Their reviewer actually has one with him (picked it up at the nVidia NV35 presentation at Santa Clara), it's just that it's at E3 with Dave, so he won't be able to do anything with it until he gets back to London. 
if that's the case then we would all jus buy ti 4200's and nothing higher. we should all buy civics instead of accords or chevy nova's instead of camaro's or go to state colleges and ignore the ivy leagues or .... you get my drift im sure.
Mem
Lifer
- Apr 23, 2000
- 21,476
- 13
- 81
I don`t know about you but I don`t buy a video card based on 3Dmark score,besides we don`t know if Nvidia did actually cheat and according to Extremetech it has to do with FX cards only with 3Dmark.Remember this is new hardware by Nvidia we are talking about, not a very old TNT or Geforce card,I`m still not 100% certain eitherway.
Bottom line is this,as long as my games run fine I`m one happy gamer and I still want more conculsive facts before I make a final judgement,we just don`t know if it`s a driver bug or cheat by Nvidia at this time,remember it`s very easy to point the finger eitherway.
Will it stop me buying a FX or Nvidia card in the future? Nope,as I said 3dmark means Jackshxx to me,in the end playing games that run fine is all I care about not how fast my 3dMark score is.
Mem, even though your points have been refuted before in this thread, I'll repeat myself (and others) one more time, just so casual gamers like yourself will understand how serious this could be. You will care when nV uses this same technique for a real game that's benchmarked in the same way 3DM03 is ("on rails," like in a flyby), and you buy the card thinking it'll get 100fps at a certain setting, but find it gets only 80fps. Some people and, more importantly, companies (like Dell and other PC makers) choose cards due to industry-standard benchmarks (which 3DM clearly is). Finally, the FX is no longer new hardware--the FX 5800 was demoed last November, and we're already on the second generation of that technology. (New hardware is not an excuse for cheating, BTW, so I'm not sure why you even brought that up.)
I, too, await more facts. I think we have enough to be conclusive, but I'd like to see more corroborating evidence and hear nVidia's reply before I pass final judgement on this as yet another nVidia lie/cheat (don't forget either xbit labs or digit life exposed their pseudo-4x FSAA shortcut on the GF4MX a few months ago, not to mention the previous IQ corner-cutting with most FX drivers).
It's all the more baffling since they claimed to think so little of 3DM03--they could've just told everyone that they're simply not going to optimize for it, and that's why their cards are scoring so poorly. But this is the second time they've sacrificed IQ for speed in this benchmark they seem to care so little for.
I, too, await more facts. I think we have enough to be conclusive, but I'd like to see more corroborating evidence and hear nVidia's reply before I pass final judgement on this as yet another nVidia lie/cheat (don't forget either xbit labs or digit life exposed their pseudo-4x FSAA shortcut on the GF4MX a few months ago, not to mention the previous IQ corner-cutting with most FX drivers).
It's all the more baffling since they claimed to think so little of 3DM03--they could've just told everyone that they're simply not going to optimize for it, and that's why their cards are scoring so poorly. But this is the second time they've sacrificed IQ for speed in this benchmark they seem to care so little for.
You can take the diapers off me now whiz kid. I'm aware of the ramifications of this cheat, and I still don't give a rat's @ss. I hope ATI, Matrox, Trident, SiS, and PowerVR all do it too.
My point is, if these guys have a synthetic benchmark shoved in their face which can make or break their OEM ODM sales, they are gonna cheat. All of them. Regardless of which one you like. If they haven't cheated that just means that either they haven't been caught, or they don't have the resources to cheat because there are too many other bugs to fix.
WRT this affecting all benchmarks, I don't think so. And that's the beauty of it!!! Anand can record a new UT or Quake or DOom demo at any time, so why would nVidia try to cheat in this manner on those?!? They would be easily caught and reviewers would just switch to a new demo, making their work pointless.
But what about the Futuremark? You can't just switch the angle of the demo version in a patch, it would render your database invalid. So, either Futuremark's database get's rendered invalid or nVidia continues to cheat... neither of which concern me in the slightest.
