Does GameWorks influences AMD's Cards game performance?

[futurefields]

AMD makes GPU's that kill on synthetic benchmarks. Nvidia makes GPU's that kill in realtime game scenarios.

Personally I got sick of broken features, like not being able to use MSAA in GTA5, or several maps in Advanced Warfare never working right. That's why I bought an Nvidia card recently.

If you can't beat 'em, join 'em.

I've got games that need to be played, simple as that.

 

[Silverforce11]

Personally I got sick of broken features, like not being able to use MSAA in GTA5, or several maps in Advanced Warfare never working right. That's why I bought an Nvidia card recently.

If you can't beat 'em, join 'em.

I've got games that need to be played, simple as that.

Your 7950 runs GTA V + MSAA well? Not even Titan X can. There's always gonna be bugs whatever hardware you use frankly, as someone who used Intel, AMD, NV, that's been the truth in my experience.

But certainly this reinforces the point I raised earlier, gamers don't care why GW titles run badly on AMD, they just want to play. If they keep seeing AMD performing badly in new games, they will pay more for NV because it's "better".

Me included. If NV gets a lot of AAA titles into GameWorks, games that I want to play, I'll have to ditch AMD GPUs moving forward.

 

[Techhog]

People here should stop assuming that Gameworks cripples other IHVs unless we test out each sample to see how poorly it runs ...

Correlation DOES NOT imply causation!

You're using that term incorrectly. What we're seeing here isn't simply correlation; it's a pattern with strong supporting facts. Can you explain why GW titles all need months of optimizations from AMD and/or the developer, but Trine 3 runs as expected on both vendors without an optimized driver? Why only GW titles give AMD trouble?

 

[SirPauly]

Imagine a world where starting from GeForce 2, every single next generation feature from compute shaders to global illumination to SM3.0 to tessellation was inserted into AAA games as part of NV's proprietary GW's SDK? PC gaming would not be what it is today and you know it.

I have no problem with a few fidelity enhancements for a customer base -- I would have a problem if an IHV did actual harm.

 

[Techhog]

I have no problem with a few fidelity enhancements for a customer base -- I would have a problem if an IHV did actual harm.

What is your definition of "actual harm?"

 

[ThatBuzzkiller]

PhysX and Apex are multi-platform --- The reason why AMD doesn't support GPU physX is they don't desire to support CUDA. With Apex, nVidia is slowly moving to Compute.

No, the reason why AMD doesn't support CUDA is because Nvidia does not wish to license out the middleware to AMD ...

If Nvidia did give the chance to offer it to AMD, they would take advantage to support it as soon as possible in order to gain a leverage in the HPC market ...

 

[ThatBuzzkiller]

You're using that term incorrectly. What we're seeing here isn't simply correlation; it's a pattern with strong supporting facts. Can you explain why GW titles all need months of optimizations from AMD and/or the developer, but Trine 3 runs as expected on both vendors without an optimized driver? Why only GW titles give AMD trouble?

How do you know if it's not on the issue of the developers themselves or possibly even AMD's compiler ?

 

[Silverforce11]

How do you know if it's not on the issue of the developers themselves or possibly even AMD's compiler ?

It's probably a case of lazy sell-out devs + GameWorks closed source, obfuscated code which only NV has access to. Takes two to tango.

 

[SirPauly]

No, the reason why AMD doesn't support CUDA is because Nvidia does not wish to license out the middleware to AMD ...

If Nvidia did give the chance to offer it to AMD, they would take advantage to support it as soon as possible in order to gain a leverage in the HPC market ...

Cuda is an API, that is proprietary controlled and if AMD supported it, it would of garnered a near monopoly control for nVidia and their hardware vision. It may not be wise to invest into a competitors proprietary API based on control and vision.

 

[ThatBuzzkiller]

It's probably a case of lazy sell-out devs + GameWorks closed source, obfuscated code which only NV has access to. Takes two to tango.

Again how do we know that without any valid data to back that up ?

 

[nvgpu]

http://www.ngohq.com/news/14254-physx-gpu-acceleration-on-radeon-update.html

The bad news is we still don’t have access to any HD 4800 hardware yet. It is very important for this project to receive AMD’s support on both developer and PR levels. It seems that AMD still is not being cooperative, we get the feeling that they want this project to fail. Perhaps their plans are to strangle PhysX since AMD and Intel have Havok.

Before we get to the good news, I’m going to ask you to hold on to something steady as some of you are going to feel a bit dizzy after you hear this. The truth is… Nvidia is now helping us with the project and it seems they are giving us their blessings. It’s very impressive, inspiring and motivating to see Nvidia's view on this. Why they help us? My best guess would be: They probably want to take on Intel with CUDA and to deal with the latest Havok threat from both AMD and Intel.

AMD's and Intel's NIH syndrome working as usual.

http://en.wikipedia.org/wiki/Not_invented_here

 

[Silverforce11]

Cuda is an API, that is proprietary controlled and if AMD supported it, it would of garnered a near monopoly control for nVidia and their hardware vision. It may not be wise to invest into a competitors proprietary API based on control and vision.

I'm quoting this so nvgpu can perhaps read again why AMD doesn't want to get onboard with GPU PhysX. Hopefully he can understand when it comes from SirPauly.

 

[SirPauly]

What is your definition of "actual harm?"

To take existing developer code or add code to purposely do harm to their competitor.

All one has to do is investigate the Gamework features and see what the performance hit is.

 

[ThatBuzzkiller]

Cuda is an API, that is proprietary controlled and if AMD supported it, it would of garnered a near monopoly control for nVidia and their hardware vision. It may not be wise to invest into a competitors proprietary API based on control and vision.

It maybe a proprietary API but that doesn't mean AMD shouldn't support it when the platform has a mild software ecosystem for GPU acceleration ...

AMD may not be able to influence the specifications for compute capability but they can still control their own compilers to support the base functionality of a certain compute capability device version ...

 

[SirPauly]

http://www.ngohq.com/news/14254-physx-gpu-acceleration-on-radeon-update.html





AMD's and Intel's NIH syndrome working as usual.

http://en.wikipedia.org/wiki/Not_invented_here

AMD likes PhysX and dynamics and had silent talks with nVidia about PhysX. I believe AMD asked permission to use PhysX without Cuda and desired to port it to compute or OpenCl. Similar to what AMD did with Havok AnD Intel.

 

[ThatBuzzkiller]

AMD likes PhysX and dynamics and had silent talks with nVidia about PhysX. I believe AMD asked permission to use PhysX without Cuda and desired to port it to compute or OpenCl. Similar to what AMD did with Havok AnD Intel.

It's kinda useless at this point for AMD to support Physx when Havoks is also becoming GPU accelerated ...

Nvidia is a little too late to popularize GPU accelerated physics when they had the chance to ...

Physx could have been a common ground had Nvidia not squandered it with their stubborness ...

Now Nvidia is pretty much forced to offer automatic multicore CPU support to be even competitive with what the rest of the industry is offering or doing ...

Even game developers themselves are implementing their own physics system in their own game engines such as Ubisoft doing cloth simulation in ACU with AnvilNEXT on direct compute!

 

[Techhog]

How do you know if it's not on the issue of the developers themselves or possibly even AMD's compiler ?

Because all of those games have only one thing in common...

 

[ThatBuzzkiller]

Because all of those games have only one thing in common...

What they also have in common could be poor programming ...

 

[Techhog]

What they also have in common could be poor programming ...

So, you're trying to argue that it's pure coincidence that this only happens with GameWorks titles?

 

[Silverforce11]

Again how do we know that without any valid data to back that up ?

You won't ever get that valid data, because developers aren't going to publicly claim they are pathetic, lazy or choose not to optimize for AMD. But there's one thing that's already on public record, GW devs cannot share GW optimizations with AMD. That's coming from an NV spokes person directly. It's also on the GW contract.

What we can go on, is that EVERY SINGLE GW game so far has ran poorly on AMD on release. Also, here's the kicker. EVERY SINGLE GW game has received an AMD optimization patch 3-4 months post launch, suddenly bringing performance up massively.

Take that as you will, certainly correlation does not equal causation, but trends are open for analysis and so far, the trend is GW negatively affect AMD. If you deny that, you obviously have not paid attention.

 

[Rvenger]

Reviewers and people who run out of the box settings.

My main Titan has a 700W limit so I don't but a lot of people do.

Either way, dunno what's causing crazy power use on AMD in pCARS, but that's the problem at least for my tahiti.

Maybe I should try it on my 295x2 with the power slider maxed?

 

[ThatBuzzkiller]

So, you're trying to argue that it's pure coincidence that this only happens with GameWorks titles?

I'm not saying that ...

I'm only saying we need some ACTUAL data from the Gameworks demo samples to draw that conclusion ...

 

[ThatBuzzkiller]

You won't ever get that valid data, because developers aren't going to publicly claim they are pathetic, lazy or choose not to optimize for AMD. But there's one thing that's already on public record, GW devs cannot share GW optimizations with AMD. That's coming from an NV spokes person directly. It's also on the GW contract.

What we can go on, is that EVERY SINGLE GW game so far has ran poorly on AMD on release. Also, here's the kicker. EVERY SINGLE GW game has received an AMD optimization patch 3-4 months post launch, suddenly bringing performance up massively.

Take that as you will, certainly correlation does not equal causation, but trends are open for analysis and so far, the trend is GW negatively affect AMD. If you deny that, you obviously have not paid attention.

I'm not denying the trend, what I'm saying is that we're all too brash to deduce that it truly is hurting AMD owners ...

 

[SirPauly]

You won't ever get that valid data, because developers aren't going to publicly claim they are pathetic, lazy or choose not to optimize for AMD. But there's one thing that's already on public record, GW devs cannot share GW optimizations with AMD. That's coming from an NV spokes person directly. It's also on the GW contract.

What we can go on, is that EVERY SINGLE GW game so far has ran poorly on AMD on release. Also, here's the kicker. EVERY SINGLE GW game has received an AMD optimization patch 3-4 months post launch, suddenly bringing performance up massively.

Take that as you will, certainly correlation does not equal causation, but trends are open for analysis and so far, the trend is GW negatively affect AMD. If you deny that, you obviously have not paid attention.

When I gamed on ATI hardware, nVidia centric titles, there was an out-of-the-box performance advantage and a fidelity feature but with a few driver releases or a patch, performance was there again --- never whined -- talked and discussed about ATI's strengths and offered constructive views for a more robust developer relations for ATI.

 

[Edgy]

I can't believe there's even a debate about this issue.

I do not think there is a "smoking gun" conclusive proof available yet (e.g., GW code reveal) but certainly enough strong circumstantial evidence that there may be extremely unfair and anti-competitive practice in GW by NV as well as those developers that may be colluding to this in exchange for money.

I wish some reputable and respected tech site can do an in-depth coverage on topic such as this the attention it deserves instead of parroting the same stuff that everyone else covers.