GTA V Supporting Vulkan API (Roy doesnt know what he talking about)

[bob345]

im really hoping that my 770 will be up to running this game once it is finally released for pc.

[Monarch]

So turns out Roy actually didn't mean to say GTA V will support Vulcan.

 

https://twitter.com/amd_roy/status/583757120954314753

https://twitter.com/amd_roy/status/583757931272916992

 

Move along people, nothing to see here.

[bogus]

No it's basically Mantle with a new name.

Fixed that for you. You are welcome

[NateGSR117]

Wow, this is sweet! I just can't justify paying full price tho

[Monarch]

Wow, this is sweet! I just can't justify paying full price tho

 

https://twitter.com/amd_roy/status/583757120954314753

[Thony]

Hmmm I wonder if it will also support/update to have DX12

Games was made in DX11 I think so doubt they would put more effort into making it DX12

[JAKEBAB]

Fixed that for you. You are welcome

That's like saying DX12 is mantle but with a different name...... it's not, So no thank you.

[Sakkura]

That's like saying DX12 is mantle but with a different name...... it's not, So no thank you.

vulkan is largely based on mantle though.

[LukaP]

Fixed that for you. You are welcome

No

 

vulkan is largely based on mantle though.

And Kepler is largely based on Fermi. I dont see anyone saying a 780Ti is a 580

 

Yes, DX12 and Vulcan do largely the same thing as mantle does. And the code reflects that. you cant make code that does a particular thing look THAT different from someone elses code. Especially when that someone else has done a great job and lets you learn from it. You guys oversimplify it and think Vulcan is just the exact Mantle code and a new name, but that is ridiculous. Mantle is closed into the GCN1.0+ Arch. And its libraries are written to reflect that. It would simply not run on anything else. It had to be rewritten for a crossplatform API. But of course, the same coding principles applied to Vulcan and DX12 as they did to Mantle. Hence why people say its a part of them. In the same way as Windows 7 is basically Windows NT. Yes their kernels use the same coding principles, but in reality they are much more different than one may think

[Sakkura]

 

And Kepler is largely based on Fermi. I dont see anyone saying a 780Ti is a 580

 

Yes, DX12 and Vulcan do largely the same thing as mantle does. And the code reflects that. you cant make code that does a particular thing look THAT different from someone elses code. Especially when that someone else has done a great job and lets you learn from it. You guys oversimplify it and think Vulcan is just the exact Mantle code and a new name, but that is ridiculous. Mantle is closed into the GCN1.0+ Arch. And its libraries are written to reflect that. It would simply not run on anything else. It had to be rewritten for a crossplatform API. But of course, the same coding principles applied to Vulcan and DX12 as they did to Mantle. Hence why people say its a part of them. In the same way as Windows 7 is basically Windows NT. Yes their kernels use the same coding principles, but in reality they are much more different than one may think

 

DirectX 12 accomplishes many of the same things as Mantle and Vulkan, but is a separate project. However, Vulkan is largely derived from Mantle. The closely connected names are not a coincidence, and it has been confirmed publicly that Vulkan is built with Mantle components.

[LukaP]

DirectX 12 accomplishes many of the same things as Mantle and Vulkan, but is a separate project. However, Vulkan is largely derived from Mantle. The closely connected names are not a coincidence, and it has been confirmed publicly that Vulkan is built with Mantle components.

As I said, yes, Vulcan developers had insight into Mantle, and used it. But Vulcan is not even close to Mantle. First off, it supports features that Mantle does not, to achieve feature parity with DX12. Second off, it supports GCN, Kepler, Maxwell, Gen7(.5) and Gen8. That is one hell of a different code to do that, compared to Mantle. 

 

I know what youre trying to say, and I know that. Ive been following the API news closely, as a game developer and as a spare time programmer. AMD gave Chaos access to mantle source code, so they could work on oGLNext (or now called Vulcan) faster and catch up with DX12, which was in development simultaneously (or perhaps even before) as Mantle. But in no way did Chaos take Mantle code and strap a new name onto it.

 

You have to realise how low level mantle (and to some extent, DX12 and Vulcan) code is. Mantle simply would not run on anything but GCN. Its like trying to natively run ARM code on x86. Without binary translation or some clever hardware tricks, it doesnt work. And if they were to use that, you immediately lose all benefits of being a low level API on anything but GCN again.

 

So what im trying to say (and i seem to be just going around in circles explaining it from different angles) is that Vulcan isnt Mantle, just like DX12 isnt Mantle. In reality i would wager Vulcan resembles DX12 more than mantle by this point

[HippY]

Honestly, who is that Roy?

he seems to spread misinformation in the past year+ hype things that does not need to be hyped...

