The AMD Mantle Thread

[krumme]

And support mantle that way...

 

[psoomah]

For example Plants vs. Zombies Garden Warfare is not a graphically intensive title, but it will support Mantle, just because Johan's like it. The situation is this simple.

And Johan's told that Mantle and PS4 (libGNM API) will drive their future Frostbite development. They don't care that Intel or NVIDIA don't support it.
If something is really good, and based on the Oxide demo I think Mantle is, than the market will buy the hardwares for it. For the peoples who don't like fragmentation ... well they can choose one of the two shiny new console.

Per DICE's APU13 slide 37, Mantle on Garden Warfare will 'Focus on APU Performance'.

It's a very clever way to get mainstream AMD GCN capable APU desktop and laptop owners into PC gaming with a game they are already very familiar with from Mobile and then upsell them on other PC Mantle games that will run on their computers. Should move a lot of Kaveri, Mullins and Beema hardware.

 

[3DVagabond]

What does this have to do with AMD's Mantle?
EA is being sued because they released a broken game, and they've stated they're going to halt all current projects including Mantle until they've fixed the game.

I haven't seen that referenced.

 

[3DVagabond]

I don't know if this is an entirely fair assumption. We don't know why Mantle is delayed. It can be speculated that DICE has diverted resources to fixing current problems in the DX version, but there is no certainty.

Do we know Mantle is delayed? I haven't seen any announcements.

 

[smackababy]

Do we know Mantle is delayed? I haven't seen any announcements.

Technically, we have a week before we pass their "target". For as excited as Johan seems to be about this, you'd expect to hear more though. Who knows.

 

[OCGuy]

nvm

 

[3DVagabond]

Technically, we have a week before we pass their "target". For as excited as Johan seems to be about this, you'd expect to hear more though. Who knows.

I don't disagree that it's getting down to the wire and we've heard nothing. It's not looking positive, for sure. I was just wondering if there had in fact been any announcement or if some people were just getting giddy over the prospect.

 

[144HzGamer]

My gut tells me that BF4's Mantle renderer/build is not delayed, if that was actually the case, Johan would have address it already IMO, but we've heard absolutely nothing from either camp (AMD or DICE).

:crosses fingers:

 

[BFG10K]

Unless nVidia adopts Mantle then I predict this will be about as successful as 3D vision or hardware PhysX. Any advantages it might have won't mean much if developers don't target it, or if it splinters PC gaming.

Right now it's looking like another Glide. DirectX may not be perfect but it's responsible or getting us out of the 1990s where every vendor from 3dfx to S3 to Rendition was pushing their own "pedal-to-the-metal, ours-is-better-and-more-efficient" API.

Nobody wants a different video card for each game they own.

I think the best way forward is for Microsoft to ship something like a DirectX "lite" which lets developers forgo some of the usual niceties in exchange for lower level access.

 

[SiliconWars]

What are the differences between then and now?

 

[OCGuy]

Unless nVidia adopts Mantle then I predict this will be about as successful as 3D vision or hardware PhysX. Any advantages it might have won't mean much if developers don't target it, or if it splinters PC gaming.

Right now it's looking like another Glide. DirectX may not be perfect but it's responsible or getting us out of the 1990s where every vendor from 3dfx to S3 to Rendition was pushing their own "pedal-to-the-metal, ours-is-better-and-more-efficient" API.

Nobody wants a different video card for each game they own.

I think the best way forward is for Microsoft to ship something like a DirectX "lite" which lets developers forgo some of the usual niceties in exchange for lower level access.

Judging by the lack of FPS performance numbers and statements by AMD brass regarding expectations, I think Mantle is a solution looking for a problem.

The reason the announcement even got legs at that terrible AMD event was because it was initially assumed that developers would naturally want to port to PC from a unified console API. Then we learn that Mantle isn't actually in consoles, as they already have their own "close to the metal" programming.

The fact is the PC market does not have to extend the life of their hardware as long as consoles do. Mantle will never make an unplayable game playable on a modern gaming computer with a discreet graphics solution.

It looks like Mantle isn't going to quite be "AMD's Iphone" like some predicted.

