Why does nvidia cheat so much?

[Seero]

For me, had trouble with it because I am a firm believer in GPU Physics and desire to see more adoption and more dynamic gaming. This is a gamer stand-point, without the behind the scenes information, because I desire not only to see PhysX available to other vendors with a discrete nVidia PhysX card, have a desire to see PhysX be ported to OpenCL as well some day. This is my idealism. May be wrong or even ignorant but that's the way I feel.

I strongly believe that it won't be long until Nvidia once again re-enables the support as this action limits the growth of PhysX. If no game devs use PhysX, then the investment on acquiring PhysX will change from "worth something" to "worth nothing." Limiting users who actually have the hardware to use it isn't a good long term plan. IMO, PhysX itself can not be the dominate factor when it comes to purchase decision.

 

[Seero]

I also agree with the general sentiment regarding Nvidia locking out people from using an AMD/Nvidia mix for physx. Don't Q&A for it, fine. Don't provide technical support for it, fine. But don't lock it out - it's a crappy business move and prevents physx from getting higher adoption rates. I do not believe the argument that it would cut into sales of higher end Nvidia cards. I think, in fact, it would INCREASE sales of all Nvidia cards.

It will certainly increase sales, but not necessarily increase revenue. It was known that Cypress hits the market when Nvidia has nothing good to sale. It is not hard for me to believe that every employees were held at gun point searching for ways to stop potential customers from buying something else because they don't have anything to sale. If they did not do something, then potential customer will no longer be potential customer when they eventually have something to sale. The real economical problem is that they have already spend money on Fermi. If everyone buys 5xxx, then no one will buy 4xx and the entire production line must be put on hold as there will be not enough demand. You can see that as blood sucking vampire, or you can see it as the last resort before laying off people.

Look at the current bulldozer situation. It is the same problem, where competitor has something to sale, but not AMD. Other than leaking positive info, what can it be done? Bulldozer can be a good source of revenue ONLY if there is a market, and that market is shrinking as we speak.

If they have a solution that may delay the shrink, they will use it. This is business. Do they even care if it is a dirty trick? No, they should not, their employees have mouths to feed, and we, as consumers, are what they feast upon. A customer, we do not donate our money on garbages. We heard employees jumping off buildings, and we feel sad, but that does not change the will of getting an iphone. In fact, it helps the sale.

 

[amenx]

So, back to the original topic.

After reading through the Tech Report's article, I have to conclude that the author is pretty ignorant about information theory and computer graphics.

First of all, the quickest way to do fixed tessellation is to use the object's bounding volume's center or projected area-to-polygon-density ratio as the determining factor for when to tessellate. Hence you see the far side of those objects as densely tessellated as the front. When you have enough power to not care, you just don't care, which is probably the case here.

Secondly, does the author know how ANNOYING it is to write code to support punching holes in a water sim? There's a huge amount of border conditions and continuity problems if one assumes holes can occur anywhere. The simplest solution is to just always apply the tessellation to the whole water plane -- which, I'm sure, originally appeared under the city anyway.

And this leads to my third point: does the author even know how difficult the hidden object removal problem is in computer graphics? The usual "if I can't see it, then don't draw it" (which is what the author's statement boils down to) is probably as informative as the statement "physics can't be that hard if I'm living in it!"

In other words, if he can come up with an algorithm that does occlusion culling in O(logN) time, please go write a paper and publish it in SIGGRAPH -- I'd be VERY interested in reading it. And for those of you who think that this is already solved via BSP, KD-tree, BVH, etc, etc.... Well, please go read the literature again. Someone already calculated that these aren't O(logN), but more like O((logN)^6) or something -- but I couldn't find that paper again.

Moral of the story is: "Never attribute to malice that which is adequately explained by stupidity [or time constraints]."

Awaiting for those more technically knowledgeable to answer this. If factual, then we should see a lot of red faces from those who have argued that malice is the only explanation when something fishy involving Nvidia is concerned. Or is it going to be a silence of the lambs (and red faces anyway)?

 

[3DVagabond]

Is there not anything that isn't a conspiracy?
I must admit this is an excellent thread and I hope everyone has the stamina to stay and read it. I feel the tide of this forum is turning a bit. People are beginning to get tired of the conspiracy theories and are starting to peer through all that haze and see what is really going on here. There is a group, who would perpetuate the conspiracy mantra. Trying to keep it alive and well. Well, it's almost time to put it all on life support because I don't think people are going to take it anymore or be manipulated by the obvious BS.
The ability to be transparent is fading for them.
I'm not a psycho Nvidia fan. I love their products, and I can clearly see that company does more for gaming and it's customers and I appreciate that and will buy from them because of that. Which is exactly why Nvidia does it. To be far better than any competition and get more customers to buy their products. That is simple business, and when you do less for your customers, you lose them. That is what is happening en-masse now I think.

