nVidia's 53.03 and 52.16 drivers in 3dMark2003

[digitalwanderer]

Hmm interesting. This for a ASUS 5900U

Results from early December, system is overclock in these tests.

52.16 - 4946
53.03 - 5719

Result from 10 minutes ago, system is not overclocked:

52.16 - 4371
53.03 - 4367

I don't know how to explain this. Both drivers where downloaded from last year as well as build 340.

Any other major system changes? Different dx version, tweakage, anything?

That's just goofy. :|

 

[Khedlar]

Ok my faith is renewed. When running this benchmark set sliders to the left and turn off vertical sync. Latest results

52.16 - 4394
53.03 - 5134


I think when we see test that show the 2 drivers with the same performance level, at least in the cases of Nvidia they made the same mistake I did.

 

[illuminati]

Ok my faith is renewed. When running this benchmark set sliders to the left and turn off vertical sync. Latest results

52.16 - 4394
53.03 - 5134


I think when we see test that show the 2 drivers with the same performance level, at least in the cases of Nvidia they made the same mistake I did.

Setting sliders to the left... does that mean set to "performance" and away from "quality"? We usually set the sliders all the way to "Quality" and also turn vertical sync off.

Very interesting though... Thanks.

 

[kyleb]

either way, i think it is pretty obvious that vsync was the real issue.

 

[Tim Murray]

Dig, I almost said, "NVIDIA IS CHEATING WITH A TIMER TO CONFUSE US ALL AND DISCREDIT US!" But then I woke up from my crazy fever dream

 

[digitalwanderer]

Dig, I almost said, "NVIDIA IS CHEATING WITH A TIMER TO CONFUSE US ALL AND DISCREDIT US!" But then I woke up from my crazy fever dream

Wow, and people call me paranoid!

 

[Tim Murray]

Dig, I almost said, "NVIDIA IS CHEATING WITH A TIMER TO CONFUSE US ALL AND DISCREDIT US!" But then I woke up from my crazy fever dream

Wow, and people call me paranoid!

I'm not paranoid--I just couldn't go to sleep as a kid because I thought Lee Harvey Oswald was in my closet.

 

[digitalwanderer]

Dig, I almost said, "NVIDIA IS CHEATING WITH A TIMER TO CONFUSE US ALL AND DISCREDIT US!" But then I woke up from my crazy fever dream

Wow, and people call me paranoid!

I'm not paranoid--I just couldn't go to sleep as a kid because I thought Lee Harvey Oswald was in my closet.

No, no, no....that was me silly! 8)

 

[Tim Murray]

Dig, I almost said, "NVIDIA IS CHEATING WITH A TIMER TO CONFUSE US ALL AND DISCREDIT US!" But then I woke up from my crazy fever dream

Wow, and people call me paranoid!

I'm not paranoid--I just couldn't go to sleep as a kid because I thought Lee Harvey Oswald was in my closet.

No, no, no....that was me silly! 8)

:O Dig is Michael Jackson?

I will never sleep again.

(see, aren't these silly little threads more fun than "Futuremark is stupid!" "NVIDIA is evil!" "Futuremark is stupid!" "You're stupid!" "You smell like poo!" ?)

 

[cthellis42]

But some of us DO smell like poo!

 

[ByteMe]

But some of us DO smell like poo!

Great, now people are gona start throwing the shithead remark around.

 

[illuminati]

Ok my faith is renewed. When running this benchmark set sliders to the left and turn off vertical sync. Latest results

52.16 - 4394
53.03 - 5134


I think when we see test that show the 2 drivers with the same performance level, at least in the cases of Nvidia they made the same mistake I did.

I still have doubts. Is the 53.03 driver you got these results from the WHQL version? or did you happen to get this driver before it was an official driver from nVidia like in this Elite Bastards article?

Is it possible for you to download a new copy of the 53.03 driver from nVidia's site if you happen to have a pre-WHQL version?

