Reply from nVidia about the 3DMark anomaly

[BoardBonobo]

I emailed nVidia and got the following response (which is more than I though I'd get ):

This was a bug with our drivers. The bug has been fixed. We are sorry but
we can not get into the details of the bug however it was not a cheat. The
fix should be introduced in our next driver release.

NVIDIA Support

-----Original Message-----
From:
Sent: Saturday, June 15, 2002 3:16 AM
To: SW-Info
Subject: 3DMark scores

Hi,
I was wondering whether you could help clarify the issue that has
been
raised with regards to the performance difference that occurs in 3dmark 2001
se when splash screens are disabled. There is quite a furore occuring in a
number of forums as to whether this is a cheat (it happens only in 3dmark)
or simply a performance optimisation that affects all DX based code.

Any reply would be helpful even if it's a simple no ).


Lead Programmer MLRDU
The Open University
Milton Keynes
UK

 

[Doomtrooper]

It was a optimization

If it was no big deal why the 'hush hush'...

 

[BoardBonobo]

It'll be interesting to see what the *fixed* drivers perform like though. It kind of invalidates quite alot of Madonions hall of fame, but I wonder what kind of affect it will have on DX based games?

 

[Hellbinder]

Id say no effect at all on games.

From what I understand, What they do when the splash screen is detected is simply clear all their caches, buffers etc so the card gets a fresh start for each new portion of the test. this does indeed give the card an extra little boost, but does not adjust lod or such. It is not the blackest of cheating but it still goes againt the nature of the test. Unless of course the next version of 3dmark has all cards do the same thing.

 

[Ailuros]

Maybe 3dmark2002 will have ONLY splash screens.....

 

[Doomtrooper]

I'm not betting nothing I almost had to eat my words with the last bet with Nvidia Cg

 

[Hellbinder]

we can not get into the details of the bug however it was not a cheat

On second thought.. I simply cant accept this as the truth. To me they should just admit what they did and move on. There is simply no bug on earth that will add 800 points to a test just because a splash screen is detected....

IMO This is far more deceptive than what ATi did with quack. It shows you just how Nvidia really works. Cry foul with the rest of the Nvidia favoring sites, while at the same time hidding optimizations with tricky methods.

Let this be a lesson to all you Nvidians out there. Just because you cant find it with an ASCII search... Doesn't mean its not there.

One last thing. I dont think that what they did is such a big deal. what I dont like is the obvious lie, and the further revelation of the net wide doubble standard that most simply refuse to acknowledge.

 

[dksuiko]

Hellbinder for President!

 

[Typedef Enum]

I'm just glad that you've clarified everything for the whole group Hellbinder...

I mean, if you say it was a lie...by God, it must be!

 

[BoardBonobo]

To be fairly honest I'm pretty suprised that I got a reply at all. I mean if it was a *real* cheat you'd think they'd just keep quiet and fix it later. There's a certain smattering of honesty in the statement which comes through in just replying to an average joe off the street. It's not like I got any the credentials to demmand a reply.

I think I'll reserve judgment till the next driver release and see whether the score drops with the splash screens on or whether it increases without them on. Madonion seem to think the latter will occur. But then again that's just idle rumour and speculation and we have no truck with that do we

 

[Crusher]

You all seem fairly certain that the score with the screens on will drop. Is this because you have factual evidence that that will be the case, or is it just based on your negative perception of NVIDIA?

edit: started typing this before the last couple of posts, glad to see some people are patient enough to see the results before making judgements

 

[Doomtrooper]

To be fairly honest I'm pretty suprised that I got a reply at all. I mean if it was a *real* cheat you'd think they'd just keep quiet and fix it later. There's a certain smattering of honesty in the statement which comes through in just replying to an average joe off the street. It's not like I got any the credentials to demmand a reply.

I think I'll reserve judgment till the next driver release and see whether the score drops with the splash screens on or whether it increases without them on. Madonion seem to think the latter will occur. But then again that's just idle rumour and speculation and we have no truck with that do we

Why the name change SurfMonkey ??

 

[Fred]

They should rename these boards to BeyondNvda.

Everyone here seems to think they have a better idea about what the company does than the employees themselves.

(Oh of course, All Nvidia employees lie to protect deep and dark secrets)

 

[Doomtrooper]

Yes I know Nvidia plays the game 'squeaky clean' just like all big corporations.

Doh !! how'd that get leaked

Go live in your fantasy world

 

[Fred]

Right, so you pull out an internal PR document to prove your point.

Gads, that proves me wrong.

Point is. If a company claims a certain performance. I'm skeptical. If a company says there squeecky clean and doesn't cheat. I would be skeptical (see David Kirks comments).

That a major company goes on record and claims that a specific issue is a certain way. Then I start accepting it more. Why? B/c the particular issue is ultimately easily verifiable. The loss of credibility would be enormous if they were found out, right after saying the opposite. It would be a lose lose situation for them, were that the case.

Has it happened before? Sure. 'I'm not a crook' by the late, great R. Nixon. But that was over something that was of vital importance to him. In this case, Nvidia has nothing to gain by stating what they emailed.

So, I've now switched my skepticism onto people like you. And frankly, I find it very easy to believe that you're mistaken.

 

[Oompa Loompa]

Yes I know Nvidia plays the game 'squeaky clean' just like all big corporations.

Anyone who thinks most companies don't badmouth their competitors behind closed doors is very, very naive.

 

[Joe DeFuria]

I think that's more or less Doom's point. He's just trying to convince ceratin nVidia faithful that "most companies" includes nVidia. Whether it's driver "optimizations", bad-mouthing bechinf closed dorrs (or out in the open, for that matter), etc.

 

[Doomtrooper]

Its different to badmouth behind closed doors vs. sending out trash PR to your resellers .. and thats what that is TRASH PR... lies, lies and more lies.ohhh and lies
Sure this is a internal document, this wasn't just meant for Nvidia employees to see either was it...Dell, Compaq, Boardmakers and the list goes on.
My point is if some people get a 'bad feeling' about Nvidia business practices, they have a good reason to do so

 

[OpenGL guy]

Thanks, Doomtrooper, for proving that ATI is number one!

 

[Doomtrooper]

I better stop I'm getting flamed by the 'Superior Intelligence' people on this forum in the off-topic section...
I'm just gonna play checkers and shoot some pidgeons