 

[AnandThenMan]

Some amazing predictions without even a single Mantle game being released yet.

 

[bystander36]

Some amazing predictions without even a single Mantle game being released yet.

That's what I'm thinking.

 

[BFG10K]

What are the differences between then and now?

Back then it was a fledgling industry that wasn't sure what it wanted to do. Glide was actually a very good API back in the day, especially given the terrible OpenGL/DirectX implementations at the time. But history has shown that unified standards are undoubtedly the way forward.

Imagine if instead of USB, PCIe, DDR, (etc) every motherboard vendor shipped their own connectors and sockets because they were "better". That certainly wouldn't benefit consumers.

DirectX is propriety, but it's also a unified standard on Microsoft systems. That means any hardware that implements a DirectX driver automatically generates potential benefit for customers.

AMD may say Mantle is "open" to the same degree, but the reality is that no vendor is going to support a competitor's technology. You need a neutral technology like DirectX or OpenGL to promote true competition.

 

[3DVagabond]

Unless nVidia adopts Mantle then I predict this will be about as successful as 3D vision or hardware PhysX. Any advantages it might have won't mean much if developers don't target it, or if it splinters PC gaming.

Right now it's looking like another Glide. DirectX may not be perfect but it's responsible or getting us out of the 1990s where every vendor from 3dfx to S3 to Rendition was pushing their own "pedal-to-the-metal, ours-is-better-and-more-efficient" API.

Nobody wants a different video card for each game they own.

I think the best way forward is for Microsoft to ship something like a DirectX "lite" which lets developers forgo some of the usual niceties in exchange for lower level access.

These guys said that they've been begging M$ for something like that and they simply won't do it.

 

[bystander36]

These guys said that they've been begging M$ for something like that and they simply won't do it.

Won't or can't? Remember, what ever they create, has to be compatible with at least 3 different architectures. The lower level you go, the more specific the hardware must be, or you have to write for 3 different architectures.

They may be able to give a little more control, but to stay compatible with 3+ archs, you can only go so far. The reality of it is they kind of have to be able to work with 10+ different archs, as they do need to be compatible with old architectures too.

 

[3DVagabond]

Won't or can't? Remember, what ever they create, has to be compatible with at least 3 different architectures. The lower level you go, the more specific the hardware must be, or you have to write for 3 different architectures.

They may be able to give a little more control, but to stay compatible with 3+ archs, you can only go so far. The reality of it is they kind of have to be able to work with 10+ different archs, as they do need to be compatible with old architectures too.

Seems like the biggest complaint is that DX decides too much, like what work goes to what threads, in what order, and supplies no feedback to the devs so they can code for it. There's likely other things, but that seems to be their #1 complaint. I don't know how opening up control over that would break compatibility.

Thing is, according to AMD M$ is fine with Mantle. You could be right and this is M$'s way out by allowing Mantle to supplant DX. I know it sounds crazy, but they don't seem to be too concerned about Mantle.

 

[Falafil]

Won't or can't? Remember, what ever they create, has to be compatible with at least 3 different architectures. The lower level you go, the more specific the hardware must be, or you have to write for 3 different architectures.

They may be able to give a little more control, but to stay compatible with 3+ archs, you can only go so far. The reality of it is they kind of have to be able to work with 10+ different archs, as they do need to be compatible with old architectures too.

That's why they said Mantle cannot work on Nvidia cards... oh wait

 

[krumme]

Seems like the biggest complaint is that DX decides too much, like what work goes to what threads, in what order, and supplies no feedback to the devs so they can code for it. There's likely other things, but that seems to be their #1 complaint. I don't know how opening up control over that would break compatibility.

Thing is, according to AMD M$ is fine with Mantle. You could be right and this is M$'s way out by allowing Mantle to supplant DX. I know it sounds crazy, but they don't seem to be too concerned about Mantle.

There is a reason ms dont react.
The reason beeing ms needs Mantle to keep users on the windows platform instead of going android. Windows and office is their cashcow on b2c side.
The threat for ms is not mantle beeing superior and excluding dx. Who cares? The potential problem is mantle makes arm transistion far more viable in the future. Its long term platform threat.

Mantle is extremely in Intel interest for the short and mid term. X86 winning consoles is really a huge win for them. Mantle is strengthening their platform. The problem beeing in the 3-5 year term they need support for their apu.

 

[Carfax83]

Won't or can't? Remember, what ever they create, has to be compatible with at least 3 different architectures. The lower level you go, the more specific the hardware must be, or you have to write for 3 different architectures.

It's neither won't nor can't. It's that they just couldn't be bothered. When you're in Microsoft's position and completely dominate the PC gaming landscape, there isn't much incentive to innovate.

With the introduction of Mantle however, we now see Microsoft all of a sudden becoming interested in porting the Xbox One's low level API to Windows.

If they do such a thing as planned, I doubt it will be as thin an API as Mantle, but it will be low level enough to stop devs from griping whilst still retaining enough abstraction necessary to create and run games on various hardware.

 

[krumme]

Try looking at some of the prior ports from consoles to pc or the other way. Often a third party specialized in porting is needed to get the job done. It takes huge ressources. Despite the efforts and cost the ports was often very bad. If we got the game portet at all.

Some people keep saying fragmentation. But pc gaming constitutes about 40% of the gaming market. Of the games sold here only a part is performance games. Perhaps we talk 20% performance games for pc while consoles is 50%. Calling fragmentation from pc side is narrowminded.

The devs needs incentives to port to pc or develop games for pc at all. For the future performance games the future apu of amd plus all dgpu is mantle enabled.

We have seen oxide make a mantle demo within 2 months without prior knowledge. Build into their engine solving a huge part of the batch and multicore cgallenge. Bf4 mantle comes aprox 2 months after start.

Calling this fragmentation is a reflection of a small pc world. From the devs viewpoint i would say the new gen consoles and mantle combined is far less fragmentation. People still think pc is the huge only important market it was in the 90 ties. The world have very much moved since then. Mantle could be what pc gaming needs to evolve again instead of the stagnation we have seen the last years for high performance games.

 

[bystander36]

You guys seem to think you know everything about how to code with these API and their internal workings. I don't know everything about the API's, but I've done my share of coding, even some x86 assembler coding back in the DOS days. Low level coding is very hardware dependent. Now when they talk about Mantle being low level, they probably mean "lower" level coding. How low the code goes, would determine just how hardware dependent Mantle is.

I doubt any of you know the specifics of coding with Mantle, and just how depending on GCN it is. AMD says it requires GCN. A few dev's made remarks that it could work for other venders. How true is it? Are they making assumptions? How much performance loss would there be if it did? Will it work on older hardware?

As of now, it is AMD only. That is about all we really know for now.

Dev's wanting Mantle on other venders is not a matter of DirectX failing us. It is simply a matter of Dev's wanting more control. Just because they want more control, does not mean it is practical, or possible on a single API.

 

[Lonbjerg]

I would hate the days of glide and different patches depending on what graphics card you had.



It seems a lot of people are far to young to remember the annoying diversions of that day.



DX fixed that...if AMD's plan is to return to those days...they have lost their marbles....been there, done that...not again.



DX saved us from API hell...people wanting to go back to times before that, really need to take a look at history again...

 

[ShintaiDK]

If DX is so terrible, and its all due to MS fault. Then why not use OpenGL? Or maybe the fault isnt MS, but rather the universal issue.

There is simply no easy way around a unified system.

Lonbjerg is right about the past, it was an anoying hell. And only 2 APIs emerged, DX and OpenGL. And OpenGL is barely used on the platform.

Also in terms of DX. Remember its much more than a graphics API, its also network, sound, input and so on.

 

[Falafil]

I would hate the days of glide and different patches depending on what graphics card you had.



It seems a lot of people are far to young to remember the annoying diversions of that day.



DX fixed that...if AMD's plan is to return to those days...they have lost their marbles....been there, done that...not again.



DX saved us from API hell...people wanting to go back to times before that, really need to take a look at history again...

I remember Glide. I loved it. I was disappointed to see it go.
Right now the purpose of DX is to kill PC gaming to turn everyone to consoles by making PC optimization crap and making game developers' jobs as difficult as possible. Mantle to the rescue!