AMD announces GPUOpen - Open Sourced Gaming Development

[GoogalyMoogaly]

Following a sentace claim of superior nvidia support in ogl.

Don't Nvidia do better in OpenGL and have lots of custom extensions?
Even if they don't I don't see how he was suggesting they should get preferential treatment.

 

[humbug]

Don't Nvidia do better in OpenGL and have lots of custom extensions?
Even if they don't I don't see how he was suggesting they should get preferential treatment.

Nivida only extensions are not open source. They have nothing to do with opengl, they are proprietary extensions attached to opengl, that's their level of support for open source, as long as it can be used to give them an advantage.
Give me one example of a significant technology by nvidia and willingly used in the mainstream by others that is open source and ask me to do the same for other hardware vendors.

Think perhaps the very reason opengl is nothing more than an abstract API is because its used as a technology vessel by nvidia.
No one has any interest in the crap that's left over after nvidia are done butchering it.

 

[GoogalyMoogaly]

Nivida only extensions are not open source. They have nothing to do with opengl, they are proprietary extensions attached to opengl, that's their level of support for open source, as long as it can be used to give them an advantage.
Give me one example of a significant technology by nvidia and willingly used in the mainstream by others that is open source and ask me to do the same for other hardware vendors.

Think perhaps the very reason opengl is nothing more than an abstract API is because its used as a technology vessel by nvidia.
No one has any interest in the crap that's left over after nvidia are done butchering it.

Didn't explain how you thought he said Nvida should be given preferential treatment. That was just a rant about nvidia/opengl.

 

[tommybhoy]

Out of all the dross posted in this sub googaly, are you making a fuss over bugs interpretation of one of D.P's posts, is everyone meant to give him the benefit of the doubt, after you cut through the heavy negative chaff, there's nothing left.

I know you like to give people that don't have an issue with AMD's openness stick but come on mate, give us a bit of balance and go after some of the bs going the other way, ask them some of the hard questions and mix it up some.

 

[Insanties_birth]

You do realize most industries are closed, and most software is closed source. You are almost certainly writing the above post form a closed source operating system, quite possibly form a closed source browser, using closed source hardware, supported by closed source drivers, compiled on a closed source compiler.

Nvidia is a big supporter of open source, their OpenGL support has been far superior to AMD's. The president of the Khronous group that oversees the development of open source royalty free software like openGL and Vulkan APIs is none other than Nvidia vice-president Neil Trevett.

It doesn't really matter if most are closed, it's still better for the consumer that it be open. You might like the idea of companies nailing each tech behind a paywall like gsynch and getting overcharged but most informed consumers would prefer they have more open choice. It's not even just a benefit to us, it's a benefit to the devs so why would you argue against it unless you really get your kicks off of companies relying on closed tech that forces you to into a worse off situation? So the stuff I use is closed source in some regards, does that mean I wouldn't benefit from it being open or (if it was realistic and had more options in the world with comparably performing open source systems) that I would get open source if it was available?

I don't feel Nvidia is that big a supporter of open source really. They might have the odd project going and I'm not sure whether the vulkan project is being headed more as Neil's own back rather than Nvidia directly involving him (as I'm not sure on that point admittedly) but it seems they went against freesynch for cash, they never supported tressfx and if opengl was really to be promoted rather than sidelined in favour of gsynch it might show there more clear focus. Having the odd project be open while still more heavily pushing proprietary expensive tech against the consumer (gsynch), pushing less than open software that damages performance on other systems whenever the game is sponsored by them (gameworks) to the point where a lot refer to it as gimpworks etc. kind of highlights where the real push is. Nvidia could have easily tried to help improve freesynch for example but chose to go against this (which is arguably more an open standard than open source but toes the same line of being open for all). As stated before though, I like open source and open standard, if you prefer to get overcharged for gsynch, worry if buying another vendors card means you'll get shocking performance from Nvidia features when it's in there games and watch as AMD pushes the open GPU platform while Nvidia lag behind on this sort of stuff then that's your choice. Just console yourself that when it suits them they have the opengl thing that is pushed so little that there gameworks closed source stuff eclipses it. I'm not really here to argue anyway, it's my opinion that open source is good, I feel it's stupid to argue otherwise and you can argue how open other companies are but that's completely irrelevant to the fact AMD are open and Nvidia are lagging in that regard so I can praise there efforts. Nothing wrong with being happy at a positive change in the industry so why complain?

