AMD: Mantle -> DX12 Porting Very Easy

[TeknoBug]

Knowing Microsoft, they'll pretty much copy'paste Mantle into DX12 with a few changed codes.

 

[PPB]

If AMD actually pull their finger out and port Mantle to Linux, this would be very interesting. Sounds like DX12 and Mantle are much closer than OpenGL and DX11.

More developers adopting Mantle, Mantle allowing Linux support, so when the dev wants a Linux version (for SteamOS) they just change a few lines of their DX12 version to make it Mantle API applicable.


One can only dream of such scenario, what will really happen is that while DX12 will be a carbon copy of Mantle in it's functionality, DX will just bloat the code to [deceive] the developers into thinking they will get an easy port, specially if AMD decides to go the Linux way with Mantle.

Profanity isn't allowed in the technical forums.
-- stahlhart

 

[3DVagabond]

Just because APIs are similar does not mean that porting is easy. I have used a tonne of APIs claiming this and I can tell you first hand that its complete and utter lies that porting from one to the other is easy. It never is. The printf/Println example from Java/C is far closer to the truth and the reality kicks in when you start embedding names where what looks very similar turns out completely different:

System.out.println("Hello " + name);

printf("Hello %s",name);

Made worse by the fact that the name in the c code has to be a null terminated String, in a language where there a lot of ways to represent a String, whereas in Java there is only one way (and its not null terminated). These sorts of on the surface similarities result in significant porting time.

Just look at the PS3 and PS4, they basically use OpenGL so you would expect porting to PC or from PC to PS3/PS4 was easy because the API was similar. Yet we haven't seen many OpenGL games at all. The Xbox 360 on the other hand runs something very similar to DirectX, infact its almost precisely DirectX. Yet we already know it can take 6 months or more for development teams to port. These are APIs with the same, exact same versions and just somewhat modified. The difference of DX12 to Mantle is going to be larger than these differences. Porting will not be easy and its not something developers are going to do unless they have to.

OpenGL and DX on consoles are not the same as they are on PC. I would even assume that they are heavily modified versions. Remember, on PC you can't execute draw calls from more than one core (something Mantle changes drastically), yet on consoles you can.

Also, keep in mind that AMD is also involved in the development of both Mantle and DirectX 12, they would know best how easy it is to port from one to the other.

But yes, despite all that, porting still takes time.

"AMD (or VESA, insert any AMD employee, insert Game dev who is actually working with the API) is lying." It's the same response you get all of the time. Pretty convenient way of dismissing anything.

 

[dacostafilipe]

Just because APIs are similar does not mean that porting is easy.

It's not only about the API calls, it's mostly because of the way you have to prepare your engine.

Mantle does have a more async way to do stuff, something that you do not have in OGL and DX<12. DX12 will also be more async, and having Mantle will certainly help out.

The same can be said about the command queues and the more simple resources types.

 

[Dribble]

Really i dont understand if DX12 will have better or equal performance of Mantle than why in the hell Developers will port Mantle and waste time and money i dont understand this logic of AMD.

They won't, unless AMD pays them which they won't as soon as DX12 gets close. Having served it's purpose (mostly marketing) mantle will quietly disappear over the next year or two, and a little while later AMD will quietly drop support and that will be that.

 

[zlatan]

Really i dont understand if DX12 will have better or equal performance of Mantle than why in the hell Developers will port Mantle and waste time and money i dont understand this logic of AMD.

D3D12 is a giant leap forward, but Mantle provide more functionality and less CPU overhead. So Mantle is a better API. It can be ported to Linux and Mac. This is a huge opportunity for developers who want to support Mac/Linux, but don't want to use OpenGL.

 

[OCGuy]

Who would "port" Mantle to DX12?

Mantle has never proven itself to "port" over easily from D3D_11, which was one of the claims made early on.

 

[Despoiler]

Really i dont understand if DX12 will have better or equal performance of Mantle than why in the hell Developers will port Mantle and waste time and money i dont understand this logic of AMD.

DX12 will actually be slower than Mantle. They are like 95% the same and the 5% that isn't is more abstraction in DX12.

Who would "port" Mantle to DX12?

Mantle has never proven itself to "port" over easily from D3D_11, which was one of the claims made early on.

No one made that claim. The claim made was porting from consoles would be relatively straight forward because they are using AMD hardware and are already programmed low level. XB1 uses DX11.x. Note the .x. It's not the same as DX11 we have on PCs.

 

[sandorski]

Who would "port" Mantle to DX12?

Mantle has never proven itself to "port" over easily from D3D_11, which was one of the claims made early on.

Really?

 

[VulgarDisplay]

Who would "port" Mantle to DX12?

Mantle has never proven itself to "port" over easily from D3D_11, which was one of the claims made early on.

Can you even tell the difference between something not being proven yet and something 100% not being possible in you dream world?

You are making it sound like mantle definitively does not port to dx11 easily which you have no proof of. The only people with any knowledge on the matter are under NDA.

 

[DiogoDX]

Contradicting our forum API experts Apple is too moving away from Opengl creating a low level API.:whiste:

http://www.polygon.com/2014/6/2/5773128/metal-ios-gaming-wwdc

Warning issued for callout.
-- stahlhart

 

[3DVagabond]

Who would "port" Mantle to DX12?

Mantle has never proven itself to "port" over easily from D3D_11, which was one of the claims made early on.

What direction are you working from? First you say "port Mantle to DX12", then to back it up, you say "port DX11 to Mantle".

 

[VulgarDisplay]

If amd can make mantle more enjoyable yo develop with than direct x 12 then it could conceivably take over the api world. As long as it can port over to dx12 but offers better and more options for developers they could end up developing for mantle and then porting to other platforms later.

Especially if mantle isn't tied to an operating system.

 

[DiogoDX]

Tim Sweeney (Epic) on Mantle: "It's the wrong direction for the industry to go"

And now they are supporting Apple new low level proprietary API Metal. I think Mantle is bad for Epic because not came from Nvidia.

 

[sontin]

DX12 will replace Mantle within a year.
Metal will be the number one graphics API on iOS.

 

[Erenhardt]

DX12 will replace Mantle within a year.
Metal will be the number one graphics API on iOS.

me(n)tal.... mantle...
I mean, its close There was a rumor that mantle will make its way to mobile, but no one expected the slight name change

also it seems you spelled "at least two years" wrong

 

[ocre]

me(n)tal.... mantle...
I mean, its close There was a rumor that mantle will make its way to mobile, but no one expected the slight name change

also it seems you spelled "at least two years" wrong

huh?

Apple is developing metal. Its a low level API specifically for their chips.

At best you could say that Apple stole the idea from AMD. I think they are scrambling to find a way to get more competitive as their marketshare continues to be threatened by Android.

It really is no secret that OpenGL is a mess. It really could be that apple seen there was an absolute need to address this on their platform.

there is so many angles to look at. Another is the closing of the gap from Android compared to Ios gaming. At one time apple was miles ahead. Perhaps, they feel like they need to do something fast and dramatic.

So all of these things could play a factor. Apple is not only a HW company, they are just as much a SW company. This is what separated them from most of their competitors. This can be used as a competitive advantage.

 

[Despoiler]

DX12 will replace Mantle within a year.

1.5 years. DX12 is a December 2015 target assuming no delays.

 

[VulgarDisplay]

1.5 years. DX12 is a December 2015 target assuming no delays.

I think its 1.5 years for an alpha version of dx12 right?

 

[Noctifer616]

I think its 1.5 years for an alpha version of dx12 right?

They said games with DX12 support coming holidays 2015. I would assume that would be the final version, it's in alpha/beta right now.

 

[TreVader]

huh?

Apple is developing metal. Its a low level API specifically for their chips.

At best you could say that Apple stole the idea from AMD. I think they are scrambling to find a way to get more competitive as their marketshare continues to be threatened by Android.

It really is no secret that OpenGL is a mess. It really could be that apple seen there was an absolute need to address this on their platform.

there is so many angles to look at. Another is the closing of the gap from Android compared to Ios gaming. At one time apple was miles ahead. Perhaps, they feel like they need to do something fast and dramatic.

So all of these things could play a factor. Apple is not only a HW company, they are just as much a SW company. This is what separated them from most of their competitors. This can be used as a competitive advantage.

Apple is still miles ahead in gaming, android is so fragmented there is almost no chance of them keeping up.


Mantle and Metal have nothing to do with each other besides both being low-level APIs. Until AMD gets into mobile in a big way there will be no need for mantle, and frankly I don't think android is even capable of implementing any API at all. They can barely get android to run on android.

Porting Mantle games to PC/DX will be the standard until people stop using nvidia.