AMDs Mantle

[Athernar]

If Mantle does offer substantive performance benefits in addition to greater flexibility from exposing more GPU hardware functionality then NVidia will likely be forced into picking it up. I wouldn't be surprised if they pick it up in a limited fashion and try to push for some form of standardisation.

 

Really the only people that stand to lose here are Microsoft, a competing API that offers better perf/flexibility with the bonus of being cross-platform? Could be the end of DirectX as the #1 API.

[spaceelf]

people are crazy if they think NVidia will touch mantle.

So you consider major game developers crazy?  Fair enough.

[Thief000]

mantle isn't magic. NVidia will just roll their own. the way I see it playing out is, either Microsoft creates a low level api in directx,which interfaces with drivers, or a low level api which hooks into mantle and nvidias version.

Especially since the same performance can be achieved on their own hardware with OpenGL and hardware specific extensions. I said it before, Mantle will not achieve what AMD hopes it will.

[psionicinversion]

Especially since the same performance can be achieved on their own hardware with OpenGL and hardware specific extensions. I said it before, Mantle will not achieve what AMD hopes it will.

 

 i think it will i watched the 45 minute video and seems like it should be quite good. eliminates sli crossfire scaling problems cus you manage the queues to which cards yourself etc theres tons of stuff which you can do with it but its new and will take time to perfect

[Athernar]

Especially since the same performance can be achieved on their own hardware with OpenGL and hardware specific extensions. I said it before, Mantle will not achieve what AMD hopes it will.

 

They said they would support Mantle feature parity for OpenGL with extensions, that says nothing about performance.

[Thief000]

They said they would support Mantle feature parity for OpenGL with extensions, that says nothing about performance.

I'm talking about John Carmack's quote and I agree with him. We'll all have to see how it plays out, but I'm not betting on Mantle being the next big API, rather that its strenghts will be translated to the other API's sooner rather than later.

[Athernar]

I'm talking about John Carmack's quote and I agree with him. We'll all have to see how it plays out, but I'm not betting on Mantle being the next big API, rather that its strenghts will be translated to the other API's sooner rather than later.

 

I don't agree with Carmack here simply because with the performance claims being thrown around, OpenGL should have sparked an exodus away from Direct3D.

 

If the technical side isn't a problem, the lack of momentum behind OpenGL certainly is. If Mantle is the momentum needed to push a cross-platform API to the #1 spot, then that's a fair trade in my opinion.

[Thief000]

I don't agree with Carmack here simply because with the performance claims being thrown around, OpenGL should have sparked an exodus away from Direct3D.

 

If the technical side isn't a problem, the lack of momentum behind OpenGL certainly is. If Mantle is the momentum needed to push a cross-platform API to the #1 spot, then that's a fair trade in my opinion.

Yeah, but that's the thing, it isn't cross platform yet as it hasn't been adopted by nVidia and very likely won't if they can achieve what they claim with OpenGL.

It's kind of a chicken and the egg thing, but I don't see it happening for Mantle.

[Athernar]

Yeah, but that's the thing, it isn't cross platform yet as it hasn't been adopted by nVidia and very likely won't if they can achieve what they claim with OpenGL.

 

That's my point, if "they" could of done so with OpenGL then why hasn't anyone?

 

As I said, even if OpenGL is technically capable it seemingly lacks the momentum needed to do so. If Mantle does have that momentum then NVidia will be forced into supporting it.

[Thief000]

That's my point, if "they" could of done so with OpenGL then why hasn't anyone?

 

As I said, even if OpenGL is technically capable it seemingly lacks the momentum needed to do so. If Mantle does have that momentum then NVidia will be forced into supporting it.

Right now Mantle doesn't have any momentum unless more developers pick it up and VERY FAST. It's only got hype and one developer implementing it in engine. You're just jumping ahead of things.

But because there is a push for more draw calls, those functions will be adopted by the ARB faster than Mantle will ever gain traction among more developers and I'm betting MS is working on a similar lower level solution for DX as well (at least for the Xbox One for now). Competition is great and helps improve things, but it's not going to happen for Mantle, but we'll see how far they can get.

[spaceelf]

Right now Mantle doesn't have any momentum unless more developers pick it up and VERY FAST. It's only got hype and one developer implementing it in engine. You're just jumping ahead of things.

But because there is a push for more draw calls, those functions will be adopted by the ARB faster than Mantle will ever gain traction among more developers and I'm betting MS is working on a similar lower level solution for DX as well (at least for the Xbox One for now). Competition is great and helps improve things, but it's not going to happen for Mantle, but we'll see how far they can get.