You claim conspiracy theory (buzz word of the week). To be honest you sound really naive with these kind of responses.

Define psycho? /joke

 

[Firestorm007]

You claim conspiracy theory (buzz word of the week). To be honest you sound really naive with these kind of responses.

Define psycho? /joke

If this is what Keys looks like in real life; I think that we may have a problem.

 

[mosox]

Awaiting for those more technically knowledgeable to answer this. If factual, then we should see a lot of red faces from those who have argued that malice is the only explanation when something fishy involving Nvidia is concerned. Or is it going to be a silence of the lambs (and red faces anyway)?

Aw, COME ON. There's lots of DX11 games but it seems only the recent TWIMTBP ones behave very differently than the rest. For everybody the 6970 is better than the 560Ti in DX11 but for them.



I don't have Crysis 2 but from the pics the famous concrete barriers are very similar to the ones in Mafia 2 that I have. Does the Nvidia Search and Destroy software team have a standard crippling package (because Mafia 2 is made by some other developer than Crytek)? Can't wait to see them barriers in BF3 AND of course, the 560Ti being AGAIN faster than the HD 6970.

 

[Keysplayr]

You claim conspiracy theory (buzz word of the week). To be honest you sound really naive with these kind of responses.

Define psycho? /joke

Ok?

 

[Keysplayr]

Awaiting for those more technically knowledgeable to answer this. If factual, then we should see a lot of red faces from those who have argued that malice is the only explanation when something fishy involving Nvidia is concerned. Or is it going to be a silence of the lambs (and red faces anyway)?

It will always be silence. At least until they come to their senses.

 

[3DVagabond]

It will always be silence. At least until they come to their senses.

I said it was done the way it was because it's the easiest way to accomplish the task. Which is what the reference post says as well. We just have disagreement on exactly what "the task" was. I got told that I was a dude trying to conduct brain surgery. Now someone else says basically the same thing, asking if people understand how difficult it would be and if they want to write the code themselves. Now it's OK.

Putting that ocean mesh that covers the entire game level is easier. tessellating models that weren't made with tessellation in mind is easier with simple models with flat surfaces. In the end though, tessellating a cube isn't going to really add a lot of detail. It does accomplish the bullet point check mark though, and low and behold, it makes the sponsor look good to boot.

 

[Keysplayr]

I said it was done the way it was because it's the easiest way to accomplish the task. Which is what the reference post says as well. We just have disagreement on exactly what "the task" was. I got told that I was a dude trying to conduct brain surgery. Now someone else says basically the same thing, asking if people understand how difficult it would be and if they want to write the code themselves. Now it's OK.

Putting that ocean mesh that covers the entire game level is easier. tessellating models that weren't made with tessellation in mind is easier with simple models with flat surfaces. In the end though, tessellating a cube isn't going to really add a lot of detail. It does accomplish the bullet point check mark though, and low and behold, it makes the sponsor look good to boot.

Ah, then you can turn the tesselation off.

 

[sontin]

tessellating models that weren't made with tessellation in mind is easier with simple models with flat surfaces. In the end though, tessellating a cube isn't going to really add a lot of detail. It does accomplish the bullet point check mark though, and low and behold, it makes the sponsor look good to boot.

How many times will you repeat this? In every thread i'm reading something about nVidia i see you and your complains about Tessellation in Crysis 2 at the moment.
So tessellating a "cube" doesn't add more details? Wow, then Crytek must have a magic pen:

 

[3DVagabond]

Ah, then you can turn the tesselation off.

And when reviewers do this are you going to be OK with that?

Please, stop pretending that you don't get it. This benefits nVidia in benchmarks. Most gamers, whether nVidia or AMD hardware users, are not going to be able to use this patch maxed out. Reviewers will though and it will skew the results between the two brands.

 

[96Firebird]

And when reviewers do this are you going to be OK with that?

Please, stop pretending that you don't get it. This benefits nVidia in benchmarks. Most gamers, whether nVidia or AMD hardware users, are not going to be able to use this patch maxed out. Reviewers will though and it will skew the results between the two brands.

Are you really that concerned about what the benchmarks show? Is it because they show AMD being slower than Nvidia? Why does that matter so much? The benchmarks show what the user will experience, whether or not they like what they see is up to the reader.

