nVidia vs Futuremark Continued - Guess what nVidia's doing!
- Thread starter Arun
- Start date
I'm also having trouble divining ED's problem, but I think it may be a product of misunderstanding due to imperfect translation. Perhaps ED (mis)took your initial comment as haughtier than intended?DaveBaumann said:Evildeus said:So are you giving up intigrity to Nv?
I fail to understand what you're trying to say.
What is the truth Reverend? That is definately my perception, what actually happened?Reverend said:
The "truth" is that FM is made up of more than one individual -- the official public statements were made (hence the understandable public perceptions) but not necessarily agreed upon by all. After some mulling over, I see no harm in posting the below exchange :
I think I said in some thread that I wished FM had implemented what I suggested to them when I read their audit report draft, which was to not use the word "cheat", which should be left to the discretion and decisions of the public (and websites), because I recognized the potential legal danger... but the audit report was scheduled to be public within the next day after I read the draft, so things were already in motion before what I had to say had any, er, importance. I even gave FM my edited draft, substituting all instances of the word "cheat" with what was actually going on. Although I respected the fact that this was entirely in FM's hands in what they choose to call it officially ("cheat" or statements of facts about what's actually happening).
The result is not one which I wished for, for all parties concerned.
Rev to FM personnels after knowing about the joint FM/NV press release said:
A FM personnel in response to Rev said:
I think I said in some thread that I wished FM had implemented what I suggested to them when I read their audit report draft, which was to not use the word "cheat", which should be left to the discretion and decisions of the public (and websites), because I recognized the potential legal danger... but the audit report was scheduled to be public within the next day after I read the draft, so things were already in motion before what I had to say had any, er, importance. I even gave FM my edited draft, substituting all instances of the word "cheat" with what was actually going on. Although I respected the fact that this was entirely in FM's hands in what they choose to call it officially ("cheat" or statements of facts about what's actually happening).
The result is not one which I wished for, for all parties concerned.
Thanks Reverand, I appreciate you sharing that.
I always sort of felt that the controversial joint press release was the result of an individual or a couple of individuals panicking, but once it was done it was too late to change.
I agree that FM shouldn't have used the word "cheat" in the first document. They could have avoided giving nVidia any ammunition to use against them in the first place and it wouldn't have been hard to get the gist across without using that particular word with all it's various legal implications.
I always sort of felt that the controversial joint press release was the result of an individual or a couple of individuals panicking, but once it was done it was too late to change.
I agree that FM shouldn't have used the word "cheat" in the first document. They could have avoided giving nVidia any ammunition to use against them in the first place and it wouldn't have been hard to get the gist across without using that particular word with all it's various legal implications.
Nv is cheating, everyone knows it but none says it anymore, officially at least. Nv could reenter FM beta program, but there is some issues that are aknolesge but not resolve if i understand correctly.DaveBaumann said:Evildeus said:So are you giving up intigrity to Nv?
I fail to understand what you're trying to say.
You have always said that if you were not happy with the policy of FM, you would go out. I'm asking you if you (aka B3D) will stay at FM, if those issues are not resolved, or you would prefer pretend that they are, and get rid of your integrety (as FM as already done) to the benefit of being a FM beta member?
Evildeus said:
Just a personal opinion, but I don't think the language used to describe nVidia's actions matters any more - Everyone knows what they're doing, so it doesn't matter if you call them 'cheats', 'optimisations' or 'teapots', the story is still the same.
I'd also say that in my eyes, regardless of how Beyond3D feel about FutureMark's actions, I'd rather have them in the program than out of it. It wouldn't benefit the community one iota if Beyond3D and/or ExtremeTech dropped out, we'd just be left even more in the dark as to what's going on.
First:
Treating all instances of cheating as the "same" for the convenience of avoiding discussing the distinction between specific instances, and trying to propose that evaluation as a necessary given (when discussion can proceed independently of that evaluation), is, IMO, pure laziness, and an excuse for proposing moral apathy without having to actually defend such a stance under scrutiny. Plenty of commentary concerning my...hmm..."more than mild disagreement" with that particular sentiment are abundant for address elsewhere, if anyone who happens to be proposing such moral apathy (again) wishes to actually try and defend it as being more than a tool of personal convenience.
But, in any case, the matter of cheating is only one part of the issue, and discussion is more (IMO) a matter of the context of evaluating Futuermark's actions in regards of the unique things nVidia did in that regard (that treating ATI and nVidia as equivalent in their actions serves to neatly gloss over) in light of the specific action in cooperation with nVidia, on Futuermark's part, that is being discussed.
