Okay, fine they cheated. It was the thing to do in their situation no doubt. But the biggest problem is that they got caught and that's very bad. I think it was a bad business decision, because they know that dozens of little review sites are going to be scrutinizing their product and its performance. I guess the alternative was to release two lacklustre flagship products in a row - and that is almost suicide in this business - so made it was a justified decision. All this having been said, as a consumer I'm going think ten times before I buy a flagship Nvidia in the future.
Okay, fine they cheated. It was the thing to do in their situation no doubt. But the biggest problem is that they got caught and that's very bad. I think it was a bad business decision, because they know that dozens of little review sites are going to be scrutinizing their product and its performance. I guess the alternative was to release two lacklustre flagship products in a row - and that is almost suicide in this business - so made it was a justified decision. All this having been said, as a consumer I'm going think ten times before I buy a flagship Nvidia in the future.
well, the 5900 peforms admirably in alot of the standard game benchmarks, I just choose to disregard both the doom 3 bench and the 3dmark bench.
It's still one hell of a card, just as the 9800 pro 256 mb is a great card.
rogo
It's still one hell of a card, just as the 9800 pro 256 mb is a great card.
rogo
I'm not in any way trying to support their actions, all I'm saying is that the nv35 is a worthy opponent for the 9800pro.
Rogo
Rogo
Agreed, I think their actions are irrelevant if the product performs well without cheating. We just have to be more careful that the claimed performance is the actual performance. Being more savvy consumers we should hopefully not rely much on 3Dmark as we know its limitations.
Before I comment, how do you pause the scene in 3dmark and move around (as they did to see the "errors" with the 5900).
And does it work with 3dmark2k1?
And does it work with 3dmark2k1?
Here is my stand, right or wrong, but at least its based on something besides stupidity.
1. 3D Mark 03 is crap, useless. People can't play it, they only care about highest score.
nVidia gets the highest score. They don't screw up the scene as you see it. They cut out polys you DON'T see to increase the score. SCORE is all that matters in 03. If not, run it at 16x12 everything turned up max. To think that it is relevant for anything other than bragging about, is stupidity.
ET even said themselves: YOU CAN NOT SEE IT W/O BLAH BLAH VERSION.
Do you have it? Does anyone really? NO! That's right. Higher score = good. nVidia is providing you a service, and you spite them for it.
2. If it does it in games, and it isn't noticeable. GOOD. If they didn't do it in games. GOOD.
Why?
Because if they do it in games and it doesn't affect gameplay/iq but makes it faster, then do it. There is no cheating folks. Its software. Get there faster w/o sacrificing iq. That is the key.
This isn't the Contra code.
So see, this whole argument is pointless. 3D Marks ONLY purpose is for the score. You can say its to compare video cards. That isn't a good thing to do. You need to use GAMES to compare video cards, not 03. If nVidia can do this, so can anyone. Who knows what else they can do.
Why do you think ET took the time to find this NOW? Coincidence?
Hmm, lets fire up the Beta version of 03 and look all over for some way that nVidia is cheating. You don't just investigate stuff. Have to have a reason.
If there was a reasons how come other sites aren't on to this.
Who is to say ATi, Matrox, Trident, ANYONE doesn't do something like this in 2003? Have any sites EVEN LOOKED at the Beta version for things like this, or something entirely different?
NO. If they had, it'd make news by now.
Hopefully this will make 3D Mark 2003 irrelevant. It does to me.
The 5900 Ultra is faster in games. It has comparable image quality. That is the only thing that matters.
I don't care how they do it, I just want them to do it.
Buy ATi, its fast as hell to. By nVidia, that's great too.
This happens with lots of things (My mobo defaults to 137mhz fsb) among others.
You do what you have to do to win.
If it isn't sacrificing IQ/Compatibility/Gameplay, then DO IT.
This isn't misleading the customer, your nVidia card will do the same for you, as it does for them.
Read some. Gain an understanding. You need to realize it isn't that big of a deal, and it isn't something that hasn't been done before, or will be done again.