[LukaP]

Honestly, who is that Roy?

he seems to spread misinformation in the past year+ hype things that does not need to be hyped...

Some AMD Marketing employee i think. Maybe technical marketing

[Sakkura]

As I said, yes, Vulcan developers had insight into Mantle, and used it. But Vulcan is not even close to Mantle. First off, it supports features that Mantle does not, to achieve feature parity with DX12. Second off, it supports GCN, Kepler, Maxwell, Gen7(.5) and Gen8. That is one hell of a different code to do that, compared to Mantle.

 

That's not much of a difference. Mantle was not GCN-exclusive, it's just that Nvidia didn't want to use an AMD-controlled API.

 

Vulkan is based on Mantle components, not just as a source of inspiration but actual components being taken from Mantle and put in Vulkan. Vulkan is like Mantle 1.1.

 

I know what youre trying to say, and I know that. Ive been following the API news closely, as a game developer and as a spare time programmer. AMD gave Chaos access to mantle source code, so they could work on oGLNext (or now called Vulcan) faster and catch up with DX12, which was in development simultaneously (or perhaps even before) as Mantle. But in no way did Chaos take Mantle code and strap a new name onto it.

 

You have to realise how low level mantle (and to some extent, DX12 and Vulcan) code is. Mantle simply would not run on anything but GCN. Its like trying to natively run ARM code on x86. Without binary translation or some clever hardware tricks, it doesnt work. And if they were to use that, you immediately lose all benefits of being a low level API on anything but GCN again.

 

So what im trying to say (and i seem to be just going around in circles explaining it from different angles) is that Vulcan isnt Mantle, just like DX12 isnt Mantle. In reality i would wager Vulcan resembles DX12 more than mantle by this point

 

Chaos? Vulkan is made by Khronos.

 

And yes, Mantle would certainly support architectures apart from GCN, it just never got there because of a lack of support from Nvidia. Intel was tinkering with Mantle support, but that work went into Vulkan instead.

[LukaP]

That's not much of a difference. Mantle was not GCN-exclusive, it's just that Nvidia didn't want to use an AMD-controlled API.

 

Vulkan is based on Mantle components, not just as a source of inspiration but actual components being taken from Mantle and put in Vulkan. Vulkan is like Mantle 1.1.

 

 

Chaos? Vulkan is made by Khronos.

 

And yes, Mantle would certainly support architectures apart from GCN, it just never got there because of a lack of support from Nvidia. Intel was tinkering with Mantle support, but that work went into Vulkan instead.

Sorry, meant Khronos... 

 

Mantle was and still is GCN exclusive. Mantle was developed as a closed source, with a promise of an SDK and open sourcing it in the future. none of that happened. NV didnt want to do anytihng with mantle, because they were hard at work with MS on DX12. (also they got a look into mantle when AMD opened it to KG)

 

 

 

Vulkan is based on Mantle components, not just as a source of inspiration but actual components being taken from Mantle and put in Vulkan. Vulkan is like Mantle 1.1.

No. Because what youre saying makes no sense. To give you an analogy, its the same as BMW making a fast sports car, and while developing it, taking a look at their old designs. And then, instead of learning from those and making a great car (like they always do) they would take an engine from a 1 series and put it in there. then they would realise its not really fast, but its has great economy when youre driving slow and deciding to just add another more powerful engine in there, and then switch between them. Instead of rather building a new engine that is both powerful as well as economic. Would make more sense right?

 

Its the same with Vulcan. They saw what Mantle can do, but it can only do it in one special case, which is GCN. So why would they possibly take mantle source and put it into their API, when they then have to write source for every other architecture too... No, they write code that effectively abstracts the silicon itself, but is still very low level. so they get the best of both worlds.

 

What mantle was, was an experiment to push the industry, nothing more. and when that was done, AMD handed the source to someone who needed it to stay competitive. But i can assure you, no matter how much of the codebase Khronos has seen, Copy pasting wasnt a thing they did. and neither is Vulcan Mantle1.1...

 

 

 

And yes, Mantle would certainly support architectures apart from GCN, it just never got there because of a lack of support from Nvidia. Intel was tinkering with Mantle support, but that work went into Vulkan instead.

Mantle could support other GPU architectures outside of GCN (it already supports GCN1.1, 1.2 and 2.0). but it does not support those radically different ones. and it never will. It could support them, if AMD opened it up, and let NV add libraries to support their cards. But AMD wont open it, and even if they do, NV has no benefit in using it, with DX12 and Vulcan here.