Doing so:
I've proposed that the software is not Futuremark's only product, and even that, IMO, providing information is their major "product" and that the software was in some regard just part of the toolset in delivering that. That, atleast, relates to my perspective of what they have done wrong (as far as I'm aware with the information available to me, which is part of the point), from the time of the joint FM/NV announcement up until this point.
Viewed in this way, I have a slightly different impression than that I think has been expressed: that Futuremark is actually obligated to do something along these lines if nVidia hardware has unique functionality in this regard (i.e., PPP, or atleast efficient PPP), and that not doing so, in such a case, would just hurt objectivity more in the long run. And further, that's not to say that there isn't a major issue with maintaining (non OEM) customer trust within the outlined situation, it is just that I think the focus of the problem is what they have not done, not what they are proposing to do...IOW, that what they are planning to do just highlights a deficiency in some of what has (recently) been done before.
Remember, not everyone reads these forums...I sometimes think people forget how important press releases and benchmark results are, and that there are people who don't visit such forums who can still form solid informed opinions if the information is actually made visible (outside of forums like this). From the perspective of providing information and considering both the proposed "demo" and the lack of a label like "cheating" for some of nVidia's actions, I think my reasons for the above viewpoint can be summarized in this way: You can misinform by selectively presenting and suppressing perfectly accurate factual information...but without just simply changing the type of misinformation, the focus for preventing misinformation should more likely be concerned with correcting what has been suppressed, not suppressing new opportunities for providing information.
Well, that is, in an ideal world...in the actual world, this demo discussion sounds like a pretty significant opportunity for misinformation, simply because there isn't any indication given that the prior information suppression will be corrected...
All this is, of course, is with the information commonly available at this time, which from my perspective is Futuermark's product (as far as their feature and API selection for evaluation is concerned)...from the perspective of information being Futuremark's product, it seems fairly evident that the "commonly available information" represents a specific deficiency on their part at the moment (See my prior discussion of OEM versus consumer information). What it comes down to is that I think this demo is only a problem if it occurs without addressing the currently exisiting issues with nVidia's "unique driver behaviors", and that the problem would be allowing the selective absence of information, not necessarily the new information provided (demo) itself.
...
This viewpoint tends to be precluded by both viewing nVidia's cheating as being part and parcel of Futuremark executing a demo to highlight nVidia's (presumably unique) hardware capabilities, and als precluded by viewing cheating as a universal either/or state of affairs without details warranting individual evaluation. It can be considered disagreement to both of these other viewpoints.
...
The question of whether such a demo is a good thing or a bad thing of itself would depend on exactly what it is (which we, or atleast I, don't know yet).
The question of whether the creation of such a demo serves providing accurate information in a broader sense of Futuremark's "product", depends on what other information is provided by Futuremark/nVidia's drivers at that time (which indeed might be, and hopefully will be, different than now). The "nVidia's drivers" comment is to make note of the possibility that a change in this aspect might be something Futuremark is working towards (and remind that there is a specific IHV deserving particular recognition for misinformation), though, again, without informing consumers of such, Futuremark are still deficient in delivering their "product", IMO.
Treating all instances of cheating as the "same" for the convenience of avoiding discussing the distinction between specific instances, and trying to propose that evaluation as a necessary given (when discussion can proceed independently of that evaluation), is, IMO, pure laziness, and an excuse for proposing moral apathy without having to actually defend such a stance under scrutiny. Plenty of commentary concerning my...hmm..."more than mild disagreement" with that particular sentiment are abundant for address elsewhere, if anyone who happens to be proposing such moral apathy (again) wishes to actually try and defend it as being more than a tool of personal convenience.
But, in any case, the matter of cheating is only one part of the issue, and discussion is more (IMO) a matter of the context of evaluating Futuermark's actions in regards of the unique things nVidia did in that regard (that treating ATI and nVidia as equivalent in their actions serves to neatly gloss over) in light of the specific action in cooperation with nVidia, on Futuermark's part, that is being discussed.
Doing so:
I've proposed that the software is not Futuremark's only product, and even that, IMO, providing information is their major "product" and that the software was in some regard just part of the toolset in delivering that. That, atleast, relates to my perspective of what they have done wrong (as far as I'm aware with the information available to me, which is part of the point), from the time of the joint FM/NV announcement up until this point.