Just my ramblings and i don't care if you respond to it all but the simple question is, if things are becoming more open and that's better for the consumer then why is there a group in here just arguing and frowning at the idea of things potentially becoming better for the consumer? Is it really that tied into the old AMD vs Nvidia battle that even positive news is a pain to you guys?

 

[humbug]

Whats the point of the Khronous Group?
Seriously, their biggest achievements are OpenGL and OpenCL, OpenGL is a failure by their own admittance and OpenCL is lagging behind CUDA, an Nvidia proprietary tech, Nvidia are on the Khronous BOD.

The only Khronous Group member who has gained anything out of the organization are Nvidia, if they are a Group apparently sporting Open Source technology then why is it only one of them benefit?

And why is it that one of their members built a technology entirely independently of the Khronous Group and then handed it over to bail them out, yes Mantle. They publicly admitted OpenGL was going nowhere, GL-Next was not working and that they couldn't have done it without AMD, It was AMD who at least atempted to work with the likes of Adobe ect.... to make OpenCL work, meanwhile Nvidia are still pushing CUDA against it.
What a shocking state to be in for a a group of Vendors apparently working together for the good of the collective and those of us using their hardware.

AMD are not perfect, but it does look like they are the only ones who try.
Its not just Mantle, they are capable of bringing together other vendors and setting up cooperative organisations on their own, the HSA foundation, GPUOpen is a collective of several independent entities.
AMD are the ones who worked with VESA to standardize ports making Adaptive Sync viable for everyone, AMD are the ones who in partnership with Hynix designed a new memory architecture in GDDR when DDR could no longer cut it, and now HBM, are Nvidia or anyone excluded from the technology? no.

Meanwhile Nvidia are rubbing the likes of Samsung up the wrong way, for what? Money and control, like megalomaniacs.

 

[GoogalyMoogaly]

Out of all the dross posted in this sub googaly, are you making a fuss over bugs interpretation of one of D.P's posts, is everyone meant to give him the benefit of the doubt, after you cut through the heavy negative chaff, there's nothing left.

I know you like to give people that don't have an issue with AMD's openness stick but come on mate, give us a bit of balance and go after some of the bs going the other way, ask them some of the hard questions and mix it up some.

Yeah, it went on longer than I expected. Sorry.
I was basically expecting Humbug to say he'd misread and that would be that.

I really don't care if something is open source or not. i use so much closed source software (Microsoft Windows, various games, etc.) that I can't really say it's a bad thing. I'm not saying there's no benefit to open-source software but I do feel it's been blown way out of proportion and is now just being used as something else to argue over. Closed source isn't inherently bad and open source isn't inherently good.

I like these libraries from the point of view that it gives people to use these effects that might otherwise not be included. If they're optional I see no harm in it even if they hurt performance. You can just turn them off and it'd be the same as if the effect wasn't included at all.

I think it was Humbug that said that these libraries are unnecessary (or at least it was when it was an Nvidia thing, wonder if they're still pointless now AMD have them). It stuff developers can do for themselves and from what Humbug said it can be done pretty quickly. If these effects can be done natively by the engine, that has to be better than a 3rd party library surely?

As for performance hits differing across vendors, I think that's to be expected to some degree. the cards are all different and one vendor's may do some things better than the other's. Should they be forced to implement an effect in a way that works equally well on both or should they be allowed to implement it in the best way possible? (I'm not saying either GameWorks or GPUOpen does this, but I think they should be allowed to.)

This meaning that some games may run better on one vendor while other games run better on another, doesn't really bother me. Games do that anyway. I don't want games to be delayed while developers tweak performance to bring the faster card in line the the equivalent card from the other vendor. I don't expect performance to always be equal.
For me, I've got a rig with AMD cards in and a rig with Nvidia cards in, so I'll play it on whichever works best for me.

 

[tommybhoy]

Having both vendors gpu's is a fortunate position to be in, perhaps if you only had AMD or Kepler and watching it crumble at times with GW's, you might have had a different pov idk.

 

[humbug]

Yeah, it went on longer than I expected. Sorry.
I was basically expecting Humbug to say he'd misread and that would be that.

I really don't care if something is open source or not. i use so much closed source software (Microsoft Windows, various games, etc.) that I can't really say it's a bad thing. I'm not saying there's no benefit to open-source software but I do feel it's been blown way out of proportion and is now just being used as something else to argue over. Closed source isn't inherently bad and open source isn't inherently good.