Pete:
Of course, at that point you could just return the card. You say "when" nVidia uses this technique in real games. I've seen no mention of them doing so, and I have to think reviewers and others are looking hard at their drivers for this. (due to the "fame" of breaking the "nVidia cheats" story)
As far as Kyle goes, he's been pretty pro ATI of late, I thought his 5800 review was totally skewed in ATIs favor.
I think I'll:
A. Wait to see nVidia's response to this
B. Not care about 3dmark still
C. Buy a 5900 Ultra and see if it kicks a$$
Of course, at that point you could just return the card. You say "when" nVidia uses this technique in real games. I've seen no mention of them doing so, and I have to think reviewers and others are looking hard at their drivers for this. (due to the "fame" of breaking the "nVidia cheats" story)
As far as Kyle goes, he's been pretty pro ATI of late, I thought his 5800 review was totally skewed in ATIs favor.
I think I'll:
A. Wait to see nVidia's response to this
B. Not care about 3dmark still
C. Buy a 5900 Ultra and see if it kicks a$$
Regs
Lifer
- Aug 9, 2002
- 16,666
- 21
- 81
Benchmarking programs do have there strong points but also have their downs. Take a game like Doom 3, you are not going to need 50+ fps walking down corridors or playing cut scene movies. You are not going to need 50+ fps controlling armies in Wc3, civilization, ext..
Wow ATI and Nvidia "cheated" by adding maybe a few extra 100 points in a score that won't matter dick playing a game. BOooooOO hooo, learn some logic and stop crying. If you do not know the logistics of marketing: you should turn off the tv, unplug your computer and phone, and lock yourself in your room before you end up investing money into condo's that happen to be built inside a USA air force bomb testing zone.
Wow ATI and Nvidia "cheated" by adding maybe a few extra 100 points in a score that won't matter dick playing a game. BOooooOO hooo, learn some logic and stop crying. If you do not know the logistics of marketing: you should turn off the tv, unplug your computer and phone, and lock yourself in your room before you end up investing money into condo's that happen to be built inside a USA air force bomb testing zone.
Nah....PC makers care about compatibility and drivers before performance. I doubt you hear "Dude my Dell is only getting 80 FPS instead of 100 FPS" before you hear "Dude my Dell won't play any games". I also doubt that computer companies look at 3DM03 as the final say on what card goes into a their Dell or Compaq machine. But everyone here....well....that's a different story..(ad nauseum)...
Self
I'll throw in an additional $0.02 here.
1) Benchmarking is a tool used to determine which IHV's product performs better in various application. It's done by websites and reviewers who GET PAID because they make observations and conlusions the average consumer doesn't have the time, ability, or resources to make.
2) There are benchmarks which take a snapshot of an actual application (such as a Quake timedemo), and there are synthetic benchmarks which are written to make the process of #1 above easier and more fair across the board, such as 3DMARK03.
3) If the benchmarks in #2 above can be easily thwarted, such that the results don't reflect typical "real world" performance, then they are no longer good tools for the journalists in #1 above to use in their publications.
4) You must assume that IHV's will do anything in their power to "optimize" for a benchmark, as it will drive the journalists to demonstrate their products as superior.
5) If the journalists rely on flawed tools to provide data to consumers, then so be it. Their credibility as good journalists will soon decay and the more intelligent consumers will end up reading reviews on sites like B3D where they have the capability to sort between the "wheat and the shaft" for lack of a better term.
My conclusion is that regardless of whether nVidia did this on purpose of not (my assumption is that they did this on purpose), this serves as a nice wake-up call to not only consumers but to the journalists who's opinions we pay (by our clicks or by our magazine subscriptions, whatever) for that they need to be able to be more intelligent about how they come to their conclusions.
If nVidia is spoiling the Futuremark milk, so be it. I've never placed much faith in a program which spits out a "3dMark" number that people put in their signatures on BBS's.