 

Intel is in a very interesting position, as their GPU arch is still "basic" enough so that they emulate a lot of features in their driver. which means they already have driver code for emulating things. meaning they could simply modify the driver to emulate hardware needed for mantle to run, and they had it running well enough to fiddle with it. but they still merely fiddled with it.

 

Now if you can give me definitive proof that there is mantle source code inside Vulcan, then you have a point. otherwise, facts(vulcan being crossplatform and having feature parity with DX) and common sense make my case

[TwistedDictator]

"Mantle was and still is GCN exclusive. Mantle was developed as a closed source, with a promise of an SDK and open sourcing it in the future. none of that happened. NV didnt want to do anytihng with mantle, because they were hard at work with MS on DX12 because of pride..."

 

Fixed that for you

[Sakkura]

Sorry, meant Khronos... 

 

Mantle was and still is GCN exclusive. Mantle was developed as a closed source, with a promise of an SDK and open sourcing it in the future. none of that happened. NV didnt want to do anytihng with mantle, because they were hard at work with MS on DX12. (also they got a look into mantle when AMD opened it to KG)

 

 

No. Because what youre saying makes no sense. To give you an analogy, its the same as BMW making a fast sports car, and while developing it, taking a look at their old designs. And then, instead of learning from those and making a great car (like they always do) they would take an engine from a 1 series and put it in there. then they would realise its not really fast, but its has great economy when youre driving slow and deciding to just add another more powerful engine in there, and then switch between them. Instead of rather building a new engine that is both powerful as well as economic. Would make more sense right?

 

Its the same with Vulcan. They saw what Mantle can do, but it can only do it in one special case, which is GCN. So why would they possibly take mantle source and put it into their API, when they then have to write source for every other architecture too... No, they write code that effectively abstracts the silicon itself, but is still very low level. so they get the best of both worlds.

 

What mantle was, was an experiment to push the industry, nothing more. and when that was done, AMD handed the source to someone who needed it to stay competitive. But i can assure you, no matter how much of the codebase Khronos has seen, Copy pasting wasnt a thing they did. and neither is Vulcan Mantle1.1...

 

Mantle could support other GPU architectures outside of GCN (it already supports GCN1.1, 1.2 and 2.0). but it does not support those radically different ones. and it never will. It could support them, if AMD opened it up, and let NV add libraries to support their cards. But AMD wont open it, and even if they do, NV has no benefit in using it, with DX12 and Vulcan here.

 

Intel is in a very interesting position, as their GPU arch is still "basic" enough so that they emulate a lot of features in their driver. which means they already have driver code for emulating things. meaning they could simply modify the driver to emulate hardware needed for mantle to run, and they had it running well enough to fiddle with it. but they still merely fiddled with it.

 

Now if you can give me definitive proof that there is mantle source code inside Vulcan, then you have a point. otherwise, facts(vulcan being crossplatform and having feature parity with DX) and common sense make my case

 

Mantle was only temporarily GCN-exclusive, because AMD was still developing it. They just decided to hand it over to Khronos instead of making an open Mantle, because Mantle would never have gained the support that Vulkan will. Nvidia wouldn't accept it.

 

Your BMW analogy is what makes no sense. Khronos decided to ditch all their old OpenGL code in favor of Mantle and what they built on top of that. It's like working on piston-engined planes and then getting a jet engine design to build their new plane around.

 

Your claim that Mantle could never support architectures apart from GCN is demonstrably wrong. Why would Intel ask for access to Mantle if it wouldn't be able to run on their GPU architectures? It's nonsense, of course Mantle could be used with other architectures.

 

Here is what Khronos has said:

 

 

Many companies have made great contributions to Vulkan, including AMD who contributed Mantle. Being able to start with the Mantle design definitely helped us get rolling quickly

 

So Vulkan is clearly built on Mantle. As Pcper puts it:

 

 

the Vulkan API was definitely started with Mantle and grew from there

[LukaP]

Mantle was only temporarily GCN-exclusive, because AMD was still developing it. They just decided to hand it over to Khronos instead of making an open Mantle, because Mantle would never have gained the support that Vulkan will. Nvidia wouldn't accept it.

 

Your BMW analogy is what makes no sense. Khronos decided to ditch all their old OpenGL code in favor of Mantle and what they built on top of that. It's like working on piston-engined planes and then getting a jet engine design to build their new plane around.