The 5900 Ultra and 3D mark 2003 don't mean squat at all.
nVidia will be here tomorrow, and ppl will buy their cards.
Will YOU still be complaining about it.
1. 3D Mark 03 is crap, useless. People can't play it, they only care about highest score.
nVidia gets the highest score. They don't screw up the scene as you see it. They cut out polys you DON'T see to increase the score. SCORE is all that matters in 03. If not, run it at 16x12 everything turned up max. To think that it is relevant for anything other than bragging about, is stupidity.
ET even said themselves: YOU CAN NOT SEE IT W/O BLAH BLAH VERSION.
Do you have it? Does anyone really? NO! That's right. Higher score = good. nVidia is providing you a service, and you spite them for it.
2. If it does it in games, and it isn't noticeable. GOOD. If they didn't do it in games. GOOD.
Why?
Because if they do it in games and it doesn't affect gameplay/iq but makes it faster, then do it. There is no cheating folks. Its software. Get there faster w/o sacrificing iq. That is the key.
This isn't the Contra code.
So see, this whole argument is pointless. 3D Marks ONLY purpose is for the score. You can say its to compare video cards. That isn't a good thing to do. You need to use GAMES to compare video cards, not 03. If nVidia can do this, so can anyone. Who knows what else they can do.
Why do you think ET took the time to find this NOW? Coincidence?
Hmm, lets fire up the Beta version of 03 and look all over for some way that nVidia is cheating. You don't just investigate stuff. Have to have a reason.
If there was a reasons how come other sites aren't on to this.
Who is to say ATi, Matrox, Trident, ANYONE doesn't do something like this in 2003? Have any sites EVEN LOOKED at the Beta version for things like this, or something entirely different?
NO. If they had, it'd make news by now.
Hopefully this will make 3D Mark 2003 irrelevant. It does to me.
The 5900 Ultra is faster in games. It has comparable image quality. That is the only thing that matters.
I don't care how they do it, I just want them to do it.
Buy ATi, its fast as hell to. By nVidia, that's great too.
This happens with lots of things (My mobo defaults to 137mhz fsb) among others.
You do what you have to do to win.
If it isn't sacrificing IQ/Compatibility/Gameplay, then DO IT.
This isn't misleading the customer, your nVidia card will do the same for you, as it does for them.
Read some. Gain an understanding. You need to realize it isn't that big of a deal, and it isn't something that hasn't been done before, or will be done again.
The 5900 Ultra and 3D mark 2003 don't mean squat at all.
nVidia will be here tomorrow, and ppl will buy their cards.
Will YOU still be complaining about it.
OT mini-rant: Please be forewarned that typing [H] will initiate the highlight function of the bulletin board system here. There's a trick to get around it, but I'm not sure if I should tell you or not...
Opstar
Have you been reading the posts?
This has nothing to do with playing the game.
It is all about misreprestenting the benchmarks in a HUGE demographic!
All of us would agree that time demos and 3dmark don't represent games in their quiddity, but as mentioned earlier that is not the point.
The point is that these benchmarks (that don't reflect the "truth" of a video card's performance) are published accross the web, periodicals, and reviews, the public is DECIEVED, and this influences their purchase. PR is a monstorous concern for all buisnesses and this is why it is very very bad.
The only reason why nvidia would do this is because the benchmarks are higher and therefore they can sell more cards PERIOD.
They don't expect that the hardcore users (who read video card forums) will buy into their bs (unless they don't get caught).
This has everthing to do about "joe sixpack" and not you since you are the minority.
Rogo
Have you been reading the posts?
This has nothing to do with playing the game.
It is all about misreprestenting the benchmarks in a HUGE demographic!
All of us would agree that time demos and 3dmark don't represent games in their quiddity, but as mentioned earlier that is not the point.
The point is that these benchmarks (that don't reflect the "truth" of a video card's performance) are published accross the web, periodicals, and reviews, the public is DECIEVED, and this influences their purchase. PR is a monstorous concern for all buisnesses and this is why it is very very bad.