15 Frostbite 3 games, Thief, Star Citizen, and Oxide Games upcoming engine are all confirmed.  Hardly one developer.

[chrisj1968]

15 Frostbite 3 games, Thief, Star Citizen, and Oxide Games upcoming engine are all confirmed.  Hardly one developer.

well said

[Athernar]

Right now Mantle doesn't have any momentum unless more developers pick it up and VERY FAST. It's only got hype and one developer implementing it in engine. You're just jumping ahead of things.

But because there is a push for more draw calls, those functions will be adopted by the ARB faster than Mantle will ever gain traction among more developers and I'm betting MS is working on a similar lower level solution for DX as well (at least for the Xbox One for now). Competition is great and helps improve things, but it's not going to happen for Mantle, but we'll see how far they can get.

 

15 Frostbite 3 games, Thief, Star Citizen, and Oxide Games upcoming engine are all confirmed.  Hardly one developer.

 

^ 'Nuff said pretty much.

 

You accuse me of jumping ahead of things, personally I think you've made your mind up already that it's going to be a failure regardless. (See how this works both ways?)

 

We've had years and multiple revisions of both DirectX and OpenGL since GPUs moved to general purpose designs over fixed function, and yet neither API vendor has taken the initiative to bring these performance levels to mainline. And yet we've seen squat.

[ViperAFK]

I guess it's more of a long-term investment on what forms the basis of next-gen consoles and their whole APU strategy. By the time Mantle gets any kind of significant adoption, the HD 6950 will be 4 years old; I'm not sure how it's in AMD's interest to pour money into new features for GPUs that are already a few generations behind, and I don't expect NVIDIA to do anything similar. It's not like they were selling anymore of these. That doesn't mean it wouldn't make sense for NVIDIA to support it in their actual and next-gen GPUs, with big names in the gaming industry like Johan Andersson asking for it and taking advantage of it on AMD's cards and APUs.

I can understand that, but it is still disappointing as an owner of one of those cards. Oh well...guess I'll just have to wait and see what happens with mantle.

[Noir Angel]

Then as far as I'm concerned, AMD's showboating about how mantle can be supported on anything rings very hollow, when they won't even support their own hardware properly. This is still a perfectly good and pretty powerful card, I'm quite disappointed if it won't be supported.

 

I own the same card, it's not a bad card so it would be disappointing for it not to be supported

 

As long as this is not a threat to DX I am fine with it existing.

 

If the performance numbers are anything to go by it will be.

[+Red King Subscriber²]

If the performance numbers are anything to go by it will be.

Hopefully DX will emerge victorious, just like with OGL vs. DX.

 

Needless fragmentation. Drivers and support are subpar as is, especially with AMD's support for "old" graphic cards.

[Athernar]

Hopefully DX will emerge victorious, just like with OGL vs. DX.

 

Needless fragmentation. Drivers and support are subpar as is, especially with AMD's support for "old" graphic cards.

 

As opposed to needless OS version restrictions, complete lack of cross-platform support and being tied to Microsoft?

 

I don't think you really understand the meaning of fragmentation either.

[+Red King Subscriber²]

As opposed to needless OS version restrictions, complete lack of cross-platform support and being tied to Microsoft?

 

I don't think you really understand the meaning of fragmentation either.

Wait, but that is the same as mantle then.

Anyhow, I don't think there are any DX forks so there is no fragmentation. Not sure about OGL.

 

It is just giving developers / GPU manufacturers more work - supporting three architectures.

[Athernar]

Wait, but that is the same as mantle then.

Anyhow, I don't think there are any DX forks so there is no fragmentation. Not sure about OGL.

 

It is just giving developers / GPU manufacturers more work - supporting three architectures.

 

It's not like Mantle at all, Mantle can run on NVidia hardware and non-Windows OSes. Meanwhile DirectX is still needlessly tied to various Windows versions in an attempt to try and drive OS sales.

 

OpenGL is a standard, standardisation as a process exists to combat fragmentation. So to bring it up in this context shows you don't really understand what fragmentation is, as arguably DirectX is/was by far the most fragmented API.

[Max Norris]

Meanwhile DirectX is still needlessly tied to various Windows versions in an attempt to try and drive OS sales.

It's just adding new features.. can you list any games that require DirectX 11.2?  I can't.  A handful are starting to make version 10 the bare minimum..IE, Vista.  The overwhelming majority still support XP's version 9.  Not exactly driving OS sales.