The end result is a catch-22. If journalists really want to serve their customers when they review this type of hardware, they will have to learn to use tools which do the following
1) represent actual application performance
2) can be easily customized to different levels of IQ
3) allow custom timedemo's to be recorded, such that a sequence cannot be "pre-rendered" by the driver
4) be willing to accept that there is no single benchmark which will summarize a products performance
5) be willing to eliminate the database approach of benchmarking, where a single test be used over and over for convience sake.
I hate to say it, but I almost agree with Kyle's opinion on this. If you stick a synthetic benchmark into the face of a IHV, they are going to do everything they can to end up with the highest result. They are engineers afterall, and that's what engineers do.
If you send the message to the IHV's that you can and will "randomize" the benchmarking, then they will be forced to do the right thing, which is to optimize their products for the most popular game titles. This is what consumers want. This is what IHV's want. So what's the problem? It's the journalists, they want a standarized tool to make their lives easier. Hence the catch-22.
1) Benchmarking is a tool used to determine which IHV's product performs better in various application. It's done by websites and reviewers who GET PAID because they make observations and conlusions the average consumer doesn't have the time, ability, or resources to make.
2) There are benchmarks which take a snapshot of an actual application (such as a Quake timedemo), and there are synthetic benchmarks which are written to make the process of #1 above easier and more fair across the board, such as 3DMARK03.
3) If the benchmarks in #2 above can be easily thwarted, such that the results don't reflect typical "real world" performance, then they are no longer good tools for the journalists in #1 above to use in their publications.
4) You must assume that IHV's will do anything in their power to "optimize" for a benchmark, as it will drive the journalists to demonstrate their products as superior.
5) If the journalists rely on flawed tools to provide data to consumers, then so be it. Their credibility as good journalists will soon decay and the more intelligent consumers will end up reading reviews on sites like B3D where they have the capability to sort between the "wheat and the shaft" for lack of a better term.
My conclusion is that regardless of whether nVidia did this on purpose of not (my assumption is that they did this on purpose), this serves as a nice wake-up call to not only consumers but to the journalists who's opinions we pay (by our clicks or by our magazine subscriptions, whatever) for that they need to be able to be more intelligent about how they come to their conclusions.
If nVidia is spoiling the Futuremark milk, so be it. I've never placed much faith in a program which spits out a "3dMark" number that people put in their signatures on BBS's.
The end result is a catch-22. If journalists really want to serve their customers when they review this type of hardware, they will have to learn to use tools which do the following
1) represent actual application performance
2) can be easily customized to different levels of IQ
3) allow custom timedemo's to be recorded, such that a sequence cannot be "pre-rendered" by the driver
4) be willing to accept that there is no single benchmark which will summarize a products performance
5) be willing to eliminate the database approach of benchmarking, where a single test be used over and over for convience sake.
I hate to say it, but I almost agree with Kyle's opinion on this. If you stick a synthetic benchmark into the face of a IHV, they are going to do everything they can to end up with the highest result. They are engineers afterall, and that's what engineers do.
If you send the message to the IHV's that you can and will "randomize" the benchmarking, then they will be forced to do the right thing, which is to optimize their products for the most popular game titles. This is what consumers want. This is what IHV's want. So what's the problem? It's the journalists, they want a standarized tool to make their lives easier. Hence the catch-22.
what is the problem with recording a new Q3 demo for example for a card review? it only takes a quick jump to a server, doesn't it?
then you can just ask forum peeps to send THEIR demos and use all of these.
then you can just ask forum peeps to send THEIR demos and use all of these.
Jeff7181
Lifer
- Aug 21, 2002
- 18,368
- 11
- 81
That's why we have places like AnandTech... to show us what the card can do in real games. I'd actually like to see a comparison done with games that aren't normally used... play some Ghost Recon... record the game... then play it back using different video cards and see what the frame rates are.
