AMD Halts Optimizations for Mantle API

[TheRealDeal]

That was way before Mantle was dumped in favour of DX12. I'd be *very* surprised if they didn't switch their focus to DX12 now...

Yea i was just highlighting the fact that it was not just AMD that said Star Citizen would use Mantle. The man himself said it and highlighted why.

It would be foolish now to use anything other than dx12 or Vulkan with the game so far from release.

 

[GoogalyMoogaly]

Really, it sets an unfortunately precedent? In that AMD bring forth an API and all work towards enabling your engine for it carries over to an industry standard you know will be supported for years to come?

This establishes that working on something AMD is working on will be turned sensibly into a industry standard. If I was a dev working on a feature like API support I would know something Nvidia do that is proprietary will never make it into AMD hardware or gain industry wide support, if AMD make an API or a new good hardware feature you can be significantly more certain it will gain industry wide support as an open standard that ALL users can benefit from in the future and thus be far more confident in spending the time to support it.

All the work done towards Mantle now carries forward to ALL hardware that supports Vulkan... how is that bad for devs? All devs could possibly learn from this situation is that not a single second of working on Mantle will be wasted for future games. It gave them a huge headstart in support for Vulkan and DX12.

Are you referring to things like the industry standard that is TressFX?
Or TrueAudio? Do the Fury cards have TrueAudio?

Yes, In case he still doesn't get it...

The Fact that DICE will be moving their Frostbite engine to Vulkan means Mantle will live on.

Why would DICE need to move their Frostbite engine to Vulkan if Mantle and Vulkan are exactly the same thing? Surely it already supports Vulkan, but it's just called Mantle?
Why use the word transition if they just mean change the name on the documentation?

Do your research.. Mantle is very much Vulkan.. They tuck the best parts from mantle api into the Vulkan API.

"AMD then further went on to reveal that Vulkan had actually absorbed all the best features, hinting at Vulkan being the true progress after Mantle API"
“The cross-vendor Khronos Group has chosen the best and brightest parts of Mantle to serve as the foundation for Vulkan,” said Robert Hallock, the head of global technical marketing at AMD. “Vulkan paves the way for a renaissance in cross-platform and cross-vendor PC games with exceptional performance, image quality and features,” stressed Mr. Hallock"

They is a very good reason why khronos thank amd a lot for the api.. It helped them build Vulkan.

Taking parts from and basing something on is quite different from being the same thing. I don't think there's any doubt that AMD deserve credit for their contribution, but I don't think Vulkan IS Mantle.
Just because a game uses CryEngine doesn't mean it IS Crysis.

 

[Phixsator]

PSML, Mantle is absolutely not in Volkan or DX12 in a pure form. Amd contributed code and ideas, just like Intel, Nvidia, Apple, SGI, Imagination , Valve, and others.

And your second point contradicts your first. If Vulkan/DX12 really is Mantle then there would be absolutely no need for Liquid VR to use Mantle, it would just use Vulkan and be done.

The fact is Vulkan is not Mantle, it exist long before Mantle came to exist and is a renamed OpenGL Next, which has been on the works pretty much since OGL 3.2. Just look at home many Nvidia extensions existed in OpenGL and how many of them were incorporated into the core OpenGL profile to realize how much Nvidia has contributed to OpenGL and GlNext/Vulkan.

Im sorry friend but i dont believe you are not correct on this one. I think you are mergin to things together. Take a look at this Khronos video https://youtu.be/qKbtrVEhaw8?t=10m37s . AMD gave Mantle to khronos as a foundation and a starting point for Vulkan. The guy in the video says so himself.

 

[CAT-THE-FIFTH]

I believe Nvidia were on record talking about how important Mantle was to Vulkan,and Nvidia would not just idly say that.

Plus,interestingly enough AMD kept to its promise of opening up Mantle to other companies,which basically their contributions to Vulkan have done.

 

[Phixsator]

I believe Nvidia were on record talking about how important Mantle was to Vulkan,and Nvidia would not just idly say that.

Plus,interestingly enough AMD kept to its promise of opening up Mantle to other companies,which basically their contributions to Vulkan have done.

Indeed they did i just don't think people had imagined for AMD to go this route with it. So many nay sayers to Mantle yet they all miss the freaking point and what potential good that API has done in its very short lifetime.

I think its perfectly fine to come up with a working concept and put that into a limited solution to push for a better experience and then say "we dont have the resources to continue the development but here take it for free". It worked, it got people interested and now we have Vulkan thanks to it(as in it delivered a critically important part to get Vulkan off the ground) . If it had been Apple or a similar company they would have been sitting on it and let it die rather than sharing it if they couldn't continue development themselves.