 

Your claim that Mantle could never support architectures apart from GCN is demonstrably wrong. Why would Intel ask for access to Mantle if it wouldn't be able to run on their GPU architectures? It's nonsense, of course Mantle could be used with other architectures.

 

Here is what Khronos has said:

 

 

So Vulkan is clearly built on Mantle. As Pcper puts it:

Mantle still is GCN exclusive, unless you have any kind of article that states either NV or Intel cards can now use mantle... AMD claimed they were going to release an SDK and source. Then came DX12 and oGLN. We havent heard anything about releasing the source since...

 

KG ditched all oGL code. Then they looked at Mantle code. Then they wrote their own code. You seem to not be able to comprehend that Mantle code is very low level. It literally cannot even execute on anything but GCN.

 

i said Mantle doesnt support other architectures. I never said it could not support them.  It seems you have comprehension problems this whole time. 

Intel could run it, i even gave you a reason as to why Intel had an easier time getting Mantle to execute on their GPUs...

 

 

 

Many companies have made great contributions to Vulkan, including AMD who contributed Mantle. Being able to start with the Mantle design definitely helped us get rolling quickly

This is exactly what i have been saying you git... AMD gave them access to Mantle, so they could more easily see the coding practices behind low level APIs (since KG doesnt have the low level experience MS has, from working on the X Box APIs) and be able to more quickly catch up with DX12. But Mantle is not a part (or subset, like you seem to be thinking) of Vulcan in any way, apart from spiritually, if you will. 

Vulcan is its own thing, that was build by Khronos engineers after they had a look at Mantle. Never did anyone at Khronos or AMD even as much as imply that any part of Mantle has been directly used in Vulcan.

[Sakkura]

KG ditched all oGL code. Then they looked at Mantle code. Then they wrote their own code. You seem to not be able to comprehend that Mantle code is very low level. It literally cannot even execute on anything but GCN.

 

I just provided proof that they took Mantle code and built some stuff on top of that to make Vulkan.

 

As for your claim that Mantle cannot run on anything other than GCN, that is obviously completely wrong. And you've provided zero proof for that weird claim.

[Dabombinable]

I just provided proof that they took Mantle code and built some stuff on top of that to make Vulkan.

 

As for your claim that Mantle cannot run on anything other than GCN, that is obviously completely wrong. And you've provided zero proof for that weird claim.

Does this fix it for you?

 

 

-snip-

i said Mantle doesnt support other architectures. I never said it could not support them.  It seems you have comprehension problems this whole time. 

Intel could run it, i even gave you a reason as to why Intel had an easier time getting Mantle to execute on their GPUs...

-snip

[Sakkura]

Does this fix it for you?

 

 

It contradicts earlier comments like this:

 

 

Mantle could support other GPU architectures outside of GCN (it already supports GCN1.1, 1.2 and 2.0). but it does not support those radically different ones. and it never will.

[Dabombinable]

It contradicts earlier comments like this:

Source of that quote?

[LukaP]

It contradicts earlier comments like this:

how the fuck does

Mantle could support other GPU architectures outside of GCN (it already supports GCN1.1, 1.2 and 2.0). but it does not support those radically different ones. and it never will.

contradict

-snip-

i said Mantle doesnt support other architectures. I never said it could not support them.  It seems you have comprehension problems this whole time. 

Intel could run it, i even gave you a reason as to why Intel had an easier time getting Mantle to execute on their GPUs...

-snip

It literally says the same thing. Mantle could support other architectures, but it does not. And it never will, because Mantle is an AMD thing. Thats it.

[Sakkura]

Source of that quote?

 

Here

 

And here's some more in that vein:

 

You have to realise how low level mantle (and to some extent, DX12 and Vulcan) code is. Mantle simply would not run on anything but GCN. Its like trying to natively run ARM code on x86. Without binary translation or some clever hardware tricks, it doesnt work. And if they were to use that, you immediately lose all benefits of being a low level API on anything but GCN again.

 

Mantle simply would not run on anything but GCN... except of course if you made it run on anything but GCN, as AMD was planning to since 2013.

 

http://techreport.com/news/25651/mantle-to-power-15-frostbite-games-dice-calls-for-multi-vendor-support

 

http://www.dsogaming.com/news/amds-mantle-does-not-require-gpus-with-gcn-architecture/

 

Those plans did not come to fruition, probably because of resistance from Nvidia. Instead that development took place after Mantle was "forked" by Khronos.

[Albatross]

OP content has been deleted. What is the topic news exactly?