Is there more to it than TWIMTBP? (Personal commentary)

[SunnyD]

In light of all the negativity about the whole AMD vs. NVidia and TWIMTBP "lockouts" and such, I wanted to throw a little personal experience commentary in on the subject. Please note, this is not even remotely intended to be any sort of flame fest regarding either camp, nor is it an attempt at fanboy trolling. I really, truly have an interesting facet of the situation to share.

#1 - I am an AMD/ATI user, proudly running an AMD CPU (940BE) on an AMD/ATI Chipset (790GX) motherboard with an ATI GPU (4890). With that out of the way...

I know we're all tired of hearing about either group's marketing department throwing their developer relations dollars around to provide "user experiences" (eg: feature "lock outs") to specific titles. But there may actually be more to it than actually appears.

Particularly in the real of anti-aliasing, I recently had the opportunity to beta test an upcoming title which will remain unnamed due to various legalities. In any even, the latest round of testing that was done had introduced some rather severe issues in terms of stability with my particular system, to the point where the title was essentially unplayable. After consulting with the developers and community, ATI users had been directed to turn off anti-aliasing options in the title completely. This in particular had seemed to solve my issues, as I was now able to actually test the title for a length of time far exceeding the "few minutes at best" I was seeing earlier. After consulting further with the community, it appears that NVidia users were not affected, and the developers for the title did indeed state that there appeared to be some sort of bug in the anti-aliasing features exposed by the current ATI drivers (most were reporting using 9.12), or possibly in the developer's implementation of AA.

As far as I am aware, this particular title is not on the TWIMTBP program, at least not publicly as of yet. My point here is that there may actually be more of a need for these sort of relations programs behind the scenes than just feature "lock-outs" and whatnot. Again, as discussed in previous threads, it is one thing for vendors to go knocking on developers' doors providing them with an incentive to implement vendor-specific features. The onus should be on the developer to request assistance from a willing vendor if specific features are needed.

Of course, the bewildering part of all of it still lies in using what should be vendor agnostic APIs (DirectX for example) and developing for the lowest common denominator, which isn't done. But again, this isn't done in some cases for good reason - sometimes you want to use a feature that is newer than the API provides to provide that little sparkle that makes your title that much more appealing.

I guess what I am trying to say here folks is that there are differences even in what appears to be a common functionality between both camps. So before we go jumping to conclusions, we really need to see what exactly is at play here. The realm is just too gray to say anything definitively otherwise.

 

[Jovec]

A bug in a beta test? Do tell...

ATI will listen to legitimate questions about their core driver functionality. They'll even respond with a fix, slower or faster, depending on their internal development schedule and how they rate the severity of the bug. Nvidia will do the same. And of course there has never been a case where a developer was convinced his code was not to blame, insisting fault lies elsewhere.

TWIMTBP is nothing more than a marketing tactic. We understand it, and we accept it as long as they don't cross the line. TWIMTBP on a PhysX title? No problem, arguably even legitimate. TWIMTBP on a title that's feature equivalent on ATI? No problem there too, Marketing 101. Purposely restricting features on one company's product because their competitor provided you with compensation through some combination of direct payment, developer support, and marketing dollars? That's crossing the line. I don't blame Nvidia either, but rather Eidos. It would never go this far obviously, but eventually that path could lead to something like ATI cards only rendering 2d sprites on a given title because Nvidia paid truck loads of cash.

 

[DaveSimmons]

This says something about the shoddy QA work of those game developers, if they don't have a good mix of test machines with both intel and AMD CPUs, ATI and nVidia cards.

 

[tviceman]

If I was AMD I would want to be doing the same thing Nvidia is with TWIMTBP - having a team of software engineers to write code specifically to leverage my company's GPU hardware. It's wrong to have the software check and disable certain features on other vendor's hardware (if that is really happening), but beyond that it's a great program that helps developers better utilize the GPU's we have in our systems.

 

[Dribble]

This says something about the shoddy QA work of those game developers, if they don't have a good mix of test machines with both intel and AMD CPUs, ATI and nVidia cards.

It's just a fact nvidia have a much larger team working with devs and a much better programme in TWIMTBP then ati. As with most things the harder you try the better the results.

Even if this isn't a TWIMTBP title the devs still probably develop with nvidia graphics cards because nvidia gave them to them for free, and still get support from nvidia techies. Ati just give out less kit and offer less support so get worse results.

 

[Rezident]

Purposely restricting features on one company's product because their competitor provided you with compensation through some combination of direct payment, developer support, and marketing dollars? That's crossing the line. I don't blame Nvidia either, but rather Eidos.

