Futuremark Announces Patch for 3DMark03

vb said:
digitalwanderer said:
Uhm, what do you mean by "fail"? Does 3dm2k3 not run or is the scoring whacked? (Sorry, but I ain't got an FX so I really don't know what you're talking about and I want to.)

It will not do aniso as requested on texture stages 1-7

:?: What makes you say that?

FUDie said:
StealthHawk said:
So scores from the PS 2.0 test are actually legit?
How do you figure? With the results Tridamn has shown, the 5950 scores significantly lower results then the 9800 XT with the two shaders shown here. Yet the PS 2.0 results put the 5950 very close to the 9800 XT.

-FUDie

It was late when I read his post. Just ignore that stupid comment I made. The scores are clearly not legit :oops:
 
from futuremarks (updated) approved drivers page:

The 52.16 drivers have 3DMark03 specific optimization for the Pixel Shader 2.0 test and that score is solely comparable between nvidia cards.

Why do they still list it as approved?!
Is it like it is okay to cheat in the benchmark if the cheats were undetected at first but revealed later?
 
Mendel said:
from futuremarks (updated) approved drivers page:

The 52.16 drivers have 3DMark03 specific optimization for the Pixel Shader 2.0 test and that score is solely comparable between nvidia cards.

Why do they still list it as approved?!
Is it like it is okay to cheat in the benchmark if the cheats were undetected at first but revealed later?
Eeew! This is major suckage on FM's part!

Well, apparently the 52.16 drivers that have been out for months have cheats in them for the PS 2.0 test. Since this is a flagrant violation of our rules we're forced to only recomend that PS 2.0 tests be compared amongst FX cards.

How bloody hella-lame is that? :oops:
 
I guess the theory is that because it isn't a cheat in one of the tests that comprise the final 3DMark score, they are willing to let it slide this time.

Which is a bad decision in my opinion, it sets a poor precedent - Better to withdraw certification for that driver altogether.

Funnily enough though, nVidia seems to have removed the Pixel Shader 2.0 test optimisations from the 53.03 drivers - At least, that test is scoring much lower than it was in the 52.16s.
 
Hanners said:
I guess the theory is that because it isn't a cheat in one of the tests that comprise the final 3DMark score, they are willing to let it slide this time.

Which is a bad decision in my opinion, it sets a poor precedent - Better to withdraw certification for that driver altogether.
And have a public flogging of their CEO and PR dept.... :devilish:
 
I will admit that I have not been keeping up on this.

Is nvidia still cheating where Futuremark catches them and then Futuremark is STILL being a wimp?

Is this what is happening?
 
ByteMe said:
I will admit that I have not been keeping up on this.

Is nvidia still cheating where Futuremark catches them and then Futuremark is STILL being a wimp?

Is this what is happening?
It's kind of different now. FM put out a new set of guidelines after the last fiasco to try and regain their respectability which explained what they would do if any IHV broke their guidelines.

The beta 53.03 set of dets broke the guidelines (cheated), but since they were beta it was no biggy since FM only accepts official drivers.

Now that the official 53.03 set is out and appears to still contain the cheats it gets a bit more interesting. Apparently FM's rules now give nVidia one week to either remove or explain the cheats to FM's satisfaction before _________ happens. (I don't know what ______ is...I'm hoping they took my recomendations for public floggings. )

At least, that's where I THINK we are now. :)
 
digitalwanderer said:
Now that the official 53.03 set is out and appears to still contain the cheats it gets a bit more interesting. Apparently FM's rules now give nVidia one week to either remove or explain the cheats to FM's satisfaction before _________ happens. (I don't know what ______ is...I'm hoping they took my recomendations for public floggings. )

At least, that's where I THINK we are now. :)

That is correct as far as I know.

There are one of three possibilities:

1) Futuremark reviews the new drivers, finds them legit, and puts them on the "reviewed and approved" list. (Anyone think this is gonna happen?)

2) Futuremark reviews the new drivers, finds the same (or some of the same) illigitimate issues as last time. Creates ANOTHER patch for 3DMark, and puts the new drivers on the "reviewed" list.

3) Futuremark reviews the new drivers, finds the same issues, does not release a patch, but puts the drivers on the "black list."

There are pros and cons to approaches 2 and 3...I'm not sure which one I'd prefer...
 
Wow, is it me or is Futuremark doing everything they can not to confront Nvidia?

Futuremark has got to learn to stand it's ground. If they did I think they would get overwhelming public support. Atleast they would have a few months back.

It seems to me that someone at Futuremark needs to grow some balls then stand and FIGHT! Business can be war and so far Futuremark fields a small army of teenage girls with powderpuff weapons.

Am I completely off?
 
ByteMe said:
Wow, is it me or is Futuremark doing everything they can not to confront Nvidia?

Futuremark has got to learn to stand it's ground. If they did I think they would get overwhelming public support. Atleast they would have a few months back.

