AMD proprietary API: Mantle

[Mechromancer]

From TechSpot

AMD unveils revolutionary 'Mantle' API to optimize GPU performance

At AMD's GPU14 Tech Day event in Hawaii, alongside the brand new Radeon R9 290X GPU, the company has also unveiled a new development model called 'Mantle'. As a combination of both API specifications and GPU drivers, Mantle will allow developers to fully unlock the graphics potential of the GCN 2.0 architecture of the new Radeon series, through low-level high-performance drivers.

Battlefield 4 will be the first title to use Mantle, with the technology coming through a free update in December. Mantle brings greater control of the GPU and CPU to game developers, especially in the memory department, which will allow games to harness the full potential of graphics cards. This model will be more similar to what is found on consoles, where developers have huge control over what hardware they use in their games.

While the technology is still in early stages, EA says it will allow PC game developers to harness even more graphics power than ever before. Where possible, Mantle will be used instead of DirectX 11 on compatible AMD hardware, although it's still compatible with other rendering APIs. AMD claims Mantle enables nine times more draw calls per second than other APIs, which is a huge increase in performance.

We've been told at the GPU14 Tech Day event that the Mantle API is open, so theoretically Nvidia could purpose the technology in their GPUs. It should also make cross-development between PC and console games a lot easier, and also more incredible for those with a high-performance AMD GPU.

More information on Mantle will be available at the AMD Developer Summit in November.

[Mechromancer]

I guess I should have bought that HD7950 on sale instead of the GTX 760 for Battlefield 4....FUUUUUUUUUUUUUUUUUUUUUUUUU

[Iconyu]

Why do people keep saying proprietary for this? Like AMD could write low level drivers for Nvidia to make it fair.

[DefStar]

Nice! It's about time AMD took hold of the reins. It will be nice to see AMD's hardware pushed to it's full potential.

[Mechromancer]

Why do people keep saying proprietary for this? Like AMD could write low level drivers for Nvidia to make it fair.

....wait....what?

[Mechromancer]

Here is a more detailed explanation of the what, why, and how:

[Iconyu]

....wait....what?

It's open with no fees, so nvidia could write drivers for it and join in too. Nvidia jumped on board for both 3Dc and TressFX ATI/AMD, so what makes this API 'proprietary'?

The 'open' is the exact opposite to the word 'proprietary' in software, so why did you use the word?

[Boogerlad]

I don't think you understand what low level means. Low level means direct access to the hardware, without abstraction layers. Nvidia architecture != AMD's architecture.

[nossy23]

Also notice that they said that GPU compute will become a lot more important for games. Not so good news for Nvidia.

[MaddMuppet]

Will be interesting to see where this heads. I guess we'll have to stay tuned.

[Ace123]

Its about time AMD start to write their own future

[Piledriver]

I said a a few months ago, AMD should start to focus on software and it's nice to see they are doing it.



Same hardware, different software, over 100% performance improvement.

[Piledriver]

I said a a few months ago, AMD should start to focus on software and it's nice to see they are doing it.



Same hardware, different software, over 100% performance improvement.

[eric66]

if i were Amd i would kept that exclusive if it really brings that kinda of advantage

[Fottemberg]

Why do people keep saying proprietary for this? Like AMD could write low level drivers for Nvidia to make it fair.

However Mantle will be great for the APUs (= Epic Performance Boost).

[Iconyu]

However Mantle will be great for the APUs (= Epic Performance Boost).

This is not proprietary software, even OP's quote says it's not. Here, I even included a definition of the word for you:

"Proprietary software or closed source software is computer software licensed under exclusive legal right of the copyright holder with the intent that the licensee is given the right to use the software only under certain conditions, and restricted from other uses, such as modification, sharing, studying, redistribution, or reverse engineering. Usually the source code of proprietary software is not made available."

Just like Nvidia's FXAA, or AMD's 3Dc texture compression and TressFX, some technology is simply shared without much fuss. If you or the op has a rebuttal that either uses full sentences or doesn't use silly pictures I'll entertain them once you manage to put them together.

[Fottemberg]

Low Level Optimization = Born to be Radeon

https://www.youtube.com/watch?v=9UaJAnnipkY

nVidia could realize its own API, but these shouldn't be Mantle API. Do you remember Glide API?

[Iconyu]

It's low level access, not low level optimisation, there is a huge difference. Nvidia doesn't NEED to make it's own API, they can use Mantle in conjunction with their own drivers.

[Fottemberg]

I don't think so. Mantle is a huge opportunity to trade the XB1/PS4 games to PC, and viceveresa, using the same graphic engines and optimizations. Less money to invest in the expansive process of conversion, because the hardware is the same. Using Mantle with nVidia cards is a non-sense, imho.

[Hornet331]

Its quite interesting how people forget the past, remember CTM?

The only only thing I see it can get sucessfull if it is 100% compatible with opengl, meaning you just press a button and have your code in opengl. Else it will face the same fate a ctm and get replaced by opengl (for consoles)

[Iconyu]

I don't think so. Mantle is a huge opportunity to trade the XB1/PS4 games to PC, and viceveresa, using the same graphic engines and optimizations. Less money to invest in the expansive process of conversion, because the hardware is the same. Using Mantle with nVidia cards is a non-sense, imho.

AMD have pretty much created a system that removes the biggest bottleneck in PC gaming, it's not about console ports, it's about taking the shackles off developers. If you think engine optimisation is a big expense in game development in comparison to asset creation you're mistaken. Developers can just port over to PC as is, the brute force of the platform makes up for sloppy work, see Demon Souls.

Mantle is a huge advantage, but AMD left it open, with that set up it would be stupid for Nvidia to set up a rival API as EA is already fully behind Mantle, and AMD have too many developers behind them. So Nvidia has two choices step on board to limit damage, and get some performance back or step back and try to counter it some how.

But looking at the history of the two companies, if one side creates something and says it's open, the other side supports it nearly instantly.

Its quite interesting how people forget the past, remember CTM?

The only only thing I see it can get sucessfull if it is 100% compatible with opengl, meaning you just press a button and have your code in opengl. Else it will face the same fate a ctm and get replaced by opengl (for consoles)

CTM is purely for GPGPU right? It has nothing to do with OpenGL, and AMD threw their weight behind OpenCL.

[fellix_bg]

AMD literary threw another "x86-64" bomb all over again with this move. Very brave decision, indeed!

The new API isn't about some uber smart tech, but a matter of having balls to challenge the stagnant status quo in the PC market, with a solution that actually everyone has been looking for a while, but no one dared to make the first step. And it's no wonder that such actions usually come from the "underdogs".

[StAndrew]

Mantle is a huge advantage, but AMD left it open, with that set up it would be stupid for Nvidia to set up a rival API

But looking at the history of the two companies, if one side creates something and says it's open, the other side supports it nearly instantly.

Be interesting to see how NVIDIA responds. I disagree with your rendition of the history... NVIDIA tends to want its technologies "exclusive" to their hardware. Just look at Havok and PhysX.

[Stukov]

Be interesting to see how NVIDIA responds. I disagree with your rendition of the history... NVIDIA tends to want its technologies "exclusive" to their hardware. Just look at Havok and PhysX.

Havok was bought out by Intel and even though Nvidia likes their exclusive stuff, they still adopt open standards (such as DirectX features which is similar to Mantle).

[zalbard]

It's low level access, not low level optimisation, there is a huge difference.

All I've read about it points at low level access to GCN features. Please feel free to prove otherwise.

On paper, it does provide low level access, yes. And what's at that level? Hardware architecture. Which is completely different across GPU generations, not even talking about hardware manufacturers. This is why OpenGL and DirectX exist, and this is why they are "high level APIs", meaning abstraction from underlying architecture and hardware.

I don't think what you are trying to convey here is even possible.