The only reason why nvidia would do this is because the benchmarks are higher and therefore they can sell more cards PERIOD.
They don't expect that the hardcore users (who read video card forums) will buy into their bs (unless they don't get caught).
This has everthing to do about "joe sixpack" and not you since you are the minority.
Rogo
it isn't misleading. You can get the same score as they do. If you use the same driver.
I don't see why they should render part of the scene not intended to be seen in the first place.
Its efficient in my eyes.
Sure you say, the ATi card is faster w/o "cheating" or whatever.
But the nVidia card will still post the higher scores they way they render it, so the consumer is not misled.
I think its smart, and I do not think its cheating.
nVidia said a long time ago they could care less about 3D mark 03 as a relevant tool, so they are 1. proving it isn't and 2. using it for what its worth
getting a high score to show people who don't know anything about videocards that they are indeed the best.
I don't see why they should render part of the scene not intended to be seen in the first place.
Its efficient in my eyes.
Sure you say, the ATi card is faster w/o "cheating" or whatever.
But the nVidia card will still post the higher scores they way they render it, so the consumer is not misled.
I think its smart, and I do not think its cheating.
nVidia said a long time ago they could care less about 3D mark 03 as a relevant tool, so they are 1. proving it isn't and 2. using it for what its worth
getting a high score to show people who don't know anything about videocards that they are indeed the best.
and before you say its not the same benchmark.
1. anybody could do it, if they choose not too, fine.
2. it isn't a degradation of performance.
You can't see it in the normal benchmark.
1. anybody could do it, if they choose not too, fine.
2. it isn't a degradation of performance.
You can't see it in the normal benchmark.
Well, the scene is supposed to be rendered a certain way, and nvidia isn't doing it right. Why there is any disputing that point I have no idea. Whether it is due to a bug or whether it is on purpose (and of course nvidia would never admit the latter) is debatable.
Consider it this way: if a group of people had to take a test, and one person copied the answers down ahead of time and snuck them in to have while they were taking the test, would that be "efficiency" or "cheating"? Put that way, I don't see how there can be any mistake (unless it truly was a driver bug).
Not for long. Sorry, but I just don't agree with your laxity regarding ethics. And I don't care if "everyone's doing it," because ATi isn't, and they're the only other competitor in this particular benchmark. Thus half the people are not cheating.
No, it'll force everyone to waste time devising new, painstaking ways to avoid this sort of thing that they could better spend investigating IQ or testing mroe than Quake 3 and UT2K3 or doing all sorts of things. I don't mean to be melodramatic, but I can't think of another way of putting this: your reasoning is sort of like saying it's good that they crashed a plane into the WTC, because now we'll look out for that sort of thing. No. It's no good for anyone any way you look at it. And don't forget we can use the same logic you used with ET's article: nVidia likely would NOT have corrected their cheat if ET or someone else hadn't exposed them (like with the overflow of horrendous MS bugs you read about every month, where ppl are forced to take it public due to weeks of MS silence/inaction).
Then you do not understand the point of 3DMark at all, and you shouldn't be voicing your opinion in this thread if it's uninformed. 3DMark is USELESS if not all cards render the same data. It is merely a test of raw horsepower. If you don't know what it's for, don't use it and don't pay attention to it. But don't try to corrupt it into something it isn't, which is a test of how well a developer can optimize a game for a certain card--because most D3D developers will not take that time, and you have other benchmarks (like UT2K3) to test for that.
Try reading the previous posts next time. I know there are a lot of them, but I'm pretty sure this point has been made early and often, and it'll prevent more uninformed posts like yours. I'm not saying stupid or idiotic or unbelievably offensive, just ignorant, so don't take offense. You don't have the whole facts, so you shouldn't be replying in an uninformed way in a thread that's already full of that sort of thing--it helps no one.