"As far as Kyle goes, he's been pretty pro ATI of late"
Of late. Otherwise I've found him overly-pro-whoever's number one ATM, but maybe I'm just misreading his natural enthusiasm. His latest editorial seems to confirm my more cynical suspicions, though.
"Nah....PC makers care about compatibility and drivers before performance. I doubt you hear "Dude my Dell is only getting 80 FPS instead of 100 FPS" before you hear "Dude my Dell won't play any games"."
I suppose that's why Dell now offers the Radeon 9700 & 9800--not performance, but driver quality. :\ Also, I didn't say 3DM was the only or final say in terms of which cards they pick, but the fact that Dell is a FutureMark Beta member indicates they take 3DM results seriously.
"I'm aware of the ramifications of this cheat, and I still don't give a rat's @ss. I hope ATI, Matrox, Trident, SiS, and PowerVR all do it too."
I can't argue against that. You want to be deceived, then enjoy it. I prefer not to be.
"Real" engineers do not cheat. Sadly, a fair share of engineers are not "real" engineers, as in all of life.
Most journalists would not have been able to catch this cheat, no matter how diligent they were, since only a handful of review sites are part of the 3DM Beta program (B3D, CNet, and ET, IIRC). Note that most web journalists probably aren't paid as well as Anand & co., either.
I also think you are oversimplifying, or, rather, misinterpreting, Kyle's opinion. From what I gathered in reading the now-locked thread at the [ H ] forums, he thinks this is all OK because he doesn't believe in 3DM's validity as a benchmark and it doesn't impact IQ. I think he's wrong on both counts. Firstly, standardized tools like 3DM can serve to keep vendors honest in terms of features they advertise--potentially very important to companies like Dell whose customers tend to keep their PC config stable longer than most of us enthusiasts, and who'll waste Dell's time complaining if their 3D card doesn't perform as expected. Secondly, rendering without impacting IQ is not 3DM's only requirement. 3DM is meant to be a test of how quickly a 3D card accelerates a standard set of data with reference IQ, and nVidia is invalidating their results by not rendering all that data. If they didn't want cards to render everything, they would've implemented those clip planes themsevles. Kyle's especially wrong to use a tactic he himself decries ("concluding motive") to smear ET's motives in posting their article (which was at least a week in planning and discussed with Futuremark, according to a few sources).
The wiz kid in me is prompted to correct you: "separate the wheat from the chaff."
Alright, I'll back off for now. I realize I'm being a bit too aggressive with my theorizing and extrapolating. But obviously the cynic in me has taken hold WRT nVidia, given their past "ethical low-road" actions.
Of late. Otherwise I've found him overly-pro-whoever's number one ATM, but maybe I'm just misreading his natural enthusiasm. His latest editorial seems to confirm my more cynical suspicions, though.
"Nah....PC makers care about compatibility and drivers before performance. I doubt you hear "Dude my Dell is only getting 80 FPS instead of 100 FPS" before you hear "Dude my Dell won't play any games"."
I suppose that's why Dell now offers the Radeon 9700 & 9800--not performance, but driver quality. :\ Also, I didn't say 3DM was the only or final say in terms of which cards they pick, but the fact that Dell is a FutureMark Beta member indicates they take 3DM results seriously.
"I'm aware of the ramifications of this cheat, and I still don't give a rat's @ss. I hope ATI, Matrox, Trident, SiS, and PowerVR all do it too."
I can't argue against that. You want to be deceived, then enjoy it. I prefer not to be.
"Real" engineers do not cheat. Sadly, a fair share of engineers are not "real" engineers, as in all of life.
Most journalists would not have been able to catch this cheat, no matter how diligent they were, since only a handful of review sites are part of the 3DM Beta program (B3D, CNet, and ET, IIRC). Note that most web journalists probably aren't paid as well as Anand & co., either.
I also think you are oversimplifying, or, rather, misinterpreting, Kyle's opinion. From what I gathered in reading the now-locked thread at the [ H ] forums, he thinks this is all OK because he doesn't believe in 3DM's validity as a benchmark and it doesn't impact IQ. I think he's wrong on both counts. Firstly, standardized tools like 3DM can serve to keep vendors honest in terms of features they advertise--potentially very important to companies like Dell whose customers tend to keep their PC config stable longer than most of us enthusiasts, and who'll waste Dell's time complaining if their 3D card doesn't perform as expected. Secondly, rendering without impacting IQ is not 3DM's only requirement. 3DM is meant to be a test of how quickly a 3D card accelerates a standard set of data with reference IQ, and nVidia is invalidating their results by not rendering all that data. If they didn't want cards to render everything, they would've implemented those clip planes themsevles. Kyle's especially wrong to use a tactic he himself decries ("concluding motive") to smear ET's motives in posting their article (which was at least a week in planning and discussed with Futuremark, according to a few sources).
The wiz kid in me is prompted to correct you: "separate the wheat from the chaff."
Alright, I'll back off for now. I realize I'm being a bit too aggressive with my theorizing and extrapolating. But obviously the cynic in me has taken hold WRT nVidia, given their past "ethical low-road" actions.
I've not been deceived, because I've never placed any faith in the results generated by 3Dmark. Have you?
Engineers are presented by a problem (getting the highest score in a synthetic benchmark), bounded by a set of rules. In this case, they pushed the limits of the rules. That's what "real" engineers do well.
Marketing is there to promote it, and apologize if caught. It's always easier to ask for forgiveness than to ask for permission.
I knew it didn't sound right!
I can't speak for Kyle's motive in his opinion posted on Hardocp, but I've also read all 10 pages of the now locked thread and I think I understand Kyle's opinion just fine.
Standardized tools are a great concept, but they must be tempered with some intelligence. You cannot expect a company who's revenue will be impacted by their performance in a standardized SYNTHETIC benchmark to NOT bend over backwards to tweak for that benchmark. You can only hope that the tweaking they do will somehow benifit the end user.
In this case, we clearly see that this is not the case. The tweaking (cheating/bug/optimization, whatever) serves no one. You can take this as meaning 2 things...
1) nVidia is a bunch of jerks, or
2) 3Dmark is an invalid benchmark.
I simply choose #2
I consider myself cynical as well, that's why this crap doesn't suprise me.
nobody ever said you people will stop buying nvidia products due to this news. But if debating over a comparable ATI product for example, you can normally use benchmarks to finalize you decision for example. It has its uses. so shut up with the "i dont give a rat's ass about 3dmark" b$. You dont belong in this thread if you are a casual user who buy from Dell.
- Aug 22, 2001
- 32,434
- 33,439
- 146
<Cliffnotes version>The Cliffnotes version of this thread for those who don't speed read or are just to lazy to read through it
fanATIcs-->
Nvidia cheated! Damn them! damn them to hell!
Ha! A big percenatge of the overall performance gap between the FX and the 9800pro in reviews comes from the lead in 3Dmark and just proves it isn't going to lay a serious performance smackdown on my beloved 9800pro afterall Thanx be!
They cheated here so that means they might cheat on all similar "on a rail" benchies eventually those despicable bastages!
Nvidia is lying and cheating to steal the consumers money by deceiving them into believing they are getting a better price/performance ratio than they really are!
Ha! Now stop rubbing my nose in the Quack fiasco!
Another reason for me to continue buying my beloved ATI products.
This explains why Nvidia dropped out of the beta program.....because they knew that 3Dmark would not show their cards in a favorable light!
NvIDIOTs-->
I don't care!
Just proves that Nvidia was right and 3DMark does suck and no one should use it.
It's just a driver bug.... big deal!Stop trying to turn it into Watergate you conspiracy theorist!
Cheating at a stupid and useless synthetic benchie isn't nearly as bad as the Quack fiasco!
Our drivers are still better than ATI's even if they have cheats written in for certain stupid and useless benchies! na na na na nana!
