AMD Mantle API [updating]

[Arwin]

I believe that engines like Unity, Unreal, CryEngine, Frostbite et al could spend some of their resources on optimizing for hardware at a lower level, and are also the greatest benefactors of being able to experiment.

That's not to say that they will - other factors are at play too. But it's possible. There's a big gap between the power of a PC GPU and the performance it delivers, and narrowing that gap would certainly make PC more competitive with consoles, which in turn could make it more worthwhile for developers to optimise for specific PC hardware 'models', creating a spiralling effect.

 

[ams]

Optimal and better, not equal.

Well actually no, because it is not better for any non GCN-equipped system, which represents the vast majority of systems on the marketplace. At least DirectX and OpenGL advancements are accessible to any IHV.

The worst part of all of this is that AMD is giving people the impression that Mantle is an open graphics API, when in reality it is only GCN-equipped systems that stand to benefit directly from this API.

At the end of the day, I don't think there is a single serious game developer on earth who would think that IHV-specific graphics API's would be a good trend moving forward. If you read in between the lines, it is very obvious what Johan's view on this is from a long-term perspective.

 

[Dave Baumann]

In this day and age, developers need to maximize graphics performance and stability across a wide range of platforms including: PC, Xbox360, PS3, 3dS, PSP, iOS, Android, Xbox1, PS4, etc.

And I’ll raise this point again. Game developers are spending most of their time on the game, not the rendering engine. The render engine developers are now targeting widespread support across multiple target platforms and in fact targeting another renderer is not that much of an overhead when you consider your list above and when you consider the target number of applications that render will be used on – for 3rd party game engine developers having multiple optimized render targets is actually a selling feature. While Johan isn’t necessarily selling his engine externally, it’s no coincidence that the first thing he mentioned on his Mantle presentation is the depth of titles that will make use of this iteration of the Frostbite engine; build a render target once and lots of titles can make use of it with very little overhead for all the developers of those titles.

 

[DieH@rd]

Excellent news for all the fans of Frostbite 3 engine... and Mass Effect... and Dragon Age... and Mirrors Edge ... and Star Wars games:

https://twitter.com/repi/status/391958934489006080

Hopefully this is only first of many big engines that will support it.

 

[ams]

And I’ll raise this point again. Game developers are spending most of their time on the game, not the rendering engine. The render engine developers are now targeting widespread support across multiple target platforms and in fact targeting another renderer is not that much of an overhead when you consider your list above and when you consider the target number of applications that render will be used on – for 3rd party game engine developers having multiple optimized render targets is actually a selling feature.

So all other IHV's should pursue their own IHV-specific graphics API because "targeting another renderer is not that much of an overhead" and is a "selling feature"??? I'm sorry, but an IHV-specific graphics API such as Mantle is hardly a good long-term solution, even if it being opportunistically promoted by some in the short-term.

 

[Davros]

so frostbite 3 supporting mantle, will that be all fb3 games or just games that start using fb3 after its been ungraded with mantle ?
the dude on facebook ( i assume works for dice) says our 15+ games (does that mean dice's games)

 

[Dave Baumann]

So all other IHV's should pursue their own IHV-specific graphics API because "targeting another renderer is not that much of an overhead" and is a "selling feature"??? I'm sorry, but an IHV-specific graphics API such as Mantle is hardly a good long-term solution, even if it being opportunistically promoted by some in the short-term.

Why is it such an issue? We are talking about 2, maybe 3 implementations. The situation is very different now from when DX was originally introduced where we had many, many IHV's jockeying for hardware position and a very high number of games being developed with their own unique rendering engine. Now we have a small number of IHV's and a small number of middleware rendering engines covering the vast majority of high end games. Games and game engines already frequently implement quite different rendering paths and options dependent on the hardware underneath even when they are using DX anyway, this is an extension of that.

Now, granted, we don't want a situation whereby the PC is fragmented and there is no option but to implement on separated IHV API's, that would be damaging for the developers that need to hit the lowest common denominator, hence the reason this is there as an option for developers that want to go that extra step for better optimization and/or features. The common API's are still critical and nobody has said they aren't.