 

[D.P.]

Double post

 

[D.P.]

Im sorry friend but i dont believe you are not correct on this one. I think you are mergin to things together. Take a look at this Khronos video https://youtu.be/qKbtrVEhaw8?t=10m37s . AMD gave Mantle to khronos as a foundation and a starting point for Vulkan. The guy in the video says so himself.

Vulkan is just the new name for OpenGL next. Khronos wanted a new name that did not link to OpenGl because the 2 are completely different and not compatible, in fact openGl may still exist in a version 5.0.

AMD did contribute their Mantle code to the Khronos group who are indeed helping integrate some aspects of that, they may even be basing a lot of core concepts on the mantle design, but the project has existed for years and has had contributions from numerous parties since. Before AMD contributed Mantle a lot of the work was done on design exploration rather than coding any kind of software.



I you don't believe in Nvidia's continuation then consider this, Neil Trevet is the president of the Khronos Group as well as vice president of Nvidia.

 

[humbug]

Vulkan is just the new name for OpenGL next. Khronos wanted a new name that did not link to OpenGl because the 2 are completely different and not compatible, in fact openGl may still exist in a version 5.0.

AMD did contribute their Mantle code to the Khronos group who are indeed helping integrate some aspects of that, they may even be basing a lot of core concepts on the mantle design, but the project has existed for years and has had contributions from numerous parties since. Before AMD contributed Mantle a lot of the work was done on design exploration rather than coding any kind of software.



I you don't believe in Nvidia's continuation then consider this, Neil Trevet is the president of the Khronos Group as well as vice president of Nvidia.

Your putting 2 and 2 together thinking continued OGL development has simply borrowed a few bits from Mantle, in fact Khronos themselves have said that Vulkan is completely new and built on the core of Mantle.

OpenGL never got mainstream because Nvidia were involved, they use a lot of it for proprietary Driver support that isn't cross vendor, its why AMD who are also on the Khronos BOD never had much interest in OGL.

Now Khronos something that truly is Cross vendor it has a chance of making it mainstream.

Yes, In case he still doesn't get it...

The Fact that DICE will be moving their Frostbite engine to Vulkan means Mantle will live on.

Exciting stuff.

 

[SkeeterUK]

Wonder if AMD was payed to drop it from MS. Since mantle was pretty much dx12 MS had a interest to keep support more for dx than mantle. Maybe not true of course but i do wonder and its not like AMD couldnt do with the money.

 

[D.P.]

Your putting 2 and 2 together thinking continued OGL development has simply borrowed a few bits from Mantle, in fact Khronos themselves have said that Vulkan is completely new and built on the core of Mantle.

OpenGL never got mainstream because Nvidia were involved, they use a lot of it for proprietary Driver support that isn't cross vendor, its why AMD who are also on the Khronos BOD never had much interest in OGL.

Now Khronos something that truly is Cross vendor it has a chance of making it mainstream.

Exciting stuff.

You are simply wrong again,. GLnext was in development long before mantle came to being, but the develop,net was in design ideas,architectures, features, philosophy, paradigm, theory and minimal code. Opengl ARB is a slow beast, that is the main criticism of opengl. The idea of Glbext was soe thing akin to mantle, not an evolution of OGL, that is why they called it GLnext, and not gl5.0., and why they changed the name to Volkan and not OGLnext. Everyone wanted a fresh start, everyone wanted to drop legacy support. Everyone one a modern future looking API. Everyone wanted multi threading and efficiency at its core. Everyone wanted to reduce overhead. That was the whole philosophy of the project long befor AMD contributed any code. The design goals was something akin to mantle but cross hardware, cross vendor, compatible, open,
And developed and maintained by a consortium and not a single interested party, e.g. AMD. That matter part is a. Very critical point and is why neither intro nor nvidia or anything else like Samsung, Apple, Qualcomm etc. touched mantle with a barge pole.

Eventually AMD gave up on on Mantle and rather than waste all that development they gave it to the Kronos group to learn from and take advantage of. That is great, I really applauded AMD for doing that and I really doubt nvidia would have done the same so quickly. But Manlte was a hardware and vendor specific bets I eolentation of some of the ideas of GLnext that had already been widely developed upon. This made a useful code base to start building open, and nvidia admit to this the selves. But what Khronos clearly states that the code had radii ally developed and changed to bring in line with what glNext's vision was and allow it to be cross hardware and architecture agnostic, this is like the Greek ship myth, if you replace every Olank of wood, the sail, the mast, the keel, the rudder, etc, is it still the same ship, even although know instead of a sail it has a twin V12 10liter Diesel engine.