If this turns out to be cheating afterall I'll still buy my beloved Nvidia products and justify and rationalize anything they do as savy business practice'
Voices of reason slipping in occassionally-->
Let's reserve judgement till all the facts are known.
This is a good time to re-evaluate the way benchmarks are performed and attempt to devise a better method of doing them in order to help make the results better reflect real world performance while being more difficult to tamper with.
All corporations engage in what many would consider to be unsavory and unethical practices in order to further their own goals so this should not come as a surprise to anyone so stop discriminating against one corp in particular or be prepared to stop purchasing almost everything you need/want because otherwise it's hypocracy.
</Cliffnotes version>
fanATIcs-->
Nvidia cheated! Damn them! damn them to hell!
Ha! A big percenatge of the overall performance gap between the FX and the 9800pro in reviews comes from the lead in 3Dmark and just proves it isn't going to lay a serious performance smackdown on my beloved 9800pro afterall Thanx be!
They cheated here so that means they might cheat on all similar "on a rail" benchies eventually those despicable bastages!
Nvidia is lying and cheating to steal the consumers money by deceiving them into believing they are getting a better price/performance ratio than they really are!
Ha! Now stop rubbing my nose in the Quack fiasco!
Another reason for me to continue buying my beloved ATI products.
This explains why Nvidia dropped out of the beta program.....because they knew that 3Dmark would not show their cards in a favorable light!
NvIDIOTs-->
I don't care!
Just proves that Nvidia was right and 3DMark does suck and no one should use it.
It's just a driver bug.... big deal!Stop trying to turn it into Watergate you conspiracy theorist!
Cheating at a stupid and useless synthetic benchie isn't nearly as bad as the Quack fiasco!
Our drivers are still better than ATI's even if they have cheats written in for certain stupid and useless benchies! na na na na nana!
If this turns out to be cheating afterall I'll still buy my beloved Nvidia products and justify and rationalize anything they do as savy business practice'
Voices of reason slipping in occassionally-->
Let's reserve judgement till all the facts are known.
This is a good time to re-evaluate the way benchmarks are performed and attempt to devise a better method of doing them in order to help make the results better reflect real world performance while being more difficult to tamper with.
All corporations engage in what many would consider to be unsavory and unethical practices in order to further their own goals so this should not come as a surprise to anyone so stop discriminating against one corp in particular or be prepared to stop purchasing almost everything you need/want because otherwise it's hypocracy.
</Cliffnotes version>
Dapunisher, it's too bad you didn't post that earlier, would've saved me a half hour. I want to introduce the third crowd: The I Don't Care people.
IMO, we haven't got a sure fire way of knowing if nVidia did do up their drivers to intentionally skew 3DMark03, but I wouldn't put it past them. I wouldn't put it past any company for that matter. The fact of the matter is it happened, whether intentionally or unintentionally. Arguing about why it happened really won't accomplish anything. If nVidia comes out and says publicly "It was a bug, sorry", then of course you're going to see the nVidia crowd yelling that in the streets, while the ATI crowd will be saying "Yeah right!". The story won't change much if nVidia said openly "yeah, we did it to pad the score, and the actual performance is lower". nVidia supporters will be out in force claiming that it was there to show everyone that Synthetic benchmarks don't count, and the ATI supporters will be there to chant the opposite.
The same would hold true if the situation were reversed. You know what that makes them all? People who can't see that most of the time, there is a different tool for each job.
This is not going to affect anyone but the uninformed in any great fashion, IMO.
I've never been a fan of brand loyalty, and as I see it, little side stories like this are really just irrelevant to the grand scheme of things. I highly doubt someone's decision on whether to buy an FX5900Ultra will be swayed either way by 'news'. If something like that does happen, the person in question is a fool for changing their decision based on something that is basically irrelevant to how the card's real world performance is going to be, IMO.