It isn't necessarily rendered, but it is supposed to pass through a certain process to be eliminated; this process is supposed to be done on the video card (requiring processing power, as it does in all real life games). Nvidia has given their video card the unfair advantage of knowing the answers ahead of time, so they don't have to go through the process. Thus, their video cards are not stressed in the same way as other manufacturers' video cards are, and the test is not fair.
Pete-
Found it interesting that these two quotes were next to each other Under Doom3, the NV30 is an eight pixel board.
I posted a question over at B3D and perhaps someone there can answer it, I don't see the clipping issue possible on game benchmarks in any way that could be remotely hidden. BTW- The SeriousSam issue is fixed with an adjustment to the game cfg file, performance seems to be completely unchanged.
Found it interesting that these two quotes were next to each other
Under Doom3, the NV30 is an eight pixel board. I posted a question over at B3D and perhaps someone there can answer it, I don't see the clipping issue possible on game benchmarks in any way that could be remotely hidden. BTW- The SeriousSam issue is fixed with an adjustment to the game cfg file, performance seems to be completely unchanged.
Right, their tactic would be legit if they did such "optimizations" for every game ever made. As it stands a video card needs to go through all those steps to render any game that you might use with it.
so whats the latest word since ET posted that article? anything from nvidia yet?
i'm surprised that not many other websites have checked this out but have been clear
to foolishly knock the claim (hardocp).
i'm surprised that not many other websites have checked this out but have been clear
to foolishly knock the claim (hardocp).
Mem
Lifer
- Apr 23, 2000
- 21,476
- 13
- 81
Pete I think you`re making too much importance in 3Dmark benchmark and its software,I can tell you for a fact most gamers don`t take it serious and if it went the way of the dodo tomorrow I for one wouldn`t lose any sleep.
The $64,000 question is did Nvidia really cheat or not? I for one still can`t say 100% if they did or not.Also officially Nvidia doesn`t have access to the 3Dmark developer version software since they refuse to pay for it , so does that really make any difference to their driver coding,I still can`t say for sure.
I was thinking if Nvidia did cheat,it means that 3Dmark performance is important to them so why didn`t they just pay for the 3Dmark developer version so they could really tweak their drivers to get the best out of it without cheating?
The $64,000 question is did Nvidia really cheat or not? I for one still can`t say 100% if they did or not.Also officially Nvidia doesn`t have access to the 3Dmark developer version software since they refuse to pay for it , so does that really make any difference to their driver coding,I still can`t say for sure.
I was thinking if Nvidia did cheat,it means that 3Dmark performance is important to them so why didn`t they just pay for the 3Dmark developer version so they could really tweak their drivers to get the best out of it without cheating?
Not for long. Sorry, but I just don't agree with your laxity regarding ethics.
So you're saying you cannot disagree while remaining civil? Interesting...
your reasoning is sort of like saying its good that they crashed a plane into the WTC, because now we'll look out for that sort of thing.
I now see why you said this wasn't going to be civil...
I personally can't justify equating the deaths of thousands of innocent civilians to a device driver which "cheats" on a synthetic benchmark. To each their own, I suppose.
Despite this horribly offensive analogy, I understand the point you are trying to make and it's still rubbish.
nVidia cheating on a synthetic benchmark (which has been validly argued to NOT be a fair representation of overall game performance in current and future titles) raises awareness to the fact that the benchmark is NOT a good overall representation of game performance in current and future titles. Why? Because if it was, it would be more difficult to include driver optimizations which improve ONLY 3Dmark scores and NOT performance in actual game titles. Yet all IHV's do "cheat" in one way or another.
As far as "cheating" goes, I see no difference in what nVidia has done to what ATI did in their original Catalyst release (when 3Dmark 2001 scores jumped considerably but actual game performance remained the same). I thought ATI did a great thing there, because it began to raise awareness to the fact that 3dmark scores might not be telling you the whole story.
Then you do not understand the point of 3DMark at all, and you shouldn't be voicing your opinion in this thread if it's uninformed. 3DMark is USELESS if not all cards render the same data. It is merely a test of raw horsepower. If you don't know what it's for, don't use it and don't pay attention to it. But don't try to corrupt it into something it isn't, which is a test of how well a developer can optimize a game for a certain card--because most D3D developers will not take that time, and you have other benchmarks (like UT2K3) to test for that.