I like these libraries from the point of view that it gives people to use these effects that might otherwise not be included. If they're optional I see no harm in it even if they hurt performance. You can just turn them off and it'd be the same as if the effect wasn't included at all.

I think it was Humbug that said that these libraries are unnecessary (or at least it was when it was an Nvidia thing, wonder if they're still pointless now AMD have them). It stuff developers can do for themselves and from what Humbug said it can be done pretty quickly. If these effects can be done natively by the engine, that has to be better than a 3rd party library surely?

As for performance hits differing across vendors, I think that's to be expected to some degree. the cards are all different and one vendor's may do some things better than the other's. Should they be forced to implement an effect in a way that works equally well on both or should they be allowed to implement it in the best way possible? (I'm not saying either GameWorks or GPUOpen does this, but I think they should be allowed to.)

This meaning that some games may run better on one vendor while other games run better on another, doesn't really bother me. Games do that anyway. I don't want games to be delayed while developers tweak performance to bring the faster card in line the the equivalent card from the other vendor. I don't expect performance to always be equal.
For me, I've got a rig with AMD cards in and a rig with Nvidia cards in, so I'll play it on whichever works best for me.

I said Gameworks and TressFX ect... was not bringing anything new, its whats already in existence as Open Source packaged up, branded and flogged.

There is good and bad about it. bad because it dumbs developers down, can be abused by vendors for their own agenda and actually make the experience of gamer's worse rather than better.

Good because its faster and cheaper for developers, they don't need the talent, more widely used.

Its a bit like the difference between an engineer and a fitter, an engineer understands the workings of his project, a fitter knows how to cut and paste..

 

[GoogalyMoogaly]

I said Gameworks and TressFX ect... was not bringing anything new, its whats already in existence as Open Source packaged up, branded and flogged.

There is good and bad about it. bad because it dumbs developers down, can be abused by vendors for their own agenda and actually make the experience of gamer's worse rather than better.

Good because its faster and cheaper for developers, they don't need the talent, more widely used.

Its a bit like the difference between an engineer and a fitter, an engineer understands the workings of his project, a fitter knows how to cut and paste..

I'm sure you said they should just do the effects themselves, like you do in your demo. No existing physics engines or anything required.

 

[humbug]

I'm sure you said they should just do the effects themselves, like you do in your demo. No existing physics engines or anything required.

No.
My demo is based on bullet, it can't be done without a physics engine.

 

[GoogalyMoogaly]

No.
My demo is based on bullet, it can't be done without a physics engine.

Oh yeah, that's right, I remember those arguments now.
Other than that I'm sure you said they should just do the effects themselves.

 

[drunkenmaster]

You do realize most industries are closed, and most software is closed source. You are almost certainly writing the above post form a closed source operating system, quite possibly form a closed source browser, using closed source hardware, supported by closed source drivers, compiled on a closed source compiler.

Nvidia is a big supporter of open source, their OpenGL support has been far superior to AMD's. The president of the Khronous group that oversees the development of open source royalty free software like openGL and Vulkan APIs is none other than Nvidia vice-president Neil Trevett.

This is entirely misleading. Using an open source api doesn't make Nvidia a bit supporter of open source. Having an open source driver would be and Nvidia's openGL/Linux/general open source openGL support is far far BEHIND Nvidia.

AMD provides an open source Linux driver that isn't the best, but is extremely well received by the linux community. Nvidia, I can't remember if they provide an open source driver at all, I think they don't or they do but it's basically awful and not much behind say a standard VGA driver.

Supporting openGL is in no way an indication of a company being big on open source support and suggesting so is complete nonsense. Dozens, hundreds, maybe thousands of companies provide drivers compatible with open source. Not many of those companies provide open source drivers, AMD afaik provides one of the most complex open source drivers out there.


Anyway, something people consistently forget, or ignore, is that an open source piece of code can be checked up and down for slow or sabotaging code. Intel's compiler shouldn't, but actively does turn off optimisations for AMD hardware, not enough to make it too obvious but enough to make them ****'s about it. Closed code that Nvidia provides only black box access to that any devs appear to use(there is NOW a source code option, that from appearance no devs have taken up and could well be priced in such a way that no dev can afford it so is left with only the black box version) could both have lack of optimisation for AMD, flat out sabotaging code and is still hard to optimise for.