 

[DieH@rd]

so frostbite 3 supporting mantle, will that be all fb3 games or just games that start using fb3 after its been ungraded with mantle ?
the dude on facebook ( i assume works for dice) says our 15+ games (does that mean dice's games)

#1 its twitter not facebook
#2 he is engine lead for Frostibite 3
#3 he meant all upcoming EA games based on FB3, not only DICE games


I wonder if that means that NFS Rivals will also get it...

 

[Rodéric]

Like Dave said, you want as "simple" API you use Direct3D or OpenGL, you want high performance low level API you use Mantle. Even if NV and Intel release theirs it won't be a problem, and if it's not too different from Mantle it will be even easier...

Why are people upset about a tool empowering developers ?

 

[jimbo75]

Why are people upset about a tool empowering developers ?

That's the easiest question to answer in this whole thread.

By the sounds of things BF4 is going to be a real showcase of what Mantle can do with specific hand-crafting and a bit of work put into it. I wonder if similar gains will carry over to the other games in the Frostbite engine if it's just using the "baked in" stuff.

 

[pjbliverpool]

#1 its twitter not facebook
#2 he is engine lead for Frostibite 3
#3 he meant all upcoming EA games based on FB3, not only DICE games


I wonder if that means that NFS Rivals will also get it...

I've only just realised that Johan is repi. Cool!

 

[xDxD]

Like Dave said, you want as "simple" API you use Direct3D or OpenGL, you want high performance low level API you use Mantle. Even if NV and Intel release theirs it won't be a problem, and if it's not too different from Mantle it will be even easier...

Why are people upset about a tool empowering developers ?

perhaps many versions means more time to develop a game/less optimization...here is a problem. I don't like mantle and and I would don't like a low level's Api by nvidia, i'd prefer an evolution of directx or Ogl.

The only thing I like of mantle is that perhaps those api will put pressure on Microsoft to improve directx

 

[Davros]

Not sure if I want a company that havnt mastered keyboard remapping to be programing without the safeguards of directx

 

[Ethatron]

I don't like mantle and and I would don't like a low level's Api by nvidia, i'd prefer an evolution of directx or Ogl.

<hypothetical>I make a game, have hours to burn, port to Mantle. You don't like my game I also make a port for intuition/graphics.library because I love it. You don't like my game </hypothetical>

 

[xDxD]

<hypothetical>I make a game, have hours to burn, port to Mantle. You don't like my game I also make a port for intuition/graphics.library because I love it. You don't like my game </hypothetical>

sorry?

 

[Ethatron]

You say "I don't like mantle", I want to understand what t's you don't like.

 

[willardjuice]

You say "I don't like mantle", I want to understand what t's you don't like.

The fear would be if I play game X (uses only D3D and Mantle) and game Y (uses only D3D and "Nvidia's Mantle"), I would essentially need to switch GPUs to ensure I get the "best experience" for each game. That's not fun as a consumer.

 

[xDxD]

The fear would be if I play game X (uses only D3D and Mantle) and game Y (uses only D3D and "Nvidia's Mantle"), I would essentially need to switch GPUs to ensure I get the "best experience" for each game. That's not fun as a consumer.

exactly, as a consumer and pc gamer i don't like this perspective.

 

[Malo]

Haven't Nvidia been trying to do this for years with "custom" AA, PhysX and CUDA in games? Oh look, all this extra fancy stuff BUT only if you're using one of our cards! I certainly haven't switched to Nvidia for years just for those features but I'm sure it's somewhat of a selling point for them. I definitely wouldn't see those benefits as essential, and to me that's what Mantle is about, benefits of working with an AMD card on Mantle supported games.

Out of all my friends with gaming PC's that I make for them and they use all the time for gaming, I'm the only one who knows anything about GPUs, APIs, differences between Nv/AMD from a gaming perspective, what antialiasing is (and even cares about it) etc. "Consumers" don't care, enthusiasts do.

 

[flopper]

The fear would be if I play game X (uses only D3D and Mantle) and game Y (uses only D3D and "Nvidia's Mantle"), I would essentially need to switch GPUs to ensure I get the "best experience" for each game. That's not fun as a consumer.

already true with other features today
you dont buy new cards beacuse of that do ya?
Mantle is a great thing and coming.