Viewed in this way, I have a slightly different impression than that I think has been expressed: that Futuremark is actually obligated to do something along these lines if nVidia hardware has unique functionality in this regard (i.e., PPP, or atleast efficient PPP), and that not doing so, in such a case, would just hurt objectivity more in the long run. And further, that's not to say that there isn't a major issue with maintaining (non OEM) customer trust within the outlined situation, it is just that I think the focus of the problem is what they have not done, not what they are proposing to do...IOW, that what they are planning to do just highlights a deficiency in some of what has (recently) been done before.
Remember, not everyone reads these forums...I sometimes think people forget how important press releases and benchmark results are, and that there are people who don't visit such forums who can still form solid informed opinions if the information is actually made visible (outside of forums like this). From the perspective of providing information and considering both the proposed "demo" and the lack of a label like "cheating" for some of nVidia's actions, I think my reasons for the above viewpoint can be summarized in this way: You can misinform by selectively presenting and suppressing perfectly accurate factual information...but without just simply changing the type of misinformation, the focus for preventing misinformation should more likely be concerned with correcting what has been suppressed, not suppressing new opportunities for providing information.
Well, that is, in an ideal world...in the actual world, this demo discussion sounds like a pretty significant opportunity for misinformation, simply because there isn't any indication given that the prior information suppression will be corrected...
All this is, of course, is with the information commonly available at this time, which from my perspective is Futuermark's product (as far as their feature and API selection for evaluation is concerned)...from the perspective of information being Futuremark's product, it seems fairly evident that the "commonly available information" represents a specific deficiency on their part at the moment (See my prior discussion of OEM versus consumer information). What it comes down to is that I think this demo is only a problem if it occurs without addressing the currently exisiting issues with nVidia's "unique driver behaviors", and that the problem would be allowing the selective absence of information, not necessarily the new information provided (demo) itself.
...
This viewpoint tends to be precluded by both viewing nVidia's cheating as being part and parcel of Futuremark executing a demo to highlight nVidia's (presumably unique) hardware capabilities, and als precluded by viewing cheating as a universal either/or state of affairs without details warranting individual evaluation. It can be considered disagreement to both of these other viewpoints.
...
The question of whether such a demo is a good thing or a bad thing of itself would depend on exactly what it is (which we, or atleast I, don't know yet).
The question of whether the creation of such a demo serves providing accurate information in a broader sense of Futuremark's "product", depends on what other information is provided by Futuremark/nVidia's drivers at that time (which indeed might be, and hopefully will be, different than now). The "nVidia's drivers" comment is to make note of the possibility that a change in this aspect might be something Futuremark is working towards (and remind that there is a specific IHV deserving particular recognition for misinformation), though, again, without informing consumers of such, Futuremark are still deficient in delivering their "product", IMO.
nooneyouknow
Newcomer
I personally think ATI should leave the FM program permanently. The major reason being that FM backed down against legal pressure from Nvidia. No balls, no use. Down with 3D Mark forever I say!
Doomtrooper
Veteran
Interesting
Reverend: Since you shared that, I feel like I'm required to share something else, eh...
Before the initial 3DMark cheating debacle, I asked someone who I consider very reliable to say me if he knew anything about potential cheats in 3DMark 2003 by nVidia, since there were rumors of a ExtremeTech article.
He replied sometimes later, I think just after it all came up ( simply because he doesn't have much time to check ) - and he said he personally didn't know anything about it, but that he'd ask around.
And he did. The result? Nobody knew anything about it. And it's not just in a department of a certain company - he asked several people in different areas, including ones which are more likely to know such things ( read: driver people ) - nothing. All were taken by surprise, including some important peoples ( I assume a director of a segment related to this stuff ) which said he was still left in the dark: about one week after the original ET article came up!
Also, I've had the opportunity to see a few statements of things nVidia sends to their employees.
And well, they aren't only lying to their customers, don't worry, there's no discrimination going on - the employee memos are pretty much just as bad, with everything done in such a BS way.
Things like "We retrieved two optimizations considered as overly aggresive from ( driver release ) , in relation to 3DMark, but the majority of our optimizations have been left in because they are legal optimizations as described by John Carmack and Tim Sweeney" ( paraphrase )
The two optimizations in question are the back buffer clearing, and the sky clipping. The other optimizations are the shading program replacements. And the driver release is the 44.67 ( leaked today ).
And considering those shading optimizations as "legal as described by John Carmack and Tim Sweeney" ( paraphrase again ) is completely and utterly ridiculous, since it has been proved they decrease IQ.