It seems to me that someone at Futuremark needs to grow some balls then stand and FIGHT! Business can be war and so far Futuremark fields a small army of teenage girls with powderpuff weapons.

Am I completely off?
Not completely, but a bit....but don't feel too bad 'cause I share your opinion. FM seems to have grown a set of nuggers since the last brew-ha-ha with nVidia, but they still have to give nVidia a chance to correct the "problem" before they can take action.

I'm hoping for a patch meself, I wanna see the "patch-new cheat drivers-patch-new cheat drivers" pattern happen for a bit just to drive home the fact that nVidia is still cheating.

SIDENOTE: Just to keep you up to snuff, nVidia is claiming that optimizing for 3dm2k3 is legitimate because it's a "synthetic gaming benchmark" which makes it ok to optimize for just like a game. The only problem with that is it makes absolutely no sense and no one in their right mind agrees with 'em. (Well, at least FM and most of the folks here don't. :) )

Just thought I'd kick ya up to speed, it's about to get fun in a few days again. :D
 
Joe DeFuria said:
There are one of three possibilities:

1) Futuremark reviews the new drivers, finds them legit, and puts them on the "reviewed and approved" list. (Anyone think this is gonna happen?)

2) Futuremark reviews the new drivers, finds the same (or some of the same) illigitimate issues as last time. Creates ANOTHER patch for 3DMark, and puts the new drivers on the "reviewed" list.

3) Futuremark reviews the new drivers, finds the same issues, does not release a patch, but puts the drivers on the "black list."

There are pros and cons to approaches 2 and 3...I'm not sure which one I'd prefer...

As far as I see it, option three is the way to go - Another patch can be circumvented by nVidia in about a week, whereas there is nothing they can do to get around having blacklisted drivers.
 
What would the effects be of their drivers being blacklisted? Couldn't nVidia just claim that FM is unfairly prejudiced and insist that review sites use the blacklisted drivers anyway as they represent how REAL games will play on their hardware? (I don't think that way, but that's what I think nVidia will say.)

We've been hearing about BB's strong-arming & blacklisting in that other thread, what's to keep them from continuing it?
 
digitalwanderer said:
What would the effects be of their drivers being blacklisted? Couldn't nVidia just claim that FM is unfairly prejudiced and insist that review sites use the blacklisted drivers anyway as they represent how REAL games will play on their hardware? (I don't think that way, but that's what I think nVidia will say.)

We've been hearing about BB's strong-arming & blacklisting in that other thread, what's to keep them from continuing it?

Well, for one thing scores from blacklisted drivers aren't supposed to show up in FM's rankings. As for review sites, I dunno if they are under any obligation to follow FM's rules about what drivers to use. I suppose the ones that would be most likely to ignore FM are the same ones that wouldn't use 3DMark in a review anyway :)
 
Hanners said:
As far as I see it, option three is the way to go - Another patch can be circumvented by nVidia in about a week, whereas there is nothing they can do to get around having blacklisted drivers.

There are, however, cons to that approach.

The most significant being web reviewer inconvenience. If 3DMark just "black lists" the drivers, web reviewers can still run "the latest 3D mark" patch, and get results...just not publish them to the database.

Reviewers are not going to "like" having to install a specific (and older) set of nvidia drivers just to benchmark 3DMark. So I'd predict one of two things could happen:

1) Reviewers just don't use 3DMark
2) Reviewers just use 3DMark with black-listed drivers.

Both of which are obviously bad.

And rather than be mad at nVidia for this situation (and refuse to review their card), they'll just stop using 3DMark.

By going with the "new patch" approach, at least web reviewers can freely use the "most up to date" drivers and 3DMark patch. The downsides to this approach:

1) More work for FutureMark
2) Slightly more inconvenience for everyone by having to download new 3DMark patches with every official driver release
3) The point of nvidia's drivers being "cheaters" is not as effectively driven home, IMO, vs. blacklisting drivers.
 
Right Joe, nV puts Futuremark in somewhat of a no win situation. What I would like to see is a rule from FutureMark saying that any site that uses 3Dmark will publish the results as they are seen on screen. This would allow FutureMark to add a banner across that states something like "FutureMark has not verified if these drivers contain Illegal optimizations".
 
I was talking with Patric about it, and he let me know when they added that bit. Seems the PS2.0 test was still being looked at after they changed the version to cover the game tests properly, since they needed the primary attention first. Releasing ANOTHER version for the PS2.0 fix would be detrimental, though, so they can just inform people as possible.

Hardly matters at this point, though. How they react to the newly-WHQL'd 53.03's are what really matters right now. PS2.0 is an unfortunately side-concern, but of much less importance than what they do regarding nVidia's new drivers. (Since by all accounts I've heard so far the optimizations are back in, right?)
 
This just goes to show that the patch method is as flawed now as it was when build 330 was put out. Seems nVidia aren't the only ones breaching their guidelines.
 
Back
Top