The Vice President of nvidia is the president of. The Khronos group. Don't under estimate how invested in Vulcan voids is, how much support they give, and how much they have contributed.

Your whip at OGL is patently worms. OGL never got mainstream windows use because of Microsoft not Supporting it in favor of its own DX as a competitive advantage against Apple and Linux. Nvidia's extension exist because of the nature of the library that permits vendors to extend the base functionality rather than waiting for a new version. You have to realize many of Nvida's extension were simply rename to become part of the next core, they had a huge influence in adding functionality. AMD just never had the same interet because they don't have the same profesional graphics market any more. Jus look at the sorry state of Linux OGL drivers in the last 20 years. Nvidia dominate here and have done since I was a spotty teenager. AMD were on the opengl ARB, which is now called the Khronos BoD, that is the exact same damn thing. They contributed, just less, they cared less, had worse drivers even in windows

I repay this agin, the president of the Khronos group is the Vice President of nvidia. That should tell you just how involved nvidia is with Volkan. Volkan is the successor of opengl, of which nvidia was a much bigger supporter.

 

[D.P.]

Wonder if AMD was payed to drop it from MS. Since mantle was pretty much dx12 MS had a interest to keep support more for dx than mantle. Maybe not true of course but i do wonder and its not like AMD couldnt do with the money.

Nvidia, AMD, and a load of others like game developers were pressuring Microsoft since 2005 to develop a DX12 like API. MS politics likely delaysed things, the xbox360 performance liekly made them realize they need a 21st century grpshics API.

AMD dropped mantle because frankly it was utterly pointless. Vendor specific APIs was the death of companies like 3DfX, S3, rendition, matrix, power VR. Vendors specific APIs were all the rage in the mid 90s, there is a reason thing like glide, s3Metal, redline, inline, sgl, nvlib, dont exist anymore. Writing a bed or specific API is dead easy, every man and his dog has done one before.

 

[humbug]

They didn't drop it ^^^ its now Vulkan. Although you are trying hard to argue that Mantle has little to do with Vulkan despite the Khronos group saying Mantle is the core of Vulkan and would not have been possible without it.

You are simply wrong again,. GLnext was in development long before mantle came to being, but the develop,net was in design ideas,architectures, features, philosophy, paradigm, theory and minimal code. Opengl ARB is a slow beast, that is the main criticism of opengl. The idea of Glbext was soe thing akin to mantle, not an evolution of OGL, that is why they called it GLnext, and not gl5.0., and why they changed the name to Volkan and not OGLnext. Everyone wanted a fresh start, everyone wanted to drop legacy support. Everyone one a modern future looking API. Everyone wanted multi threading and efficiency at its core. Everyone wanted to reduce overhead. That was the whole philosophy of the project long befor AMD contributed any code. The design goals was something akin to mantle but cross hardware, cross vendor, compatible, open,
And developed and maintained by a consortium and not a single interested party, e.g. AMD. That matter part is a. Very critical point and is why neither intro nor nvidia or anything else like Samsung, Apple, Qualcomm etc. touched mantle with a barge pole.

Eventually AMD gave up on on Mantle and rather than waste all that development they gave it to the Kronos group to learn from and take advantage of. That is great, I really applauded AMD for doing that and I really doubt nvidia would have done the same so quickly. But Manlte was a hardware and vendor specific bets I eolentation of some of the ideas of GLnext that had already been widely developed upon. This made a useful code base to start building open, and nvidia admit to this the selves. But what Khronos clearly states that the code had radii ally developed and changed to bring in line with what glNext's vision was and allow it to be cross hardware and architecture agnostic, this is like the Greek ship myth, if you replace every Olank of wood, the sail, the mast, the keel, the rudder, etc, is it still the same ship, even although know instead of a sail it has a twin V12 10liter Diesel engine.

The Vice President of nvidia is the president of. The Khronos group. Don't under estimate how invested in Vulcan voids is, how much support they give, and how much they have contributed.