 

 

So to bring it up in this context shows you don't really understand what fragmentation is, as arguably DirectX is/was by far the most fragmented API.

It's no different than a game requiring a certain version of OpenGL.  How is one ok and one not?

[PGHammer]

Hopefully DX will emerge victorious, just like with OGL vs. DX.

 

Needless fragmentation. Drivers and support are subpar as is, especially with AMD's support for "old" graphic cards.

How old?

 

HD5xxx is supported via the current driver (same age as nV GT/GTS/GTX5xx, which is supported by the mainstream version of Forceware).  Nothing older (AMD HD4xxx OR nV GT/GTS/GTX4xx) is supported by either company in their mainstream drivers.

 

What's the real issue here?  You don't want to spend a lot of money, or you don't want to spend money at all?

 

If you own a portable PC (laptop, notebook, etc.) then you are screwed, plain and simple.  It's not a GPU issue, but a form-factor issue - portable GPUs are NOT (by and large) user-upgradable.

 

If you own a desktop, then you have options by the pound (if not the ton).

 

For AMD, you have HD5xxx and newer - all of which can use the current (as in mainstream) version of Catalyst Unified.  

 

nVidia has GT/GTS/GTX5xx and newer - all of which can use the current (mainstream) version of Forceware.  While GT/GTS/GTX6xx and newer can use more features in the current Forceware driver set, that does NOT mean that support for the 5xx series is lacking.  (I have a GTX550 Ti, and run the *current* version of Forceware - specifically, 331.65.)

 

I'm not campaigning for or against either nVidia or AMD - my position is, surprisingly, rather neutral, despite this being my first nVidia GPU for my own use.  (That is because I did not switch due to issues with Catalyst - AMD's mainstream graphics driver - in fact, I have recommended AMD *since* my crossgrade.  I went with nVidia this time strictly due to availability and price.  Not features, and not fanaticism.  Had I found an AMD GPU that met my criteria, I would have purchased it instead.  No regrets choosing nVidia this go-round, but no hatred of AMD, either.)

 

Therefore, I ask again - what's the real issue?

[Matthew S.]

Actually required an API version is hugely different from requiring a certain OS version...

[vcfan]

They said they would support Mantle feature parity for OpenGL with extensions, that says nothing about performance.

if mantle is supposed to support other types of cards, then there is going to be an abstraction layer. how is this different than what can be achieved with DX or OGL?

 

hardware->driver->HAL->mantle

hardware->driver->HAL->DX/OGL

 

in my opinion, if AMD wants to create a cross platform API, its a tough hill to climb. DX/OGL will just plug in the low level stuff,and these are already well established solutions. If they want to target only AMD cards, well the fragmentation will make this a failure. I just don't see how they are going to come out of this on top, I just don't. 

 

[PGHammer]

Actually required an API version is hugely different from requiring a certain OS version...

If you are requiring a specific API to be supported, then the issue is self-explanatory - you purchase hardware that supports that API.

 

Doesn't matter what API you are referring to (D3D, GLide, RRedline, etc.).

 

D3D was a most-agnostic-API approach in that it supported all the features of other APIs at a lowest-requirement basis (including OpenGL, RRedline, etc.) without using specific-to-individual-API calls - it was why you could use ATI (later AMD) 3D Rage chipsets and Voodoo I chipsets in tandem in D3D games (such as Forsaken, or even Hellbender).

 

The bugbear with GLide (and RRedline, and other brand-specific APIs, including console APIs prior to both the original Xbox and PS2) is that they were TOO specific - you basically got "hemmed in".

 

What Mantle has to overcome is that rather nasty tendency for any API that is supported by a hardware manufacturer to be tied TO that hardware manufacturer - either by real resistance from the manufacturer in question (AMD) or from the competition (nVidia, Intel, etc.).  Either possibility would eventually doom Mantle to failure as an API.

[Athernar]

It's just adding new features.. can you list any games that require DirectX 11.2?  I can't.  A handful are starting to make version 10 the bare minimum..IE, Vista.  The overwhelming majority still support XP's version 9.  Not exactly driving OS sales.

 

 

It's no different than a game requiring a certain version of OpenGL.  How is one ok and one not?

 

Paid OS-tied updates are the same as free open API spec releases? Riiiight.

 

if mantle is supposed to support other types of cards, then there is going to be an abstraction layer. how is this different than what can be achieved with DX or OGL?

 

hardware->driver->HAL->mantle

hardware->driver->HAL->DX/OGL

 

Simple, the "thickness" of the abstraction. In much the same way as low vs high level programming languages.