And Rocksteady. Also, I definitely do blame nvidia for these type of business practices.

 

[Forumpanda]

beta test an upcoming title which will remain unnamed due to various legalities. .. some sort of bug in the anti-aliasing features exposed by the current ATI drivers or possibly in the developer's implementation of AA.

I am no game developer, but even to me that is awfully vague and without any proof or substance.
'some sort of bug' is what you tell a guy that has no technical understanding, or when they know exactly what is happening but they do not wish to disclose the actual cause.
I want actual information before I can even form an opinion, the only thought that crosses my mind is that the developer responsible for that part of the code is not the most competent programmer.

But again, this isn't done in some cases for good reason - sometimes you want to use a feature that is newer than the API provides to provide that little sparkle that makes your title that much more appealing.

Yes, physx would be an example of such a feature, standard AA code is not, it has been proven that even the fabled batman AA code is using no special features of any hardware.

I guess what I am trying to say here folks is that there are differences even in what appears to be a common functionality between both camps

That may be what you are trying to say, but that is not proved anywhere in your post, the only example you give is vague and without any concrete answers.

Multiple times do you point out that you have been 'consulting' various people, if you have actually researched the issue, then why do you not provide any relevant findings.
Does this problem show up on all ATI cards regardless of generation?
Do all people using the newest drivers experience this problem?

As far as information goes, neither of our posts contain any.

 

[exar333]

If I was AMD I would want to be doing the same thing Nvidia is with TWIMTBP - having a team of software engineers to write code specifically to leverage my company's GPU hardware. It's wrong to have the software check and disable certain features on other vendor's hardware (if that is really happening), but beyond that it's a great program that helps developers better utilize the GPU's we have in our systems.

QFT.

ATI should really be more aggressive in this front, and not rely 100% on outside devs to make their HW optimized. Plus, they get their name on some marquee titles and people will realize they are just as legitimate as NV.

As HW enthusiasts here on AT, we are much more knowledgable about what product is better (e.g. 8800GT is >>>>>9400GT) but many people just see the NV logo on their games and assume thats the card to get.

 

[SunnyD]

I am no game developer, but even to me that is awfully vague and without any proof or substance.
'some sort of bug' is what you tell a guy that has no technical understanding, or when they know exactly what is happening but they do not wish to disclose the actual cause.

I want actual information before I can even form an opinion, the only thought that crosses my mind is that the developer responsible for that part of the code is not the most competent programmer.

I wish I could flat out give you that information, but as always I've already said too much with respect to the NDA that is in place. For this reason alone, I'm sure you can understand why I had to be so vague. However, given a romp over in the PC Gaming section, I'm sure you can fill in several of the blanks at your leisure.

Yes, physx would be an example of such a feature, standard AA code is not, it has been proven that even the fabled batman AA code is using no special features of any hardware.

Actually, I wasn't referring to PhysX, as that is an API in itself, and not part of any overarching framework such as OpenGL or DirectX. I was more discussing vendor extensions in OpenGL or DirectX which are exposed indirectly through an extension mechanism.

Multiple times do you point out that you have been 'consulting' various people, if you have actually researched the issue, then why do you not provide any relevant findings.

Does this problem show up on all ATI cards regardless of generation?
Do all people using the newest drivers experience this problem?

Again, as per the reason given above, I have to be vague. I have provided as much information as I can publicly. Yes, this problem shows up on a wide range of ATI cards (I have seen reports of the issue all the way down to the 2000-series). Of the information I have seen, those that have provided information publicly have stated they are using 9.12. My gut says it's not just 9.12's though, but also earlier drivers, though a large portion of the beta is too dumb to figure out what version they're using. Such is the nature of the beast.

QFT.
ATI should really be more aggressive in this front, and not rely 100% on outside devs to make their HW optimized. Plus, they get their name on some marquee titles and people will realize they are just as legitimate as NV.

As HW enthusiasts here on AT, we are much more knowledgable about what product is better (e.g. 8800GT is >>>>>9400GT) but many people just see the NV logo on their games and assume thats the card to get.

I agree completely in principle. However as a consumer and a developer myself, I also would prefer that the vendors adhere to a common standard simplifying development and removing the need for vendor-specific "features".

 

[v8envy]

And here's the view from the Linux side.

Because of the relative popularity of the nvidia hardware vs anything ATI ever made most of the interesting Linux software which makes use of 3D hardware is developed, tested and run on NV hardware. It's what most of the developers have in their machines.