Skew the results? I don't see how the results are skewed if they reflect on what happens in the game. A person looking to play BF3 isn't going to look at the Crysis 2 DX11 benchmarks to make a decision. And if they do, they should get their head checked...

 

[dunno99]

I said it was done the way it was because it's the easiest way to accomplish the task. Which is what the reference post says as well. We just have disagreement on exactly what "the task" was. I got told that I was a dude trying to conduct brain surgery. Now someone else says basically the same thing, asking if people understand how difficult it would be and if they want to write the code themselves. Now it's OK.

Putting that ocean mesh that covers the entire game level is easier. tessellating models that weren't made with tessellation in mind is easier with simple models with flat surfaces. In the end though, tessellating a cube isn't going to really add a lot of detail. It does accomplish the bullet point check mark though, and low and behold, it makes the sponsor look good to boot.

Definitely does seem like the tessellation is a "post" production process. Take the normal/parallax map and apply offsets to the highly tessellated underlying geometry. I don't see "issues" with that -- it's a simple implementation that has its own pros and cons. When this is applied to artistically uneven surfaces, it's great. When it's applied to flat surfaces, it's a waste of computation power.

I guess the takeaway lesson is, if the feature looks good for the performance hit on your card, enable it. If it doesn't, disable it.

P.S. I really don't care for nVidia or ATi cards (as in, I don't side with either). I only care if they fit in my case, is cheap, and has good performance. As a game engine programmer, I philosophically really like both what CUDA and Fusion are trying to do; but I really hate having to program through a thickass layer called DirectX (but I do recognize the need for it...except on the XBox360).

 

[3DVagabond]

Are you really that concerned about what the benchmarks show? Is it because they show AMD being slower than Nvidia? Why does that matter so much? The benchmarks show what the user will experience, whether or not they like what they see is up to the reader.

Skew the results? I don't see how the results are skewed if they reflect on what happens in the game. A person looking to play BF3 isn't going to look at the Crysis 2 DX11 benchmarks to make a decision. And if they do, they should get their head checked...

A site reviews ~10 games. Included in the reviewer's guide are ~3 that have atypical results that favor one brand by say 20% or more. It now looks like, overall, that brand's cards are faster than what they are if these games weren't used. Crysis II with the Dx11 patch is one of those games.

You are selecting individual games and saying people should look at that. I don't want to have to own different GPU's for different games. We let it go in the direction of Crysis II without voicing disapproval and it could come to that though. I would like all of the games to be optimized for everyone equally.

This would be similar to AMD's Gaming Evolved titles requiring 2 gig of RAM to run 1080 @ 4xAA for no good reason except the textures look marginally better than they do now.

 

[96Firebird]

So, instead of blaming the brand that does better, how about we blame the brand that does worse so they get their ass in gear and improve their product? This goes for AMD and their lackluster tessellation, and for Nvidia with their lack of power in the higher resolutions.

 

[3DVagabond]

As has been often pointed out, all of these designs involve compromises. AMD wants to keep die size down and efficiency, and yields up. nVidia doesn't concern themselves with that as much and has to reduce costs elsewhere. Shorter PCB, less beefy PWM, less RAM. I'll say it again, If we let game devs cater to the strengths of either brand we are the ones who will suffer in the long run. Just like the hardware has compromises for balance, so should the games to.

 

[tviceman]

Ah, then you can turn the tesselation off.

And developers should turn it off too because it if it runs faster on Nvidia hardware then it MUST BE CHEATING#(*JR@!)(#E*@!~!

 

[The Ultimate]

I also agree with the general sentiment regarding Nvidia locking out people from using an AMD/Nvidia mix for physx. Don't Q&A for it, fine. Don't provide technical support for it, fine. But don't lock it out - it's a crappy business move and prevents physx from getting higher adoption rates. I do not believe the argument that it would cut into sales of higher end Nvidia cards. I think, in fact, it would INCREASE sales of all Nvidia cards.

I think it affects more their midrange and low end market, as AMD is very competitive there and they would had lost sales of nVidia cards in that segment being bought as potential PhysX cards. A GTS 450 would be a nice PhysX card coupled with a Radeon HD 6950 or even with another nVidia card like the GTX 560 Ti, but I think that at this point, PhysX is fading away, pity as it was a nice concept.

 

[Rifter]

As you have gone to great efforts to communicate above, the topic itself is one for which you are emotionally invested to the point of eliciting emotions of anger on your behalf.