 

[Khedlar]

They are the WHQL version downloaded from the Nvidia site on 12/9/03.

 

[Tim Murray]

Ok my faith is renewed. When running this benchmark set sliders to the left and turn off vertical sync. Latest results

52.16 - 4394
53.03 - 5134


I think when we see test that show the 2 drivers with the same performance level, at least in the cases of Nvidia they made the same mistake I did.

I still have doubts. Is the 53.03 driver you got these results from the WHQL version? or did you happen to get this driver before it was an official driver from nVidia like in this Elite Bastards article?

Is it possible for you to download a new copy of the 53.03 driver from nVidia's site if you happen to have a pre-WHQL version?

I'm confused by the doubting--if you read the 53.03 changelog, it specifically mentions:

Re-enabled NVIDIA's compiler technology for 3DMark03 Patch 340
performance.

Alas, we know that the claims of 3DMark03 build 340 disabling of the shader compiler were completely false. NVIDIA includes "hand-tuned shaders" in their catch-all of the shader compiler.

 

[illuminati]

I'm confused by the doubting--if you read the 53.03 changelog, it specifically mentions:

Re-enabled NVIDIA's compiler technology for 3DMark03 Patch 340
performance.

Alas, we know that the claims of 3DMark03 build 340 disabling of the shader compiler were completely false. NVIDIA includes "hand-tuned shaders" in their catch-all of the shader compiler.

Thanks. That bit of information and the fact that Khedlar is using the WHQL version, pretty much stops my doubts.

And thanks to everyone who took time out of their day to run a few tests in order to ease the mind of a crazy man!

 

[illuminati]

Ok... my nVidia Ref 5900 Ultra results are in...

52.16 - 5371
53.03 - 6020

On AMD FX-51 system with v-Sync turned off and "Quality" setting in both cases.

Thanks again, all!

 

[digitalwanderer]

Ok... my nVidia Ref 5900 Ultra results are in...

52.16 - 5371
53.03 - 6020

On AMD FX-51 system with v-Sync turned off and "Quality" setting in both cases.

Thanks again, all!

So Justin, would you say that nVidia HAS re-enabled cheats in the 53.03 drivers then?

 

[illuminati]

Ok... my nVidia Ref 5900 Ultra results are in...

52.16 - 5371
53.03 - 6020

On AMD FX-51 system with v-Sync turned off and "Quality" setting in both cases.

Thanks again, all!

So Justin, would you say that nVidia HAS re-enabled cheats in the 53.03 drivers then?

Yes, I would agree that nVidia has re-enabled cheats in the 53.03 driver. My mind is at ease on this issue now... Not happy, but content. I think everyone, including myself, now knows the truth.

I will be posting another editorial summarizing all my research on this issue and forming a few conclusions.

Thanks again for everyone's help.

 

[demalion]

Popularity contests are much simpler than technical details

You heard it here fi...errr...Yet Again: Benchmark results are a matter of who you favor, not hardware and drivers.

Tweaktown's Julian Shapiro and/or Cameron Wilmot[/url]]
Drivers Used: ATi 3.10 and nVidia 53.03
...
Please Note: Due to recent events with the 3DMark03 series, we are adding results purely for those who are still in favor of 3DMark03. These results should not be taken too seriously and are only added for interest sakes.

Conclusions
...
As many of you already are aware of, it’s difficult to select a champion in a race with such close combatants. Both underlying GPU technologies from ASUS are cutting edge and in the end usually provide quite similar results overall -in both real-world and synthetic benchmarks.
...

Shapiro is the article author, Wilmot is the editor of record.

 

[Mendel]

Possible future nightmare scenario:

1) send futuremark a driver with no cheats in 3dmark and let Futuremark review them and conclude no cheats are in there.

2) wait a week

3) release the driver for reviewers, with same version number, only with re enabled cheats.

4) leak cheating driver version for public through board partner

5) release a non cheating driver

Results:

Futuremark has approved a driver but it still gives cheat results for reviewers.


Think this might happen`?