Your whip at OGL is patently worms. OGL never got mainstream windows use because of Microsoft not Supporting it in favor of its own DX as a competitive advantage against Apple and Linux. Nvidia's extension exist because of the nature of the library that permits vendors to extend the base functionality rather than waiting for a new version. You have to realize many of Nvida's extension were simply rename to become part of the next core, they had a huge influence in adding functionality. AMD just never had the same interet because they don't have the same profesional graphics market any more. Jus look at the sorry state of Linux OGL drivers in the last 20 years. Nvidia dominate here and have done since I was a spotty teenager. AMD were on the opengl ARB, which is now called the Khronos BoD, that is the exact same damn thing. They contributed, just less, they cared less, had worse drivers even in windows

I repay this agin, the president of the Khronos group is the Vice President of nvidia. That should tell you just how involved nvidia is with Volkan. Volkan is the successor of opengl, of which nvidia was a much bigger supporter.

You are correct that AMD had little interest in OGL, although they have worked pretty hard on it in the last year and their performance is massively higher today that it is from what you perhaps know, in the last few moths its doubled, even tripled.

OGL Gaming is tiny, this is why they had no interest in OGL, now that gaming in its self is taking an interest AMD are, hence to amount of work they are putting into it now. quite aside from Vulkan.
In professional workstations AMD are geared toward Compute so they put their resources into OpenCL, something Nvidia don't as they use CUDA. both AMD and Nvidia put resource into what they need, Nvidia its OpenGL, AMD its OpenCL and HSA.

As the Khronos Group said its extremely difficult to make an API like Mantle from scratch, especially given all the different members all wanting their own input (See whats going on there) Khronos Group thanked AMD for giving them Mantle which gave them exactly what they wanted. i bet Nvidia didn't approve that statement and if so no one in the Khronos Group will have listened to them, shut down by the rest.
GLNext in the direction it was going this is not what the Khronos Group wanted, i know as you keep saying Nvidia are a part of the Khronos Group so are a lot of others, Samsung, Apple, Qualcomm were also set dead against Nvidia's propitiatory monopoly with OGL, this is a reason as to why no one but Nvidia are interested in OGL, its not just AMD, its everyone but Nvidia. they are all in competion with Nvidia in the Open Source API industry and don't want Nvidia monopolising it for their own advancement, this is why Nvidia has put a lot of resources into OGL, its mostly for their own benefit, Mantle is the only thing that offers everyone a true for all API as that has no propitiatory lockouts and was already largely complete, true vendor cross platform, its a lot like AMD's HSA which is AMD's invention, they head it, they put almost all of the development work in but because its all for everyone Samsung, Apple, Qualcomm, LG ecte.... are members.
Nvidia are very much on their own in what they want from whatever the API ends up being, Now that AMD handed over the Mantle API they no longer needed the Nvidia favoured API. Yes Samsung, Qualcomm..... they do want it, they already have a relationship with AMD in Open API's. Even Apple are moving away from CUDA because they don't like Nvidia's lockin API's, no one does, this isn't 2002. these people want a working relationship that is based on a mutual partnership, they don't want to be tied to a vendor.

But we ain't going to agree on anything here so let's just leave it as disagreed.

 

[Final8y]

They didn't drop it ^^^ its now Vulkan. Although you are trying hard to argue that Mantle has little to do with Vulkan despite the Khronos group saying Mantle is the core of Vulkan and would not have been possible without it.



You are correct that AMD had little interest in OGL, although they have worked pretty hard on it in the last year and their performance is massively higher today that it is from what you perhaps know, in the last few moths its doubled, even tripled.

OGL Gaming is tiny, this is why they had no interest in OGL, now that gaming in its self is taking an interest AMD are, hence to amount of work they are putting into it now. quite aside from Vulkan.
In professional workstations AMD are geared toward Compute so they put their resources into OpenCL, something Nvidia don't as they use CUDA. both AMD and Nvidia put resource into what they need, Nvidia its OpenGL, AMD its OpenCL and HSA.

As the Khronos Group said its extremely difficult to make an API like Mantle from scratch, especially given all the different members all wanting their own input (See whats going on there) Khronos Group thanked AMD for giving them Mantle which gave them exactly what they wanted. i bet Nvidia didn't approve that statement and if so no one in the Khronos Group will have listened to them, shut down by the rest.
GLNext in the direction it was going this is not what the Khronos Group wanted, i know as you keep saying Nvidia are a part of the Khronos Group so are a lot of others, Samsung, Apple, Qualcomm were also set dead against Nvidia's propitiatory monopoly with OGL, this is a reason as to why no one but Nvidia are interested in OGL, its not just AMD, its everyone but Nvidia. they are all in competion with Nvidia in the Open Source API industry and don't want Nvidia monopolising it for their own advancement, this is why Nvidia has put a lot of resources into OGL, its mostly for their own benefit, Mantle is the only thing that offers everyone a true for all API as that has no propitiatory lockouts and was already largely complete, true vendor cross platform, its a lot like AMD's HSA which is AMD's invention, they head it, they put almost all of the development work in but because its all for everyone Samsung, Apple, Qualcomm, LG ecte.... are members.
Nvidia are very much on their own in what they want from whatever the API ends up being, Now that AMD handed over the Mantle API they no longer needed the Nvidia favoured API. Yes Samsung, Qualcomm..... they do want it, they already have a relationship with AMD in Open API's. Even Apple are moving away from CUDA because they don't like Nvidia's lockin API's, no one does, this isn't 2002. these people want a working relationship that is based on a mutual partnership, they don't want to be tied to a vendor.