Now lack of optimisation before someone jumps all over it, is not saying optimise for AMD hardware, but you could simply chose to use known slower method when it detects AMD hardware and doesn't enable the faster path using basic dx11/engine calls or methods. In CPU terms you might say Intel compiler doesn't enable SSE4 optimisations when it detects AMD and uses a slower SSE2 path. It wouldn't take Intel optimising the SSE4 call to make it faster on AMD, that isn't AMD specific flag, just using a slower way on purpose.

Open source code, you KNOW the dev has COMPLETE access to all code, you know that Nvidia can have complete access to all the code. Does this make certain there is no sabotaging..... well in theory no, but it would be obviously found by devs and Nvidia and they'd get called out on it. Open source code you can basically be sure there is no shenanigans going on... black box stuff that happens to cause worse performance hits on the other companies cards, lets just say anyone that says they are sure no shenanigans are going on are kidding themselves or have an agenda.

 

[GoogalyMoogaly]

This is entirely misleading. Using an open source api doesn't make Nvidia a bit supporter of open source. Having an open source driver would be and Nvidia's openGL/Linux/general open source openGL support is far far BEHIND Nvidia.

AMD provides an open source Linux driver that isn't the best, but is extremely well received by the linux community. Nvidia, I can't remember if they provide an open source driver at all, I think they don't or they do but it's basically awful and not much behind say a standard VGA driver.

Supporting openGL is in no way an indication of a company being big on open source support and suggesting so is complete nonsense. Dozens, hundreds, maybe thousands of companies provide drivers compatible with open source. Not many of those companies provide open source drivers, AMD afaik provides one of the most complex open source drivers out there.


Anyway, something people consistently forget, or ignore, is that an open source piece of code can be checked up and down for slow or sabotaging code. Intel's compiler shouldn't, but actively does turn off optimisations for AMD hardware, not enough to make it too obvious but enough to make them ****'s about it. Closed code that Nvidia provides only black box access to that any devs appear to use(there is NOW a source code option, that from appearance no devs have taken up and could well be priced in such a way that no dev can afford it so is left with only the black box version) could both have lack of optimisation for AMD, flat out sabotaging code and is still hard to optimise for.

Now lack of optimisation before someone jumps all over it, is not saying optimise for AMD hardware, but you could simply chose to use known slower method when it detects AMD hardware and doesn't enable the faster path using basic dx11/engine calls or methods. In CPU terms you might say Intel compiler doesn't enable SSE4 optimisations when it detects AMD and uses a slower SSE2 path. It wouldn't take Intel optimising the SSE4 call to make it faster on AMD, that isn't AMD specific flag, just using a slower way on purpose.

Open source code, you KNOW the dev has COMPLETE access to all code, you know that Nvidia can have complete access to all the code. Does this make certain there is no sabotaging..... well in theory no, but it would be obviously found by devs and Nvidia and they'd get called out on it. Open source code you can basically be sure there is no shenanigans going on... black box stuff that happens to cause worse performance hits on the other companies cards, lets just say anyone that says they are sure no shenanigans are going on are kidding themselves or have an agenda.

As far as sabotaging in GPUOpen, all you can tell is that there's no (obvious) sabotaging in the code in GitHub, you don't know what code is used in the game (I'm guessing they don't have to include the source for the implementation used in a game). So companies with agreements with Nvidia/AMD could in theory add code to only optimise one side or to sabotage the other side.
It seems ridiculous but so does having 2 routes in GameWorks just to screw AMD users.

 

[drunkenmaster]

It seems ridiculous, even with multiple gameworks games with absurd drops in performance on both AMD and older Nvidia architectures.... ridiculous?

 

[GoogalyMoogaly]

It seems ridiculous, even with multiple gameworks games with absurd drops in performance on both AMD and older Nvidia architectures.... ridiculous?

That's my point, it seems ridiculous but lots of people claim there are issues. So while it seems unlikely to us...
Same could happen with this, AMD could sponsor a game and in that agreement there could be a condition that they have to do something funky with the GPUOpen code to make it run worse on Nvidia or better on AMD or perhaps run a version that hasn't been released to the public which has malicious code.
No reason Nvidia couldn't do that too, except I don't see them getting involved with GPUOpen.

