'm undecided which of these two is actually worse.

[SbOlly]

As far as I'm concerned, NVidia and ATI can cheat all they like on the 3dMark scores. They didn't ask to be judged that way, rather they simply are judged by people for it. So who can blame them for boasting their scores?

But, for ATI to make a change that meant a game ran on one of their cards, used inferior settings to simply increase the frame rate, means they are ripping off the customers who bought their cards in good faith.


(Plus I'm a NVidia fanboy. They gave me a 5800 FX for free. Gotta love em. ATI on the other hand have giving me jack!)

[pablissimo]

Originally posted by SbOlly
(Plus I'm a NVidia fanboy. They gave me a 5800 FX for free. Gotta love em. ATI on the other hand have giving me jack!)

Free you say...?

[Leshy]

Originally posted by pablissimo
Free you say...?

No one would buy them.

[pablissimo]

Originally posted by Leshy
No one would buy them.

I'd happily accept one if their stocks needed shifting...

Actually, it wouldn't fit in this wee Shuttle of mine =/

[SbOlly]

Originally posted by Leshy
No one would buy them.

It was because I signed up for one of their dull developer conference's (scarily in MoS).

I didn't go, but they sent me one anyway. Thus the fanboy-ness

[Leshy]

Originally posted by pablissimo
I'd happily accept one if their stocks needed shifting...

Me too, but that's because I'm on a GF4 MX440.

[MrL_JaKiri]

Originally posted by SbOlly
As far as I'm concerned, NVidia and ATI can cheat all they like on the 3dMark scores. They didn't ask to be judged that way, rather they simply are judged by people for it. So who can blame them for boasting their scores?

I didn't asked to be judged by the law of this country, but it doesn't make it right for me to bribe a jury.

It's an amoral method to boost your scores; it just means that you feel you need to cheat your way to the top rather than get there by offering the better hardware.

[Leshy]

ATi has now admitted that they have been using driver optimalization to produce better results in 3Dmark2003 as well (boosting their score by 2%), although they claim that they are using exactly the same sort of optimizations that they would use for games, and as such the performance would still accurately reflect their card's performance in normal use.

They will remove the optimizations in the next version of the Catalyst drivers to avoid any confusion.

http://www.beyond3d.com/#news6040