But we ain't going to agree on anything here so let's just leave it as disagreed.

Well put.

 

[Marine-RX179]

They didn't drop it ^^^ its now Vulkan. Although you are trying hard to argue that Mantle has little to do with Vulkan despite the Khronos group saying Mantle is the core of Vulkan and would not have been possible without it.



You are correct that AMD had little interest in OGL, although they have worked pretty hard on it in the last year and their performance is massively higher today that it is from what you perhaps know, in the last few moths its doubled, even tripled.

OGL Gaming is tiny, this is why they had no interest in OGL, now that gaming in its self is taking an interest AMD are, hence to amount of work they are putting into it now. quite aside from Vulkan.
In professional workstations AMD are geared toward Compute so they put their resources into OpenCL, something Nvidia don't as they use CUDA. both AMD and Nvidia put resource into what they need, Nvidia its OpenGL, AMD its OpenCL and HSA.

As the Khronos Group said its extremely difficult to make an API like Mantle from scratch, especially given all the different members all wanting their own input (See whats going on there) Khronos Group thanked AMD for giving them Mantle which gave them exactly what they wanted. i bet Nvidia didn't approve that statement and if so no one in the Khronos Group will have listened to them, shut down by the rest.
GLNext in the direction it was going this is not what the Khronos Group wanted, i know as you keep saying Nvidia are a part of the Khronos Group so are a lot of others, Samsung, Apple, Qualcomm were also set dead against Nvidia's propitiatory monopoly with OGL, this is a reason as to why no one but Nvidia are interested in OGL, its not just AMD, its everyone but Nvidia. they are all in competion with Nvidia in the Open Source API industry and don't want Nvidia monopolising it for their own advancement, this is why Nvidia has put a lot of resources into OGL, its mostly for their own benefit, Mantle is the only thing that offers everyone a true for all API as that has no propitiatory lockouts and was already largely complete, true vendor cross platform, its a lot like AMD's HSA which is AMD's invention, they head it, they put almost all of the development work in but because its all for everyone Samsung, Apple, Qualcomm, LG ecte.... are members.
Nvidia are very much on their own in what they want from whatever the API ends up being, Now that AMD handed over the Mantle API they no longer needed the Nvidia favoured API. Yes Samsung, Qualcomm..... they do want it, they already have a relationship with AMD in Open API's. Even Apple are moving away from CUDA because they don't like Nvidia's lockin API's, no one does, this isn't 2002. these people want a working relationship that is based on a mutual partnership, they don't want to be tied to a vendor.

But we ain't going to agree on anything here so let's just leave it as disagreed.

Sounds about right.

The irony of someone keep praising Nvidia criticising AMD for making things proprietary, when Nvidia is the biggest fan of making things proprietary in similar way to Apple

The are many things that AMD can be criticised for by Nvidia supporters, but on proprietary? lol that's a good laugh.

 

[andybird123]

Opengl is tiny and proprietary?
Do you even read these things before you write them, what utter rubbish.

 

[CAT-THE-FIFTH]

Comparison of the functions between Vulkan and Mantle:
http://i.imgur.com/aq62YN7.jpg

 

[LambChop]

Comparison of the functions between Vulkan and Mantle:
http://i.imgur.com/aq62YN7.jpg

The one in bold is the only match ?

 

[CAT-THE-FIFTH]

The one in bold is the only match ?

No. All the ones where the names are exactly the same.

 

[humbug]

No. All the ones where the names are exactly the same.

[vkSARCASM] Careful, Vulkan and Mantle have different prefixes, thats all so massively important and proves they are not at all alike [/grSARCASM]

 

[Killer7xx]

The one in bold is the only match ?

No, the image shows the matching functions across both APIs. Since most of the vulkan functions have the same name as the mantle functions but with a "vr" prefix instead of "gr", it can be assumed that they are the same function but renamed. In a way, it shows how mantle is very much a foundation for vulkan.