I don't see why Nvidia couldn't build a new version of GameWorks around GPUOpen. They could add value by actively developing it rather than hoping the community will do it. I'm sure they could find a way to make it a closed source product by using it within a product rather than just using it itself (I mean games that use it don't have to be open source so there must be ways around having to release source code). They could keep it up to date with the main branch (assuming that changes) and keep applying their changes over the top.

I'd have thought it would be easier to add value to this than to Adaptive Sync.

 

[tommybhoy]

You/other black box proponents keep talking about Nv screwing over AMD/

On current market share, it's of a far higher benefit for Nv to screw over their own customers for coin.

Anything souring gpuopen is probably an open and shut case, none of this show us proof pish(shown it only to ignore it) as both AMD/Nv have access to the code.

At worst, I guess the release day patch code change would be about as bad as it gets-requiring a quick fix for whoever doesn't have their name on the box.

 

[GoogalyMoogaly]

You/other black box proponents keep talking about Nv screwing over AMD/

On current market share, it's of a far higher benefit for Nv to screw over their own customers for coin.

Anything souring gpuopen is probably an open and shut case, none of this show us proof pish(shown it only to ignore it) as both AMD/Nv have access to the code.

At worst, I guess the release day patch code change would be about as bad as it gets-requiring a quick fix for whoever doesn't have their name on the box.

I've not actually read the licensing around GPUOpen but does it state that the code used in the game has to be freely available? Or can, for example, Crystal Dynamics keep the changes they made for PureHair a secret so only they can use it? Cuz then you can't see proof. Looking at what's in GitHub won't help if they've modified the code and are allowed to keep the modified code private.
People could get screwed over just as badly by altered GPUOpen code as they can by GameWorks code if there's no requirement to make the altered code available to everyone. How do you prove that what you've released to everyone is what's used in the game? Do you have to make the entire game's source code freely available so people can check?

 

[tommybhoy]

AMD/Nv will produce proof of dodgy tricks, that's the whole point of it, but you know that really and just doing what you do, anything AMD-negative, negative, negative.

I don't have a problem with that as it's your prerogative, but alas, I don't think we'll ever get a decent convo going, as I don't think Iv'e ever seen anything AMD positive come from you in all my time here due to you disdain of pro AMD posters(considering it goes both ways), it's a pity really as I think you could bring way, way more to the table in here.

 

[CAT-THE-FIFTH]

Awesome AMD,this will fail entirely if your PR just CBA highlighting any use of your technologies. CD recently used TressFX to produced a modified version called Purehair for their games.

But instead of AMD PR saying anything before the game or at launch they were silent. Not a noise.

But,Nvidia devotes space on their website talking about the tech:

Yet,look at Nvidia who posted a whole piece on Purehair yesterday including a performance guide on it:

http://www.geforce.com/whats-new/gu...rmance-guide#rise-of-the-tomb-raider-purehair

PureHair is Crystal Dynamics and Square Enix's hair rendering technology, which like our own HairWorks technology adds tens of thousands of hair strands to a character model. These hairs act realistically, swaying and moving in concert with character movement, and can be affected by water, wind and snow, and are lit and shaded in real-time by the scene.

In Rise of the Tomb Raider, up to 30,000 strands of hair are applied solely to Lara, with large groups of hairs controlled by master strands that dictate their movement and properties, preventing each individual hair strand from acting independently, and keeping physics calculation costs in check.

Conveniently make it sound like AMD did not contribute anything to it. So average gamer probably thinks CD made their own effects,and since it is on the Nvidia site they probably helped too. It also makes TressFX seem like some failure since there is no mention of AMD in the second came by the company.

Who would blame them?? AMD had potential to shout from the roofs about this as it was a vindication of their approach to make things more open so devs could use and optimise parts for their own games freely.

But AMD PR is so unexcited about it,that not even their site has mention of it:

http://www.amd.com/en-us/innovations/software-technologies/technologies-gaming/tressfx
Then a day later we have a one line comment from AMD about it:

https://twitter.com/Thracks/status/693123852839182336

Seriously,for every little bit of Nvidia assistance in a game,its posted on their website and usually get some degree of news coverage.

It not only shows they are being "proactive" in them pushing "new" tech to games but it also keeps reminding people of them and what they are doing.

Yet,AMD who should have been shouting from the roofs about this,CBA even to market it. It means plenty of people thinking in the latest TR game, AMD has no involvement in the game.