IMO, we haven't got a sure fire way of knowing if nVidia did do up their drivers to intentionally skew 3DMark03, but I wouldn't put it past them. I wouldn't put it past any company for that matter. The fact of the matter is it happened, whether intentionally or unintentionally. Arguing about why it happened really won't accomplish anything. If nVidia comes out and says publicly "It was a bug, sorry", then of course you're going to see the nVidia crowd yelling that in the streets, while the ATI crowd will be saying "Yeah right!". The story won't change much if nVidia said openly "yeah, we did it to pad the score, and the actual performance is lower". nVidia supporters will be out in force claiming that it was there to show everyone that Synthetic benchmarks don't count, and the ATI supporters will be there to chant the opposite.
The same would hold true if the situation were reversed. You know what that makes them all? People who can't see that most of the time, there is a different tool for each job.
This is not going to affect anyone but the uninformed in any great fashion, IMO.
I've never been a fan of brand loyalty, and as I see it, little side stories like this are really just irrelevant to the grand scheme of things. I highly doubt someone's decision on whether to buy an FX5900Ultra will be swayed either way by 'news'. If something like that does happen, the person in question is a fool for changing their decision based on something that is basically irrelevant to how the card's real world performance is going to be, IMO.
- Aug 22, 2001
- 32,434
- 33,439
- 146
I covered the "I don't care" response chsh1ca You actually fall into the"voice of reason" catagory :beer: Look it's part of the human dynamic, IIRC it was Neitzsche who called it the heard mentality. Most everyone wants to belong to a particular heard I suppose, it's just that some of us are rogues who eschew the heards in favor of our own egocentric impulses What does this have to do with anything....not a damned thing!
You actually fall into the"voice of reason" catagory :beer: Look it's part of the human dynamic, IIRC it was Neitzsche who called it the heard mentality. Most everyone wants to belong to a particular heard I suppose, it's just that some of us are rogues who eschew the heards in favor of our own egocentric impulses What does this have to do with anything....not a damned thing! I guess if nVidia is truely cheating, fudging, or just plain dissin 3DMark/Futuremark because they dislike it, the real test would be to take an actuall game which uses industry spec. rendering, and run it thru extensive rendering tests and look for missing/poorly rendered areas.
If it can't pass industry standards actually being used in game rendering, then shame on them. They are actually screwing us all and deserve being branded cheaters.
If it can't pass industry standards actually being used in game rendering, then shame on them. They are actually screwing us all and deserve being branded cheaters.
Here is a link showing the latest 44.03 FX detonators no clip in SeriousSam:
http://homepage.ntlworld.com/paul.toler/ssclip.jpg
So the Cheat/bug is also in games. So all of you saying you didnt care beacuse it was only in benchmarks...
http://homepage.ntlworld.com/paul.toler/ssclip.jpg
So the Cheat/bug is also in games. So all of you saying you didnt care beacuse it was only in benchmarks...
Can you post the source of this?
This is fairly important, because if this is on a "FX" card with 44.03 drivers then the theory of a "camera rail" cheat of 3Dmark might not be applicable anymore.
Originally posted by: DAPUNISHER
I covered the "I don't care" response chsh1ca
Dang, I think I have elements of all three of those in my arguments... no wonder I'm such a confused individual
I hold faith in its ability to be a reasonable benchmark of comparable performance when tested by knowledgable reviewers.
If by "push the limits" you mean "exceed acceptable bounds," then I agree. I suppose real engineers "push the limits" when they neglect to use rebar in concrete foundations. They optimized for speed! ;P
I did, too, and some of his replies are fairly reasonable. A comment of his on the last page, however, seems to prove he's also missing the point: he says it's OK as long as they don't compromise IQ. He misses the point that by changing the rules of the test, nVidia compromises (read: negates) the validity of their benchmark results. This does everyone (consumers, reviewers, benchmark makers) but the maker of underperforming hardware a disservice.
#1 is premature ATM, but so far all signs point to this being a deliberate act. We'll have to wait on a definitive article comparing all three unique FX driver sets (43.00, 43.45, 44.03) on NV35 and NV30, as well as on the 9800, to draw a definitive conclusion. #2 is the easy and ultimately fruitless way out, as if 3DM03 is invalid as a benchmark, then so is every other "benchmark" most sites use (NVNews is the sole exception, AFAIK, and deserve credit for their efforts), as they're all performed in the same manner.