Oh, sure, nVidia employees aren't idiots, I'm sure they can figure out what that all means. But people who don't have the time to figure it out for good or bad reasons obviously won't - just as in the case of the customer market, actually, with maybe an higher percentage of people figuring it out internally - but still...
So, if you want to excuse FutureMark - I fear you'll have to excuse nVidia too. Trust me, some people are NOT happy about this internally.
Obviously, other people accept it, otherwise it wouldn't happen, but it is also the case for FutureMark.
But no matter what, it *is* good to see there also are people in FutureMark who don't like what's happening right now - I supposed there was, but some evidence is always nice
Uttar
Before the initial 3DMark cheating debacle, I asked someone who I consider very reliable to say me if he knew anything about potential cheats in 3DMark 2003 by nVidia, since there were rumors of a ExtremeTech article.
He replied sometimes later, I think just after it all came up ( simply because he doesn't have much time to check ) - and he said he personally didn't know anything about it, but that he'd ask around.
And he did. The result? Nobody knew anything about it. And it's not just in a department of a certain company - he asked several people in different areas, including ones which are more likely to know such things ( read: driver people ) - nothing. All were taken by surprise, including some important peoples ( I assume a director of a segment related to this stuff ) which said he was still left in the dark: about one week after the original ET article came up!
Also, I've had the opportunity to see a few statements of things nVidia sends to their employees.
And well, they aren't only lying to their customers, don't worry, there's no discrimination going on - the employee memos are pretty much just as bad, with everything done in such a BS way.
Things like "We retrieved two optimizations considered as overly aggresive from ( driver release ) , in relation to 3DMark, but the majority of our optimizations have been left in because they are legal optimizations as described by John Carmack and Tim Sweeney" ( paraphrase )
The two optimizations in question are the back buffer clearing, and the sky clipping. The other optimizations are the shading program replacements. And the driver release is the 44.67 ( leaked today ).
And considering those shading optimizations as "legal as described by John Carmack and Tim Sweeney" ( paraphrase again ) is completely and utterly ridiculous, since it has been proved they decrease IQ.
Oh, sure, nVidia employees aren't idiots, I'm sure they can figure out what that all means. But people who don't have the time to figure it out for good or bad reasons obviously won't - just as in the case of the customer market, actually, with maybe an higher percentage of people figuring it out internally - but still...
So, if you want to excuse FutureMark - I fear you'll have to excuse nVidia too. Trust me, some people are NOT happy about this internally.
Obviously, other people accept it, otherwise it wouldn't happen, but it is also the case for FutureMark.
But no matter what, it *is* good to see there also are people in FutureMark who don't like what's happening right now - I supposed there was, but some evidence is always nice
Uttar
nVIDIA's 44.67 Detonators from eVGA say that they've corrected issues with 3DMark. Here's a quote from Neowin...
The drivers are available zipped up here http://envisautos.pwp.blueyonder.co.uk/4467_xp.zip
Evildeus said:
OK, this isn’t necessarily about their present actions or inactions on the current state of play with NVIDIA’s drivers. My concern stems from the fact that FM (not all, but ultimately enough to have the deciding decision) bowed under pressure (and knowing what I’ve heard its doesn’t surprise me at all) – the concern stems from what will occur from here. If NVIDIA don’t like what they see coming up in the future are we just going to see a further cycle of threats an capitulations? Or are Futuremark going to stand up to the internal pressure? I hope it will be the latter, and I think FM consider that to be the case at present, however inevitably things often turn out different in the heat of the moment.
The one thing that would trigger myself pushing the rest of B3D to leave the program is if the 3DMark03 were to be changed to use the precision hint. Why? Because if that were to occur that would be a clear inconsistency with their stance over why they didn’t change 2001SE to include PS1.4 as it the “scores in the database wouldn’t be comparableâ€. If 03 were changed then this is just a case of altering an already present benchmark due to pressure from one IHV. (Note: this doesn’t mean to say that I’m against the use of the precision hint at all. I’ve already suggested that it should be used in 3DMark04 (or whatever the next benchmark may happen to be
) because its just as much a valid part of DX as anything else and developers certainly are going to use it, as our recent developer interview shows).The other point I’m not keen on for various reasons (although I don’t yet know if I have quite as rigid a stance on this as the other point) is that I’m really not keen on vendor specific rendering paths for the benchmark as I feel this sort of defeats the entire point about a DirectX benchmark. This point is a little wider ranging and could be debated for a long time probably.