An emotional state which is known to cloud one's ability to think clearly and rationally about the topic itself.

Furthermore, you have endeavored to create something entirely artificial in your post such that you could then vilify "it" as if "it" had anything to do with me, my posts, or the spirit of the message I have communicated in this thread.


There is nothing I can do to dispel the multiple straw men you have concocted in your post. They are of your doing, not mine, and were crafted by you to serve your own misguided purpose, one born from a state of emotional investment in the topic itself, as you amply demonstrated.

Nothing productive will come from my attempting to engage you in rational discourse with the objective of dispelling your keen desire to see shadows of strawman where none exist. Make the devil of me if you find therapeutic benefit in the exercise, release that anger so the healing can start.

Its simply not healthy, not mentally healthy, for you to be so emotionally invested in something as inane as an online forum to the extent that it elicits emotions of anger from you when you are/were faced with the prospect that there are/were marketeers in the audience. Such a co-dependence with the makeup of a forum community is not healthy.

Well said, its amazing how much some people are so passionate about video cards.

Dont get me wrong i love computer hardware and overclocking and the competitiveness of building a great system as much as the next computer enthusiast but when it gets to the point of huge emotional involvement there is a problem.

 

[Keysplayr]

And when reviewers do this are you going to be OK with that?

Please, stop pretending that you don't get it. This benefits nVidia in benchmarks. Most gamers, whether nVidia or AMD hardware users, are not going to be able to use this patch maxed out. Reviewers will though and it will skew the results between the two brands.

Reviewers won't do this unless it's for a distinct purpose. Showing performance with or without certain features have been done in the past. Isn't any reason it should be different now. Yes 3DVagabond, I'd be fine with that.

I'm not pretending that I don't get "it". I get it perfectly man. It's just that my understanding of things seems a bit crisper and clearer. Your view, to me, appears severely distorted and bent on contempt. That's cool, but holy chizz man don't expect me to take you seriously. LOL. It appears like it's a game to you or something. Well allright then.

 

[Will Robinson]

If BF3 contained a sequence requiring fast hashing/number crunching ala password cracking/bitcoin mining etc the NVDA cards would get disintegrated by ATi's architecture.
I can imagine the howls of "unfair",rigged", "OMG we wuz robbed"!...and yet when we see(sic) vast oceans of tessellation that obviously favor NVDA's hardware...its "Aww....Jen takes care of us so well"
It's disgusting.:thumbsdown:

 

[tviceman]

If BF3 contained a sequence requiring fast hashing/number crunching ala password cracking/bitcoin mining etc the NVDA cards would get disintegrated by ATi's architecture.
I can imagine the howls of "unfair",rigged", "OMG we wuz robbed"!...and yet when we see(sic) vast oceans of tessellation that obviously favor NVDA's hardware...its "Aww....Jen takes care of us so well"
It's disgusting.:thumbsdown:

If this is so easily imagined, then could you please link a thread here on anandtech forums the last time there were several nvidia users complaining about how a particular game and/or application was rigged in favor of AMD cards and nvidia users said it was unfair "OMG we wuz robbed"??? Please? Find me a thread on Anandtech where there were several nvidia users saying these things about anything. Heck, there was a six month gap when AMD had sole possession of the DX11 market AND the ability to deliver 3+ monitor gaming setups. Surely you can find something during that time frame when there was an outcry from Nvidia users complaining about how unfair things were.....right? RIGHT?

Please? Go ahead, start searching. I'm guessing it's going to take you awhile though, so if you're up to the task of backing up your own words then I recommend you start sooner than later. I imagine you'll just dismiss this whole statement and have some grand excuse for not backing up your own words. You are just as much a biased, accusation, bait machine for AMD as rollo was for Nvidia.

 

[3DVagabond]

Reviewers won't do this unless it's for a distinct purpose. Showing performance with or without certain features have been done in the past. Isn't any reason it should be different now. Yes 3DVagabond, I'd be fine with that.

I'm not pretending that I don't get "it". I get it perfectly man. It's just that my understanding of things seems a bit crisper and clearer. Your view, to me, appears severely distorted and bent on contempt. That's cool, but holy chizz man don't expect me to take you seriously. LOL. It appears like it's a game to you or something. Well allright then.

Don't respond to my posts, then. If you aren't taking me seriously then you must be just trolling me. That would explain perfectly why you respond while ignoring or twisting any points in my posts.

 

[happy medium]

Why does nvidia cheat so much?

To outdo Intel and there cheating?
Is this a serious question or did someone forget to take business 101 ?