So ATI users are constantly being broken by software and driver updates. Not completely because the ATI driver guys (open and closed source) can't get it together, but mostly because their installed user base is so much smaller. The constant breakage and lack of feature parity in turn makes educated users stay away from ATI hardware on Linux, perpetuating the problem.

In other words, even without TWIMTBP in the picture, the size difference in the user base causes a much uglier experience for the minority (ATI users). With no sign of that changing any time soon.

 

[Seero]

Here is another developer's point of view.

Use whatever works. Least cost possible. All help are welcome. Nvidia want to give me something for free? Thank you. ATI have a problem with that? Sorry.

 

[soccerballtux]

Here is another developer's point of view.

Use whatever works. Least cost possible. All help are welcome. Nvidia want to help me something for free? Thank you. ATI have a problem with that? Sorry.

If the developer turns off special effects (not PhysX) in a game if you play on ATI, then your analogy does not apply.

 

[tviceman]

If the developer turns off special effects (not PhysX) in a game if you play on ATI, then your analogy does not apply.

Are there any other games besides Batman where universally supported features were disabled when specific GPU vendors were present?

 

[Creig]

Here is another developer's point of view.

Use whatever works. Least cost possible. All help are welcome. Nvidia want to help me something for free? Thank you. ATI have a problem with that? Sorry.

Want to sell the highest number of copies of your apps as possible? Make sure they work well on both Nvidia and ATi systems, free help or no free help.

In the end, it's still the developers responsibility to ensure that their product works reliably on the greatest number of systems possible. If they don't, they only have themselves to blame.

 

[Seero]

If the developer turns off special effects (not PhysX) in a game if you play on ATI, then your analogy does not apply.

Why not?

 

[Seero]

Want to sell the highest number of copies of your apps as possible? Make sure they work well on both Nvidia and ATi systems, free help or no free help.

In the end, it's still the developers responsibility to ensure that their product works reliably on the greatest number of systems possible. If they don't, they only have themselves to blame.

Well said. AA was coded by Nvidia's people, QA by Nvidia on varies Nvidia Hardware in Nvidia's Lab, but not on ATI's hardware. Thus to ensure nothing breaks, the function is only available where the precondition is satisfied.

If you don't know, the game works fine on both vendors. To be specific, it works fine for any video card that support Dx9/10.

ATI is capable of making AA work too, but they claim it is too time consuming. That is why AA is disabled on ATI's hardware.

 

[Rezist]

Well the AA does work in Batman for ATi no work necessary except removing a block. I'm sure ATi could help with that if need be but that shouldn't be that hard I wouldn't think.

 

[Creig]

Well said. AA was coded by Nvidia's people, QA by Nvidia on varies Nvidia Hardware in Nvidia's Lab, but not on ATI's hardware. Thus to ensure nothing breaks, the function is only available where the precondition is satisfied.

If you don't know, the game works fine on both vendors. To be specific, it works fine for any video card that support Dx9/10.

ATI is capable of making AA work too, but they claim it is too time consuming. That is why AA is disabled on ATI's hardware.

B:AA AA (that's a lot of A's) works just fine on ATi hardware, as has been proven by countless people who simply overrode the Vendor_ID lockout. Nvidia blames the developer for the lockout. The developer says Nvidia legal team won't allow them to remove the lockout. One of the two isn't telling the truth.

Personally, I believe the developer. Why would a game developer purposely include code that would artificially make their game look worse on roughly 1/2 the systems out there? That is completely counter to their goal of selling as many copies of the game as possible.

The only company that's benefiting from this AA lockout is Nvidia.

 

[Shilohen]

Here is another developer's point of view.

Use whatever works. Least cost possible. All help are welcome. Nvidia want to give me something for free? Thank you. ATI have a problem with that? Sorry.

And here is my developer PoV.

All help is welcome, unless it's a Greek gift (reference to Trojan horse, dunno if that expression is used in English, hence the explanation).

Following your idea, we should go cry to Microsoft when a web page and/or JavaScript does not react as desired. Then Microsoft should help you make it work and obviously test it only in MSIE. Then, when you end up in production, the page doesn't work in Firefox, Chrome, Safari, Opera cannot be read by screen reader and isn't accessible for mobility reduced users. So that would be W3C, Mozilla, Google and Apple's fault for not coming to you and code in your place to follow the standards correctly? That sounds not only unreasonable, but also irresponsible to me.