You exemplify my argument. You say that 3Dmark is useless if cards do not render the same data, and that developers should only optimize for actual games. You should have taken the blue pill. In the real world, 3Dmark has become a benchmark which can make or break the sales of a GPU. Do you think that nVidia will decide "ahh, let's just make sure consumers have fast stable performance in all games and not worry about what Dell thinks about our low 3dMark scores!" ?
It would be great to have a fair "universal" synthetic benchmark which can be used for comparing GPU's, apple to apple, across platforms... but it will never happen. The only good benchmark is an actual game which allows the user to record timedemo's. This will always be a valid benchmark and it will drive optimizations which will only improve the end user experience.
So you're saying you cannot disagree while remaining civil? Interesting...
your reasoning is sort of like saying its good that they crashed a plane into the WTC, because now we'll look out for that sort of thing.
I now see why you said this wasn't going to be civil...
I personally can't justify equating the deaths of thousands of innocent civilians to a device driver which "cheats" on a synthetic benchmark. To each their own, I suppose.
Despite this horribly offensive analogy, I understand the point you are trying to make and it's still rubbish.
nVidia cheating on a synthetic benchmark (which has been validly argued to NOT be a fair representation of overall game performance in current and future titles) raises awareness to the fact that the benchmark is NOT a good overall representation of game performance in current and future titles. Why? Because if it was, it would be more difficult to include driver optimizations which improve ONLY 3Dmark scores and NOT performance in actual game titles. Yet all IHV's do "cheat" in one way or another.
As far as "cheating" goes, I see no difference in what nVidia has done to what ATI did in their original Catalyst release (when 3Dmark 2001 scores jumped considerably but actual game performance remained the same). I thought ATI did a great thing there, because it began to raise awareness to the fact that 3dmark scores might not be telling you the whole story.
Then you do not understand the point of 3DMark at all, and you shouldn't be voicing your opinion in this thread if it's uninformed. 3DMark is USELESS if not all cards render the same data. It is merely a test of raw horsepower. If you don't know what it's for, don't use it and don't pay attention to it. But don't try to corrupt it into something it isn't, which is a test of how well a developer can optimize a game for a certain card--because most D3D developers will not take that time, and you have other benchmarks (like UT2K3) to test for that.
You exemplify my argument. You say that 3Dmark is useless if cards do not render the same data, and that developers should only optimize for actual games. You should have taken the blue pill. In the real world, 3Dmark has become a benchmark which can make or break the sales of a GPU. Do you think that nVidia will decide "ahh, let's just make sure consumers have fast stable performance in all games and not worry about what Dell thinks about our low 3dMark scores!" ?
It would be great to have a fair "universal" synthetic benchmark which can be used for comparing GPU's, apple to apple, across platforms... but it will never happen. The only good benchmark is an actual game which allows the user to record timedemo's. This will always be a valid benchmark and it will drive optimizations which will only improve the end user experience.
3DMark WILL BE totally useless unless ALL cards render the scenes using drivers that are not written specifically to inflate the scores. Instead of the WTC comparison, think of the Consumer Reports 1996 Isuzu Trooper fiasco. Isuzu Motors Ltd. accused Consumers Reports of rigging tests to show that the 1995-96 Trooper sport-utility vehicle displayed a propensity to roll over when making emergency turns. Isuzu claims it suffered more than $200 million in damages after Consumer Reports gave an "unacceptable" rating to the Trooper in October 1996, causing a 54 percent drop in sales. Videos shown later by Isuzu showed that Consumer Reports test drivers swung wider around the test cones than with other SUVs. This made the test driver have to swerve the vehicle harder and causing it to tip onto two wheels.
So yes, the playing field needs to be level for ALL contestants for it to be a true objective test.
So yes, the playing field needs to be level for ALL contestants for it to be a true objective test.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter