AMD GPUs seems to have an advantage over the equivalent GeForce models in the BF5 alpha

[mr moose]

11 minutes ago, leadeater said:

More correctly they did not make the compiler hurt performance they just did not utilize the more advanced instruction sets in the CPUs and did not disclose this fact when doing performance comparisons between Intel and AMD CPUs with software compiled with the Intel compiler.

 

Sort of

 

http://download.intel.com/pressroom/legal/ftc/FTC_Final_Executed_Agreement.pdf

 

Nvidia would be able to avoid being obligated to fix it if they were in this situation only by successfully showing that it was a product defect. So sure they could admit it's a defect with the product and no fix it, with all the company image that brings along with it. But at least they will have admitted the GameWorks is a defective product .

You are forgetting that 1. The change Intel made was intentionally crippling AMD CPU'S and B. the Intel compilers is unavoidable.  You cannot just switch it off or have it as an option to software developers  end users.

 

http://www.agner.org/optimize/blog/read.php?i=49

 

Intel did a really big intentional dirty on AMD in the compiler. That is no where near the same as what's happening here, unless you have evidence that says otherwise.

 

 

9 minutes ago, Sierra Fox said:

But you keep saying they are "obligated" to improve how it runs on the competitions hardware. it's an optional setting, it doesn't have to be turned on, with the off, the games run as intended. the fact of the matter is that Gameworks is an NVidia proprietary software, they made it to run on GeForce cards.

 

If it was intentional, sure that's fucked up, but WHY are they obligated to fix optimization issues on a competitor's product that is ultimately an optional setting? what is that that you're effective quoting stating that they have to make sure it doesn't hamper the competition unintentionally?

 

I would like to see the answer to this too.

[leadeater]

Just now, Sierra Fox said:

But you keep saying they are "obligated" to improve how it runs on the competitions hardware. it's an optional setting, it doesn't have to be turned on, with the off, the games run as intended. the fact of the matter is that Gameworks is an NVidia proprietary software, they made it to run on GeForce cards.

 

If it was intentional, sure that's fucked up, but WHY are they obligated to fix optimization issues on a competitor's product that is ultimately an optional setting? what is that that you're effective quoting stating that they have to make sure it doesn't hamper the competition unintentionally?

 

See above post, and the ability to turn it off in games is not always an adequate solution when it's tied to presets, meaning if you want high graphics settings you must then also have GameWorks effects.

 

Pre GameWorks, even with Nvidia partner games, these large performance impacts were not a thing so for me it's not a matter of improve how it runs it a matter of not degrading which was not present before.

 

Plus for a lot of the problems we are not talking about significant work on Nvidia's end or work that wouldn't benefit Nvidia GPUs. Limiting tessellation levels at longer distances so far objects aren't x64 which hurts Nvidia and AMD is a start. After that limiting it's usage on object that don't need it or are not actually on screen, benefit for both again.

 

 

Sorry for how old this is but it does demonstrate the potential performance impact.

[leadeater]

Just now, mr moose said:

You are forgetting that 1. The change Intel made was intentionally crippling AMD CPU'S and B. the Intel compilers is unavoidable.  You cannot just switch it off or have it as an option to software developers  end users.

 

http://www.agner.org/optimize/blog/read.php?i=49

 

Intel did a really big intentional dirty on AMD in the compiler. That is no where near the same as what's happening here, unless you have evidence that says otherwise.

A precedent is still a precedent, exactly the same is not a requirement and that is where Nvidia does need to be careful. Especially as they push more developers over to using GameWorks, like the Intel compiler there were other options but many only wanted to use theirs because it was marketed as the best and believed to be. Intel compiler wasn't unavoidable and GameWorks isn't either.

 

Quote

Many software programmers consider Intel's compiler the best optimizing compiler on the market, and it is often the preferred compiler for the most critical applications. Likewise, Intel is supplying a lot of highly optimized function libraries for many different technical and scientific applications. In many cases, there are no good alternatives to Intel's function libraries.

GameWorks is checking most of the boxes here.

 

[mr moose]

1 minute ago, leadeater said:

A precedent is still a precedent, exactly the same is not a requirement and that is where Nvidia does need to be careful. Especially as they push more developers over to using GameWorks, like the Intel compiler there were other options but many only wanted to use theirs because it was marketed as the best and believed to be. Intel compiler wasn't unavoidable and GameWorks isn't either.

The difference being that Intel's was actually intentionally gimping AMD, while gameworks currently does not and currently isn't even required (yes you can make a game without gameworks).

 

1 minute ago, leadeater said:

GameWorks is checking most of the boxes here.

 

So supplying something that is the best and has no good alternatives is equal to intentionally gimping?  Not sure how your logic works. The issue with Intel was not that their compiler had libraries that were the best, but the fact they intentionally gimped AMD with it.  We don't even have evidence gameworks is the best let alone that they are intentionally gimping AMD with it.

[Starelementpoke]

On 7/4/2018 at 1:46 AM, Dan Castellaneta said:

Well the comments went down the drain pretty fast.

Was expecting it to be the first reply tbh

[Stefan Payne]

8 hours ago, veli2501 said:

It also ignores the fact that maybe AMD could work on getting better performance in more than 4x tessellation.

Why waste transistors for something that nobody really uses?
And if, they most likely to use lower tesselation factors anyway. Not 64x (or 32x??) Like it was done in Hairworks.

 

And even for normal models, you just use more polygons because you know what it does, with Tesselation its not that precise.

 

Ironically you can see that very well with Truform in some Games. AFAIR Serious Sam had an interesting implementation.

8 hours ago, Zodiark1593 said:

If tesselation was at the heart of the issue for AMD cards running Hairworks, wouldn't a dramatic improvement in tessellation performance resolve that?

It could, but why waste Transistors for something you don't really need?
Just because some nice guys misuse it to make their shit look good??

 

Especially in todays time where everything is power limited.

And using transistors for shit you don't need increases your power consumption for nothing.

 

8 hours ago, Zodiark1593 said:

It seems to me that AMD has put a lot of focus on compute performance and extracting the most from it (which DX12 and Vulkan benefit a lot from, especially with Async added in) while neglecting their fixed function portions (like the tesselator) somewhat, while Nvidia has a bit more of a balanced approach here in continuing to improve their fixed function stuff as well as compute.

 

I'm not a gpu architect, so I could just be completely off the mark here...

GPU Architects disagree with what you say...

And because the AMD GCN Design was so "bad", nVidia adopts more and more of it in their Maxwell and Pascal Generation

So it is a pretty neat design, the Problem is just that it doesn't scale that well with more rendering units...

And looking back to the 7800GTX and X1800XL, you'll see that compute is pretty important...

Because due to the architecture there are some programms that run usable on the X1800 (or rather X1900) than on the nVidia Cards due to their architecture... 

4 hours ago, D13H4RD2L1V3 said:

Been looking at other forums and I'm seeing people angry about this? 

 

I think one guy said the 1060 "should smash the 580" as it apparently did in PUBG, Fortnite and similar titles 

Yeah, that's the Problem, although the 580 should be faster. And that is the case when its well optimized for AMD Cards and not just some unoptimized bullshit thrown at them.
PUBG is one of the best examples for a game with shitty code, especially on Consoles.

 

Ni no Kuni 2 is a good example of how well a game can perform on AMD Hardware (Hint: Vega 64 >> 1080ti).

 

 

4 hours ago, D13H4RD2L1V3 said:

I own a 1060 and I've been familiar with the 580 and quite honestly, they trade blows. One doesn't necessarily "smash" the other and when they do perform better, it's not like it's a huge amount minus outliers. Said outliers include Fortnite, where the 1060 apparently had a similar difference delta but in its favor. But looking at PUBG and some others, the difference is smaller to the point where they're almost identical when averaged with one performing slightly better at times 

As stated, PUBG is the worst example because of the bad code.

 

And it all depends on the developer, how competent they are and how well they optimize for both architectures - or not.

 

3 hours ago, leadeater said:

There is a limit to freedom of business, as well as freedom of choice. That is and always has been the essence of the issue.

Exactly.

The Gouvernment has to set the rules and enforce them.

One of the Rules is that, if you are doing business in one area, you must not interfere in a closely related area to tripple competition.

 

 

 

There is just nothing like Technology/IT and the Lawmakers are just old, they don't understand shit.

Like what is the average age for Congress? Like 60 Years old??

 

But even then, there are in principle laws on the books that should prevent such behaviour, just nobody enforces them.

 

But the closest thing that might eventually fit are car manufacturers and 3rd party repair shops. But here, the lawmakers understand the situation and there are laws that force the Car Makers to provide the repair shops with repair manuals and also there are always lawsuits about Warranty and 3rd Party Car Shops...

 

 

 

 

 

3 hours ago, leadeater said:

More games are starting to use Nvidia tools in their development so that there ends the let Nvidia do what they want is best for them. In my view they either fix their tools to not hamper performance or be forced to, fixing it will not hurt them or their customers.

Yes and that is something that damages the market, even if some people try to deny it. But we here in this Forum are just a few people, most just want something to work. They don't care about manufacturers or other shit, they are too old for the Flamewars...

 

And with that said, they see all the benchmarks, the prices of the Graphics cards and some of those guys switch to console because they are fed up with all that bullshit in PC-Gaming, all the issues with Drivers and other shit...

That's the extreme consequence of all this shit...

 

And also there are no real PC-AAA Games no more. Everything is Multi Plattform and sold for either PC and Playstation or all three or even four systems.

 

2 hours ago, Sierra Fox said:

i think @mr moose 's main point in all this, is why is it up to NVidia to ensure that AMD GPUs are optimized for gameworks? So far everything in this thread has been hearsay that NVidia are intentionally using tools that will lower AMD's performance.

First its not heresay.

Second take a look at the Gameworks contracts. some of them either leaked or are available online. And they are worse than the NDA.

Some companys complained that they couldn't even send the others 

Wasn't it even someone from Ubi who took a crap on those contracts??

 

And there are some people who _KNOW_ how some things from the Gameworks Libery works, like the Hairworks thing. And that uses excessive tesselation.

 

2 hours ago, Sierra Fox said:

In the same vein we should all be mad at Apple for their OS not running on what ever hardware people want. Apple made software that works well on their hardware, sure it CAN work on non Apple hardware, but it's much more difficult and is not a guarantee that it will work correctly, if at all.

No, because that is a completely different cup of tea.

Because what Apple does is in no way different what IBM did in 1980 or Commodore with the C64 and later Amiga or Atari with the ST. Or Sony Playstation, Microsoft XBox....

 

They provide a plattform where you can develop software for (or not). They don't even use X86 hardware for that long. Just a bit over 10 years. Before that they used IBM Power PC Chips (OK, only for a short while). Before that they used Motorola PowerPC Chips. And before that the legendary Motorola 68000 series. That is what Apples MAC was based on...

And I have a PowerMAC G4/933 with a Motorola processor. That was the exclusive Chipmaker for Apple for a long time - the longest time!

 

We aren't talking about providing a Plattform with a bundle of Hardware and Software.

We are talking about influencing the software for a 3rd party Plattform (Microsoft Windows) to run only on their Hardware. That is the End Goal for them.

 

OH and Apple uses AMD GPUs, if they use GPUs at the moment. Because they have more freedom with AMD GPUs than they would have had with nVIdia (=driver development for example)...

[leadeater]

28 minutes ago, mr moose said:

The difference being that Intel's was actually intentionally gimping AMD, while gameworks currently does not and currently isn't even required (yes you can make a game without gameworks).

Intel was not utilizing SSE2 and SSE3 when they could of been when compiling code, even that isn't quite the same as the tessellation example. Intel weren't even punished for not using SSE2 and SSE3 on AMD and VIA hardware they were for refusing to state that their compiler does not leading to lower performance than which was possible and then misrepresenting the situation and the performance of the competitor products using Intel compiled software.

 

It's the second part, the part that got Intel in to trouble that applies to Nvidia. If you know your product has an adverse effect on a competitor then depending on circumstances only have two choices, disclose that it does with the how/why or fix it. Who knows maybe they are telling all the developers that GameWorks does this to AMD hardware and developers just don't care or are now fixing it.

 

We mostly only have old examples of GameWorks issues, only new one I can think of would be FF15 which I have and it still utterly destroys the performance of my AMD GPUs (not that I didn't know going in to it).

 

Quote

A Radeon RX Vega 64 roughly matches a GeForce GTX 1070 Ti at Very High settings without GameWorks' influence. Turning those features on causes the Radeon card's performance to hit a similar level as Nvidia's GeForce GTX 1060 6GB with minimum frame rates that struggle to stay over 30. As far as the Radeon RX 580 goes, GameWorks features fall outside of its reach.

https://www.tomshardware.co.uk/final-fantasy-xv-performance-analysis,review-34257-7.html

 

It's a good thing FF15 was updated so you could selectively turn of each GameWorks effect, though you do need to be aware for high presets they are on by default.

 

Also yes at the time you could compile code without the Intel compiler, as I said before both are not and were not required.

 

28 minutes ago, mr moose said:

The issue with Intel was not that their compiler had libraries that were the best, but the fact they intentionally gimped AMD with it.  We don't even have evidence gameworks is the best let alone that they are intentionally gimping AMD with it.

It was the best, that's literally what everyone thought and how Intel marketed it. Though if you did have an Intel CPU then it was the best. You know what else is viewed as the best? Nvidia GPUs, they are though, so it's not surprising that developers would want to use Nvidia tools.

 

28 minutes ago, mr moose said:

So supplying something that is the best and has no good alternatives is equal to intentionally gimping? 

You can supply the best and not have crippling performance impact on the competitor. Again like Intel unless you can demonstrate that it's a defect of the product you may be required to fix it, but only if you have such a large market share to warrant such a thing being handed down and only after a complaint is raised, which Nvidia has the market share and AMD would do if they deem it necessary to try and take such action and think they can win.

 

Nvidia does have monopoly levels of market share, like Intel did/does, and that can indeed be an unfair position to be in when it comes to regulation. Computer graphics and game development is a huge industry with a lot of money in it now and has for the most part been untouched by regulation which is a good thing so I'd rather Nvidia be more proactive and making sure it doesn't become a thing.

[Stefan Payne]

11 minutes ago, leadeater said:

t's the second part, the part that got Intel in to trouble that applies to Nvidia. If you know your product has an adverse effect on a competitor then depending on circumstances only have two choices, disclose that it does with the how/why or fix it. Who knows maybe they are telling all the developers that GameWorks does this to AMD hardware and developers just don't care or are now fixing it.

or they are not allowed to and must only use Gameworks libarys as provided from nVidia...

 

That is the Problem with gameworks!
nViida doesn't deliver code that the developer can implement, they provide a DLL or something like that that the Developer has to use.

There is an Interview with the lead developer of Witcher 3 from PCGH.

Here it is, in the Engl. Version:

http://www.pcgameshardware.de/The-Witcher-3-Spiel-38488/Specials/The-Witcher-3-welche-Grafikkarte-welche-CPU-1107469/2/

 

 

He is very diplomatic but goes as far as he can in taking a piss at Gameworks and nVidia or the decision to use Gameworks at all!

 

 

He says "It was a business Decision".  You can't phrase it much worse than that if you want to keep your job...

 

 

[Humbug]

33 minutes ago, Stefan Payne said:

He says "It was a business Decision".  You can't phrase it much worse than that if you want to keep your job...

Other things he says

-strong feels about the proper use of tessellation

-Nvidia decides whether to even let hairworks run on AMD GPUs.

[Humbug]

It's in the best interest of a game developer to make sure that their game runs as well as possible while looking as good as possible on the widest possible range of hardware. This is in the best interest of the customers as well as the developers. So the tools you use and the middleware you choose to purchase help you achieve that goal. This in turn helps you to sell the maximum number of copies of the game, win-win situation.

 

Thing is when you flip that equation on it's head; now Nvidia is rewarding you for using their middleware. Now it's a management decision because you have a financial incentive to use said black box middleware. That's what happens with gameworks, compromising the previous objective of making the correct technical decisions which lead to the best possible performance:graphics ratio for end customers.

[Jito463]

5 hours ago, mr moose said:

But Intel intentionally made their compiler hurt AMD CPU's

One could easily make the argument that Intel didn't actually "hurt" performance on AMD processors, they just didn't "optimize" for them.  In other words, much the same argument you're making in favor of Nvidia.

[D13H4RD]

On a side note, am I one of those people who prefers Vega 56 over Vega 64?

 

I know a few folks with Vega 56 GPUs, one of which got his on launch and spent quite a while tinkering and playing around with it, even flashing a V64 vBIOS. He did say he’s planning to waterblock it. Wonder how much does that card have left inside.

[Jito463]

15 minutes ago, D13H4RD2L1V3 said:

On a side note, am I one of those people who prefers Vega 56 over Vega 64?

I don't know, are you?  

 

In all seriousness, I prefer the Vega 56 myself.  I think it's more balanced between power draw and performance.

[mr moose]

10 hours ago, leadeater said:

Intel was not utilizing SSE2 and SSE3 when they could of been when compiling code, even that isn't quite the same as the tessellation example. Intel weren't even punished for not using SSE2 and SSE3 on AMD and VIA hardware they were for refusing to state that their compiler does not leading to lower performance than which was possible and then misrepresenting the situation and the performance of the competitor products using Intel compiled software.

 

It's the second part, the part that got Intel in to trouble that applies to Nvidia. If you know your product has an adverse effect on a competitor then depending on circumstances only have two choices, disclose that it does with the how/why or fix it. Who knows maybe they are telling all the developers that GameWorks does this to AMD hardware and developers just don't care or are now fixing it.

No its not the same.  The intel one was intentional to  gimp, it wasn't just a side effect of optimization. 

 

This is from the the actual court finding:

 

Quote

2.3 TECHNICAL PRACTICES

Intel shall not include any Artificial Performance Impairment in any Intel product or require any Third Party to include an Artificial Performance Impairment in the Third Party's product. As used in this Section 2.3, "Artificial Performance Impairment" means an affirmative engineering or design action by Intel (but not a failure to act) that (i) degrades the performance or operation of a Specified AMD product, (ii) is not a consequence of an Intel Product Benefit and (iii) is made intentionally to degrade the performance or operation of a Specified AMD Product. For purposes of this Section 2.3, "Product Benefit" shall mean any benefit, advantage, or improvement in terms of performance, operation, price, cost, manufacturability, reliability, compatibility, or ability to operate or enhance the operation of another product.

In other words they can choose not to look at how their code operates on competing products, but they can not make it gimp them on purpose.

 

10 hours ago, leadeater said:

We mostly only have old examples of GameWorks issues, only new one I can think of would be FF15 which I have and it still utterly destroys the performance of my AMD GPUs (not that I didn't know going in to it).

 

https://www.tomshardware.co.uk/final-fantasy-xv-performance-analysis,review-34257-7.html

 

It's a good thing FF15 was updated so you could selectively turn of each GameWorks effect, though you do need to be aware for high presets they are on by default.

 

Also yes at the time you could compile code without the Intel compiler, as I said before both are not and were not required.

 

It was the best, that's literally what everyone thought and how Intel marketed it. Though if you did have an Intel CPU then it was the best. You know what else is viewed as the best? Nvidia GPUs, they are though, so it's not surprising that developers would want to use Nvidia tools.

 

You can supply the best and not have crippling performance impact on the competitor. Again like Intel unless you can demonstrate that it's a defect of the product you may be required to fix it, but only if you have such a large market share to warrant such a thing being handed down and only after a complaint is raised, which Nvidia has the market share and AMD would do if they deem it necessary to try and take such action and think they can win.

 

Nvidia does have monopoly levels of market share, like Intel did/does, and that can indeed be an unfair position to be in when it comes to regulation. Computer graphics and game development is a huge industry with a lot of money in it now and has for the most part been untouched by regulation which is a good thing so I'd rather Nvidia be more proactive and making sure it doesn't become a thing.

Again, If you want to use the Intel as an example then you have to prove the AMD performance hit is intentional from nvidia and not just a by product of the way they have written it for their cards.

 

6 hours ago, Jito463 said:

One could easily make the argument that Intel didn't actually "hurt" performance on AMD processors, they just didn't "optimize" for them.  In other words, much the same argument you're making in favor of Nvidia.

No, Intel actually intentionally Gimped AMD, their compiler looked up the CPU identifier and applied or ignored instruction sets accordingly.  When Agner spoofed the AMD CPU identifiers his AMD CPU's ran without any of the performance hits.

 

Quote

the Intel CPU dispatcher does not only check which instruction set is supported by the CPU, it also checks the vendor ID string. If the vendor string says "GenuineIntel" then it uses the optimal code path. If the CPU is not from Intel then, in most cases, it will run the slowest possible version of the code, even if the CPU is fully compatible with a better version.

 

This is why they got in trouble,  They did intentionally hurt the performance of AMD processors. 

[leadeater]

2 hours ago, mr moose said:

In other words they can choose not to look at how their code operates on competing products, but they can not make it gimp them on purpose.

Neither does excessive amounts of Tessellation have a benefit for Nvidia products either, in fact it has a negative effect on their products too. Once you know that your product has a negative effect on your competitor and you actively not fix it you then may be challenged as to why and if you cannot show any grounds for not fixing it and the competitor can show you gain an undue market advantage from it you may indeed be required to do something about it, even if that is only to clearly disclose it.

 

2 hours ago, mr moose said:

Again, If you want to use the Intel as an example then you have to prove the AMD performance hit is intentional from nvidia and not just a by product of the way they have written it for their cards.

Their wording was a bug or product defect for unintentional. Once you know your product has said negative impact on your competitor and you actively not fix it and it has no benefit to yourself then you can be accused of intentionally impacting the performance of the competitor. You cannot be held to account for a bug or defect that you did not intend to have that is for sure but if you intentionally gain from it you can be challenged.

 

2 hours ago, mr moose said:

No, Intel actually intentionally Gimped AMD, their compiler looked up the CPU identifier and applied or ignored instruction sets accordingly.  When Agner spoofed the AMD CPU identifiers his AMD CPU's ran without any of the performance hits.

Yes exactly and even as you have said Intel doesn't have to do jack for AMD in that regard, the issue was not disclosing they were doing this and actually saying they were producing optimized code for AMD processors then when challenged on the matter and provided evidence to the contrary they still did not acknowledged it.

 

Nvidia may have to do nothing more than clearly state "Using GameWorks has a negative performance effect on non Nvidia products" as well as disclose any optimizations in the GameWorks tools for their products, like Intel. Personally I say they should be forced to fix it but what ever.

[Belgarathian]

On 7/5/2018 at 12:11 AM, Vode said:

Just wait when the game gets released it will dump AMDs performance. Something like Hairworks or some effect specificly tailored to Geforce hardware will be included.

Probably with no menu option to toggle it until 2-3 weeks post release. 

[aezakmi]

so you can watch the cringe at ultra settings

[mr moose]

39 minutes ago, leadeater said:

Once you know that your product has a negative effect on your competitor and you actively not fix it you then may be challenged as to why and if you cannot show any grounds for not fixing it and the competitor can show you gain an undue market advantage from it you may indeed be required to do something about it, even if that is only to clearly disclose it.

No, you would have to prove the intention was to cause said performance. Remember burden of proof is on the accuser. Failing to fix something to advantage a competitor is not proof of anything.

Quote

You cannot be held to account for a bug or defect that you did not intend to have that is for sure but if you intentionally gain from it you can be challenged.

Exactly, you have to prove there is intentional gain from it, in other words you have to prove Nvidia are doing this intentionally to hamper AMD.  Until then they are not obligated in any way to fix the issue for AMD.

Quote

Yes exactly and even as you have said Intel doesn't have to do jack for AMD in that regard, the issue was not disclosing they were doing this and actually saying they were producing optimized code for AMD processors then when challenged on the matter and provided evidence to the contrary they still did not acknowledged it.

Yes they do, when they intentionally compromise the performance of AMD cpu's as they did the FTC ruling made it quite clear that they had to remedy the situation.  In fact the ruling went as far to make Intel pay for the costs of any developer who used their compiler to recompile their software and make the necessary patches. 
 

Quote

 

Requiring that, with respect to those Intel customers that purchased from Intel a software compiler that had or has the design or effect of impairing the actual or apparent performance of microprocessors not manufactured by Intel ("Defective Compiler"), as described in the Complaint:

1. Intel provide them, at no additional charge, a substitute compiler that is not a Defective Compiler;
2. Intel compensate them for the cost of recompiling the software they had compiled on the Defective Compiler and of substituting, and distributing to their own customers, the recompiled software for software compiled on a Defective Compiler; and
3. Intel give public notice and warning, in a manner likely to be communicated to persons that have purchased software compiled on Defective Compilers purchased from Intel, of the possible need to replace that software.

 

 

 

These are pretty damning directives for merely optimizing for your own product.   Why is that? because Intel did not merely optimize for their own product.  This is why the Intel example is nothing like gameworks. Remember that they are not liable for incidental performance drops in the competition, only artificial (intentional) drops.

 

Quote

Nvidia may have to do nothing more than clearly state "Using GameWorks has a negative performance effect on non Nvidia products" as well as disclose any optimizations in the GameWorks tools for their products, like Intel. Personally I say they should be forced to fix it but what ever.

 

Not like Intel at all.  They are two very different situations.  Unless you have evidence that gameworks looks up what GPU you are running and then accordingly runs code intentionally to degrade performance, then they are not the same.

[leadeater]

1 hour ago, mr moose said:

Yes they do, when they intentionally compromise the performance of AMD cpu's as they did the FTC ruling made it quite clear that they had to remedy the situation.  In fact the ruling went as far to make Intel pay for the costs of any developer who used their compiler to recompile their software and make the necessary patches. 

They were not required to change anything to do with the compiler, that's what I was pointing out. They were required to remedy damages due to their actions yes.

 

1 hour ago, mr moose said:

These are pretty damning directives for merely optimizing for your own product.   Why is that? because Intel did not merely optimize for their own product.  This is why the Intel example is nothing like gameworks. Remember that they are not liable for incidental performance drops in the competition, only artificial (intentional) drops.

While Nvidia pushes developers over to using GameWorks with a known performance impact on AMD and if they do not properly disclose it then yes it is similar.

 

I'll repeated it again, Intel was never punished for not utilizing SSE2 and SSE3 for AMD processors they were punished for lying about it and not disclosing they were not when asked. The reparations awarded where to amend those actions not because the compiler did not utilize SSE2 or SSE3 for AMD processors but because people were using it thinking it did.

 

1 hour ago, mr moose said:

Not like Intel at all.  They are two very different situations.  Unless you have evidence that gameworks looks up what GPU you are running and then accordingly runs code intentionally to degrade performance, then they are not the same.

No that is not at all a requirement of anything, just because Nvidia does not actively look for vendor ID does not mean they are unable to target or know a technique or behavior will more adversely effect AMD hardware and not Nvidia.

 

1 hour ago, mr moose said:

Exactly, you have to prove there is intentional gain from it, in other words you have to prove Nvidia are doing this intentionally to hamper AMD.  Until then they are not obligated in any way to fix the issue for AMD.

While Nvidia push GameWorks to more developers which influences game performance, reviews and market perception of Nvidia hardware products they are to me at that point forward intentionally gaining from either the bug/defect or design behavior of their tool giving them an undue market advantage. That is it for me, while they do that they are intentionally gaining from it, end of for me.

 

You're welcome to not agree with my stance on that but that is it for me, they know what it does and they know they gain from it. It originally being unintentional or not stops mattering when you intentionally gain from it. Unless the next wave of GameWorks games and the tool stops exhibiting those performance impacts on AMD products of course, Nvidia may indeed be working on fixing the issue, I have no idea.

 

Edit:

Plus it's probably important to point out because it seems to not have been covered yet the use of GameWorks effects the games more than just the tag like effects like HairWorks. GameWorks splatters all areas of the game with excessive amounts of Tessellation and is the reason AMD had to implement driver level hack/override to counter that.

[mr moose]

6 hours ago, leadeater said:

They were not required to change anything to do with the compiler, that's what I was pointing out. They were required to remedy damages due to their actions yes.

 

 

Did you read the quotes I provided? the FTC said from the FTC requirements:

 

Quote

Intel provide them, at no additional charge, a substitute compiler that is not a Defective Compiler;

They had to change the compiler.  They were ordered to provide everyone who brought their compiler a new compiler that was not "defective". 

6 hours ago, leadeater said:

While Nvidia pushes developers over to using GameWorks with a known performance impact on AMD and if they do not properly disclose it then yes it is similar.

It doesn't mater if the impact is known or not, so long as it is not intentional.   I don't know why this bit is so hard to understand. Intel were punished because they did it intentionally, We have no evidence nvidia are doing it intentionally. And intention, as you have already pointed out, is key.

 

6 hours ago, leadeater said:

I'll repeated it again, Intel was never punished for not utilizing SSE2 and SSE3 for AMD processors they were punished for lying about it and not disclosing they were not when asked. The reparations awarded where to amend those actions not because the compiler did not utilize SSE2 or SSE3 for AMD processors but because people were using it thinking it did.

No, they were found guilty and punished for gimping AMD CPUs in their compiler.  Why are you arguing this didn't happen?  Here is a quote from the FTC complaint:

Quote

B. Intel Redesigned its Software to Slow Software Performance on Non-Intel CPUs.
56.
Intel sought to undercut the performance advantage of non-Intel x86 CPUs relative to Intel x86 CPUs when it redesigned and distributed software products, such as compilers and libraries.
57.
A compiler is software that translates the “source code,” programs written by programmers or software developers in high-level computer languages such as C++ or Fortran into “object code” (0’s and 1’s), the language understood by CPUs. Libraries are collections of code for performing certain functions that can be referred to by software programmers rather than rewriting the code each time the functions are performed.
58.
For example, in response to AMD introduction of its Opteron CPU for servers in 2003, Intel became concerned about the competitive threat posed by Opteron processors. Intel then designed its compiler and libraries in or about 2003 to generate software that runs slower on non-Intel x86 CPUs, such as Opteron. This decrease in the efficiency

 

So the rest of your post aside for now,  Intel intentionally gimped AMD cpus in their compiler and that's what they got in trouble for.  This whole idea that it was merely not letting people now its effects on AMD CPU's is moot. I even quoted the relevant passage that explains why that is irrelevant earlier.

 

http://www.ftc.gov/os/adjpro/d9341/091216intelcmpt.pdf

 

BTW, you keep saying "nvidia pushes" like the game developers have no choice,  they do, it's just why would you know back an offer were someone does all the hard yards for you  virtually for free.

 

[leadeater]

54 minutes ago, mr moose said:

They had to change the compiler.  They were ordered to provide everyone who brought their compiler a new compiler that was not "defective". 

The compiler was already fixed way before that time, "We order you to do something you have already done". Wait...

 

54 minutes ago, mr moose said:

It doesn't mater if the impact is known or not, so long as it is not intentional.   I don't know why this bit is so hard to understand. Intel were punished because they did it intentionally, We have no evidence nvidia are doing it intentionally. And intention, as you have already pointed out, is key.

Yes no kidding, it's AFTER knowing is the issue. You cannot un know what your tool does, you cannot un know it's going to trash your competitor, you can choose to inform developers of this fact and why it's the case or you can choose not to.

 

You can tell people that are using the Intel compiler that it does not use SSE2 and SSE3 for AMD CPUs or you can not tell them, you can even not tell them after directly asked. You can use your compiler to show how your product is better knowing full well it's a misrepresentation.

 

You can partner with game developers and do more than assist in the later stages of developer optimizing that game for your product, you can provide tools that effect the development fundamentally and provide advice on technological techniques while also knowing it'll have a massive performance penalty on your competitor.

 

Or as put this way:

21 hours ago, Humbug said:

It's in the best interest of a game developer to make sure that their game runs as well as possible while looking as good as possible on the widest possible range of hardware. This is in the best interest of the customers as well as the developers. So the tools you use and the middleware you choose to purchase help you achieve that goal. This in turn helps you to sell the maximum number of copies of the game, win-win situation.

 

Thing is when you flip that equation on it's head; now Nvidia is rewarding you for using their middleware. Now it's a management decision because you have a financial incentive to use said black box middleware. That's what happens with gameworks, compromising the previous objective of making the correct technical decisions which lead to the best possible performance:graphics ratio for end customers.

 

 Nvidia intertwining themselves so tightly in to the creation of the game that it effects it at a base level is a problem, what you end up making is an Nvidia game not a game that runs well on Nvidia hardware.

 

Edit:

 

54 minutes ago, mr moose said:

No, they were found guilty and punished for gimping AMD CPUs in their compiler.  Why are you arguing this didn't happen?  Here is a quote from the FTC complaint

Completely yes that is what they were doing, you're missing the why they were punished. They did all those things and LIED misleading people as to what it was doing. You can do all those things and not lie and not mislead and the FTC will not stop you, punish you or make you amend anything.

 

People compiling for AMD or comparing the performance of Intel vs AMD would not have used the Intel compiler otherwise, there were other compiler options.

 

All you did there was post a summary of facts, part of a larger picture.

[mr moose]

3 minutes ago, leadeater said:

The compiler was already fixed way before that time, "We order you to do something you have already done". Wait...

 

What are you talking about?  "way before that time"?

3 minutes ago, leadeater said:

Yes no kidding, it's AFTER knowing is the issue. You cannot un know what your tool does, you cannot un know it's going to trash your competitor, you can choose to inform developers of this fact and why it's the case or you can choose not to.

 

You can tell people that are using the Intel compiler that it does not use SSE2 and SSE3 for AMD CPUs or you can not tell them, you can even not tell them after directly asked. You can use your compiler to show how your product is better knowing full well it's a misrepresentation.

The FTC complaint wasn't that they didn't tell anyone after they knew, it was that they intentionally gimped it in the first place.

3 minutes ago, leadeater said:

You can partner with game developers and do more than assist in the later stages of developer optimizing that game for your product, you can provide tools that effect the development fundamentally and provide advice on technological techniques while also knowing it'll have a massive performance penalty on your competitor.

 

Or as put this way:

 

 Nvidia intertwining themselves so tightly in to the creation of the game that it effects it at a base level is a problem, what you end up making is an Nvidia game not a game that runs well on Nvidia hardware.

From there on in it is personal perspective as I pointed out right back at the start, you can not like it or like it or hate it or love it.  But it remains simply a business practice to get their branding on games and make their products look better.  Everything else is speculation.

[leadeater]

14 minutes ago, mr moose said:

From there on in it is personal perspective as I pointed out right back at the start, you can not like it or like it or hate it or love it.  But it remains simply a business practice to get their branding on games and make their products look better.  Everything else is speculation.

It is not speculation as to what the tool does, we know what it does. I can speculate the developers are not told before hand as to the specifics of why it has that effect and that it's going to have that performance impact on AMD products. This isn't like PhysX where on and off that in the off state it has truly zero impact on the game.

 

And no it's not simply a business decision to get their branding on games or look better, they can and have been perfectly capable of such things without pushing their way in to the development cycle of the games themselves at such a level. That's not a requirement to get the Nvidia logo on a game, heck you can just hand over straight money for that and nothing else.

 

14 minutes ago, mr moose said:

The FTC complaint wasn't that they didn't tell anyone after they knew, it was that they intentionally gimped it in the first place.

Which is something Intel is actually allowed to do so long as they state so. So long as they were clear to customers that "We have updated the Intel compiler so it does not utilize SSE2 and SSE3 on AMD processors" they would not have been punished. Intel compiler was not the only compiler available for Intel processors or AMD processors and you can be sure given that information it would not have been used for AMD processors.

 

See above edit. I'm not going to cover this again. The Intel story is all around misrepresentation and lying, something I obviously believe Nvidia may be doing about their tool. That is where I think they could be challenged on legally not that it results in games that run like garbage on AMD GPUs but that they are not clear that is what will happen using their tool, something that otherwise would not be the case.

 

Those effects flow on to product reviews and customer purchasing decisions, ones influenced by their tool.

[mr moose]

15 minutes ago, leadeater said:

It is not speculation as to what the tool does, we know what it does. I can speculate the developers are not told before hand as to the specifics of why it has that effect and that it's going to have that performance impact on AMD products. This isn't like PhysX where on and off that in the off state it has truly zero impact on the game.

 

And no it's not simply a business decision to get their branding on games or look better, they can and have been perfectly capable of such things without pushing there way in to the development cycle of the games themselves at such a level. That's not a requirement to get the Nvidia logo on a game, heck you can just hand over straight money for that and nothing else.

 

And all of that is speculation, you might know the outcome, but you don't know the motivation.

Quote

Which is something Intel is actually allowed to do so long as they state so. So long as they were clear to customers that "We have updated the Intel compiler so it does not utilize SSE2 and SSE3 on AMD processors" they would not have been punished. Intel compiler was not the only compiler available for Intel processors or AMD processors and you can be sure given that information it would not have been used for AMD processors.

 

See above edit. I'm not going to cover this again. The Intel story is all around misrepresentation and lying, something I obviously believe Nvidia may be doing about their tool. That is where I think they could be challenged on legally not that it results in games that run like garbage on AMD GPUs but that they are not clear that is what will happen using their tool, something that otherwise would not be the case. 

 

It's not about failing to tell people though, they literally got done because:

Quote

Intel Redesigned its Software to Slow Software Performance on Non-Intel CPUs.

That is the actual compliant from the FTC.

 

Being ordered to tell everyone about it is the result of that intentional gimping, it is not the reason they got in trouble in the first place.

And so it goes with Nvidia, unless they are intentionally gimping AMD, they are under no obligation to "fix" and issues that arise from it.

 

EDIT: and you didn't answer my question, way before what time?  perhaps you thought the complaint and the agreement we have been talking about where from different investigations?

[leadeater]

6 minutes ago, mr moose said:

And so it goes with Nvidia, unless they are intentionally gimping AMD, they are under no obligation to "fix" and issues that arise from it.

Knowing that your product does that, actively pursuing game developers to use the tool without disclosing that information is that. You can mistakenly make the tool do that from the start but you can also leverage that effect to your gain and hide that fact.

 

 

6 minutes ago, mr moose said:

It's not about failing to tell people though, they literally got done because:

 

Quote

In addition, allegedly, Intel secretly redesigned key software, known as a compiler, in a way that deliberately stunted the performance of competitors’ CPU chips. Intel told its customers and the public that software performed better on Intel CPUs than on competitors’ CPUs, but the company deceived them by failing to disclose that these differences were due largely or entirely to Intel’s compiler design.

https://www.ftc.gov/news-events/press-releases/2009/12/ftc-challenges-intels-dominance-worldwide-microprocessor-markets

 

And in the actual ruling, section Nature of the Case:

Quote

10. Fifth, Intel engaged in deceptive acts and practices that misled consumers and the public. For example, Intel failed to disclose material information about the effects of its redesigned compiler on the performance of non-Intel CPUs. Intel expressly or by implication falsely misrepresented that industry benchmarks reflected the performance of its CPUs relative to its competitors’ products. Intel also pressured independent software vendors (“ISVs”) to label their products as compatible with Intel and not to similarly label with competitor’s products’ names or logos, even though these competitor microprocessor products were compatible.

 

Section on the Compiler:

Quote

59. To the public, OEMs, ISVs, and benchmarking organizations, the slower performance of nonIntel CPUs on Intel-compiled software applications appeared to be caused by the non-Intel CPUs rather than the Intel software. Intel failed to disclose the effects of the changes it made to its software in or about 2003 and later to its customers or the public. Intel also disseminated false or misleading documentation about its compiler and libraries. Intel represented to ISVs, OEMs, benchmarking organizations, and the public that programs inherently performed better on Intel CPUs than on competing CPUs. In truth and in fact, many differences were due largely or entirely to the Intel software. Intel’s misleading or false statements and omissions about the performance of its software were material to ISVs, OEMs, benchmarking organizations, and the public in their purchase or use of CPUs. Therefore, Intel’s representations that programs inherently performed better on Intel CPUs than on competing CPUs were, and are, false or misleading. Intel’s failure to disclose that the differences were due largely to the Intel software, in light of the representations made, was, and is, a deceptive practice. Moreover, those misrepresentations and omissions were likely to harm the reputation of other x86 CPUs companies, and harmed competition.

https://www.ftc.gov/sites/default/files/documents/cases/091216intelcmpt.pdf