This is actually exactly why I think the TWIMTBP program is a dangerous one and not helping the gaming industry at all. Standards are important, developers should follow them completely even if sometimes it means working harder or take a seemingly counter-intuitive way to achieve their goals so that outside help or proprietary solution should not be needed. With TWIMTBP, nVidia is making developer lazy, preventing them to gain the experience required for future projects and as an additional evil, produces vendor locked solutions that are never good for gamers in the end. Right now it's kind of crazy, if a basic feature doesn't work in a game, the developers can whine on how much a specific company didn't give them hardware or send them free developers, instead of questioning their own skill and code which is almost 100% certainly the source of the problems as the said feature does work in other titles.

My 2¢

 

[exar333]

I agree completely in principle. However as a consumer and a developer myself, I also would prefer that the vendors adhere to a common standard simplifying development and removing the need for vendor-specific "features".[/QUOTE]

Just to be clear, I don't advocate vendor-specific features (or disabling features for other vendors) but optimizing the universal code that is there. If AMD wants to tweak the AA on a big AAA title, and help it run faster on AMD cards, awesome! NV is welcome to do this as well.

I just think it would be great for AMD to get their name out there and on more products than just their own. Personally, I don't care about brands or whatnot. If more people understood that two good brands existed they might look a little more closely at what is the better product and not just the name. Many people probably wouldn't care, and base their decision on something other than facts (like a NV card once burned-down their grandma's house or something).

 

[Shilohen]

I just think it would be great for AMD to get their name out there and on more products than just their own.

This is what I don't get. I know it's part of the culture now, but I find it so much like "This site was optimized for IE 5+ in 800x600"... Why should a game EVER be linked to a graphic card vendor in any way? Why is it so well seen in the community? It's a bit like if Ford was sponsoring a road and then tell people how great the road is, except if you're driving a Toyota... Those are two things that should be left separated like Church and State imho, else there's just too much abuse possibilities, even if the intentions are good.

 

[exar333]

This is what I don't get. I know it's part of the culture now, but I find it so much like "This site was optimized for IE 5+ in 800x600"... Why should a game EVER be linked to a graphic card vendor in any way? Why is it so well seen in the community? It's a bit like if Ford was sponsoring a road and then tell people how great the road is, except if you're driving a Toyota... Those are two things that should be left separated like Church and State imho, else there's just too much abuse possibilities, even if the intentions are good.

I don't really disagree with you in principle, but the reality is that marketing is here to stay. It would be great IMHO if no games were linked to specific vendors, but NV already has a (arguably) successful campaign, and AMD needs to follow-suit if they want to expose their product to the same audience.

Not sure how long you have been gaming, but I didn't always feel this way. If you go back 10-12 years the 3DFX Glide games really revolutionized the gaming industry. It WAS helpful because the specific vendor pushing the tech did so because the existing tools (software and hardware) did not do what they needed. Jump to the last 5-7 years, and both AMD and NV have had competitive products trading performance punches, back and forth. It is not as important now, to have a vendor-specific game to showcase the new GPU.

 

[Shilohen]

I don't really disagree with you in principle, but the reality is that marketing is here to stay. It would be great IMHO if no games were linked to specific vendors, but NV already has a (arguably) successful campaign, and AMD needs to follow-suit if they want to expose their product to the same audience.

I'm indeed an idealist. However, I don't know if it's always going to be there. Those are expenses for the card makers. The question is how much they get in return. It does seem that nVidia is getting some success, although I believe they reduced its budget for the incoming year, but was the success due to the program or simply to the fact that their card was better during that period? Cutting in such program could be a way to sell video cards cheaper (albeit really not much) and thus reach a better performance/price ratio and, from the feel of it, Fermi might need that...

Not sure how long you have been gaming, but I didn't always feel this way. If you go back 10-12 years the 3DFX Glide games really revolutionized the gaming industry. It WAS helpful because the specific vendor pushing the tech did so because the existing tools (software and hardware) did not do what they needed. Jump to the last 5-7 years, and both AMD and NV have had competitive products trading performance punches, back and forth. It is not as important now, to have a vendor-specific game to showcase the new GPU.

True, innovation is where standards cannot help, for the features rarely reach standard status right when they're released. Maybe I should tone down my opinion to allow vendors to help developers implements proprietary features, like PhysicX. However, I also think that vendors should work with the community to make those features standards and drop without trying to tax the others. After all, they do keep the technology edge then since their product most likely supported the feature before competition.


Regards,

 

[waffleironhead]

Here is another developer's point of view.

Use whatever works. Least cost possible. All help are welcome. Nvidia want to give me something for free? Thank you. ATI have a problem with that? Sorry.

and the nV apologists come out of the woodworks...

 

[Schmide]

and the nV apologists come out of the woodworks...

+ infinity