Ben Skywalker took the shot, someone (MuFu?) hosted it for him.
I hold faith in its ability to be a reasonable benchmark of comparable performance when tested by knowledgable reviewers.
I hold faith in it's ability to be a relative benchmark, but I believe it's absolute results to be too easily influenced by driver tweaks. Unfortunatly, the ROW uses it as an absolute performance reference.
Although I do proffer that 3DMark2001 did a fair job in "ranking" cards relative to how well they would perform in actual applications. I do not believe 3DMark03 is valid in this manner.
If by "push the limits" you mean "exceed acceptable bounds," then I agree.
Well, we differ on our opinion of what's acceptable. If this "cheat" only affects 3Dmark, without changing the image quality of the default benchmark, and doesn't affect any other games (i.e. all it does is inflate the 3dmark score, no other negative impact on actual consumers) then I find it to be quite clever, and I applaud nVidia for ALMOST getting away with a fast one.
If indeed the "cheat" affects other applications, then the will be forced to fix it. That I don't argue with one bit. Do you know if Ben ran that on a FX card? I though he only had a GF4.
He misses the point that by changing the rules of the test, nVidia compromises (read: negates) the validity of their benchmark results. This does everyone (consumers, reviewers, benchmark makers) but the maker of underperforming hardware a disservice.
I think that is his point... and my point to some extent. For sake of argument, let's assume that nVidia released a driver which affects nothing else but inflates the 3dMark03 score by 50%. I applaud that, because it will FORCE the community to be more intelligent about evaluating and benchmarking this hardware. I applaud it in the same way that I applaud a hacker who figures a way to comprimise security of a network. It's a bit scary when it happens, but the end result is a better network.
#1 is premature ATM, but so far all signs point to this being a deliberate act.
Well, it it's isolated to 3Dmark I'm 100% in agreement that it's deliberate.
If it affects gameplay in other titles, I wouldn't be so sure.
Wow. This is almost becoming a civilized debate.
I hold faith in it's ability to be a relative benchmark, but I believe it's absolute results to be too easily influenced by driver tweaks. Unfortunatly, the ROW uses it as an absolute performance reference.
Although I do proffer that 3DMark2001 did a fair job in "ranking" cards relative to how well they would perform in actual applications. I do not believe 3DMark03 is valid in this manner.
If by "push the limits" you mean "exceed acceptable bounds," then I agree.
Well, we differ on our opinion of what's acceptable. If this "cheat" only affects 3Dmark, without changing the image quality of the default benchmark, and doesn't affect any other games (i.e. all it does is inflate the 3dmark score, no other negative impact on actual consumers) then I find it to be quite clever, and I applaud nVidia for ALMOST getting away with a fast one.
If indeed the "cheat" affects other applications, then the will be forced to fix it. That I don't argue with one bit. Do you know if Ben ran that on a FX card? I though he only had a GF4.
He misses the point that by changing the rules of the test, nVidia compromises (read: negates) the validity of their benchmark results. This does everyone (consumers, reviewers, benchmark makers) but the maker of underperforming hardware a disservice.
I think that is his point... and my point to some extent. For sake of argument, let's assume that nVidia released a driver which affects nothing else but inflates the 3dMark03 score by 50%. I applaud that, because it will FORCE the community to be more intelligent about evaluating and benchmarking this hardware. I applaud it in the same way that I applaud a hacker who figures a way to comprimise security of a network. It's a bit scary when it happens, but the end result is a better network.
#1 is premature ATM, but so far all signs point to this being a deliberate act.
Well, it it's isolated to 3Dmark I'm 100% in agreement that it's deliberate.
If it affects gameplay in other titles, I wouldn't be so sure.
Wow. This is almost becoming a civilized debate.
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: 24K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter