AMD Introduces Mantle At GPU 14

AMD is likely anticipating another 5-10 year console life cycle as well... And as much as we all hate to admit it, console gaming comprises the vast majority of the gaming ecosystem.

If they can stay with the GCN architecture for the same length of time, who really cares if the API is superceeded by something different in half a decade or longer. It allows some substantial benefits in the interim.

If this low level API approach proves to be a boon to developers, there's a good chance they'll land another contract moving forward... and I find it hard to believe that an API - about to be adopted by 90% of the gaming audience - won't have future parts with full backwards compatibility... at least until the whole discussion is relevant.

We will most likely see a dull in hardware expanse but see an equal improvement in performance because of Mantle. Depending on how you look at that, its both great and bad. Great in the respects it may lower hardware costs, bad in the fact we won't see major hardware improvements, but if the speed is there it shouldn't mean much.

Apparently the Developers have been begging for something like this for quite some time. I personally have not heard such rumblings prior to the release of mantle "AMD: WE WERE BEGGED FOR THIS!" but I can see where the developers wanted this (consoles).

AMD did mention when the original GCN was released that they expected this standard to be around for a while.
 
AMD's actual shader processor (SPU) hasn't changed much since the 4XXX series. The biggest thing is the memory addressing/fetching and shader scheduler. Mantle has to keep all of this into consideration. If AMD added in support for VLIW4 and prior set ups (which uses very similar SPU's granted) they would further convolute the code and make that low level coding a nightmare. They are smart to support their best and current layout with this, its also their most used. I have a feeling the next GCN core type will be somewhat backwards compatible in regards to coding.
What I'm thinking is that the 5000 and 6000 series aren't as different as AMD says. Maybe to the point where you could trick the drivers and that's all you need to do. But again, I'll wait for modding community to get their hands on these drivers.
 
I finally got around to reading AnandTech's article on what's been announced about Mantle so far. It's interesting for both the obvious reasons, lower overhead, and possible reasons for existing in the first place, taking advantage of console ports better than a port to (full) DX.

Carmack's comment brings up a good point too: MS and Sony may get frowny faced about helping console developers make better PC ports.

I worry that Mantle will be as fragile as AMD's last low level proprietary (GPGPU) APIs which constantly broke between drivers until it pretty much gave up on those in favor or standards based APIs. There are ways to help mitigate that, and maybe AMD is making that a top concern.
 
Carmack's comment brings up a good point too: MS and Sony may get frowny faced about helping console developers make better PC ports.

And in Microsoft's case, double frowny faced for both better PC ports and having the audacity to use something other than DirectX.
 
Carmack's comment brings up a good point too: MS and Sony may get frowny faced about helping console developers make better PC ports.
Not a big surprise really, but again if you're AMD and Nvidia you maybe less willing to help Microsoft and Sony. Remember AMD and Nvidia's core business is selling hardware, and while they have sold hardware to MS and Sony, it's at the cost of their core market the PC.

Part of the reason why is because Xbox 360 and PS3 are 7 year old tech. So hardware from 7 years ago can still be good enough to play modern PC games. For example, the GeForce 9600 is one of the most used graphic cards according to Steam survey, and that's a 2008 graphics card. The Radeon HD 5770 is the #1 AMD graphics card in the Steam Survey, and that's made 2009.

So if you were Nvidia and AMD right now, would you really care what MS and Sony are thinking? Their consoles took sales away from their core business, and they didn't make enough money from the PS3 and 360 hardware sales to justify the loss of sales on the PC end. So AMD is in a very unique position right now, where they totally own the console market, and Sony and MS are literally stuck with them. For AMD, the release of Mantle is trying to move the market back to PC, which is their core business.

Of course this sucks for Nvidia, but they won't just sit back and watch. They'll have to release their own low level API. I wouldn't be shocked if Sony did approach Nvidia and Nvidia asked for a huge amount for their GPUs. Cause if you're Nvidia right now, you would be pissed off. Nvidia is pissed off cause they can make Nforce chipsets for Intel. They can't make X86 CPUs like AMD can. By supplying graphics to PS3 they helped slow sales of PC graphic cards. With AMD using their console supremacy to create the Mantle API, which is something Nvidia can't do, you'd feel a bit butt hurt right now.

I worry that Mantle will be as fragile as AMD's last low level proprietary (GPGPU) APIs which constantly broke between drivers until it pretty much gave up on those in favor or standards based APIs. There are ways to help mitigate that, and maybe AMD is making that a top concern.

What previous API's has AMD made? As far as how stable Mantle is, there is literally a night and day difference between it and OpenGL and Direct 3D. The reason drivers are buggy, and why they are constantly improved, and why they're well kept secrets is because the amount of work that they need to do. The amount of optimizations and tweaks that go into drivers are massive, but not low level API's. Mantle is as direct to metal as it can get. No need for constant driver updates. It's going to be less buggy then OpenGL and Direct3D drivers.

It's not just the performance increase that Mantle brings.
 
So AMD is in a very unique position right now, where they totally own the console market, and Sony and MS are literally stuck with them. For AMD, the release of Mantle is trying to move the market back to PC, which is their core business.

I'm not as optimistic about the strength of AMD in these relationships. I think Sony probably doesn't care much, and MS would likely tolerate Mantle on Windows PCs if it contributes to lock-in; but on, say, OS X and Linux, MS would consider that unacceptable and seek to put the kabosh on that IMO. Especially any potential living room Steam Machine which would directly compete with XB1.

MS is in a far stronger position than AMD (financially and in the industry). They could really put the screws on AMD and make their lives really difficult. Or they could throw them a huge wad of cash.

I wouldn't think MS feels threatened about Mantle on Windows at all. JMO. I'd be surprised if they didn't bless this beforehand. I can't imagine this is a surprise to them.
 
What I'm thinking is that the 5000 and 6000 series aren't as different as AMD says. Maybe to the point where you could trick the drivers and that's all you need to do. But again, I'll wait for modding community to get their hands on these drivers.

Surprisingly the biggest thing between VLIW5 and 4 is the amount of low level tweaking AMD did on the core. Instead of being twins, where ones obese and the other is a cross country star, its more like twins where one is a math nerd and the other is the QB.

GCN is a whole new beast, to get Mantle to cover their older core tech, they would bloat it more then DX9c was.
 
I'm not as optimistic about the strength of AMD in these relationships. I think Sony probably doesn't care much, and MS would likely tolerate Mantle on Windows PCs if it contributes to lock-in; but on, say, OS X and Linux, MS would consider that unacceptable and seek to put the kabosh on that IMO. Especially any potential living room Steam Machine which would directly compete with XB1.

MS is in a far stronger position than AMD (financially and in the industry). They could really put the screws on AMD and make their lives really difficult. Or they could throw them a huge wad of cash.

I wouldn't think MS feels threatened about Mantle on Windows at all. JMO. I'd be surprised if they didn't bless this beforehand. I can't imagine this is a surprise to them.

MS would have to make up their minds then. Apple envy and appstore tunnelvision have them moving away from the traditional PC model--"PC's are dead" and "We're in the Post-PC era" and all that bullshit--and the fratboy marketing majors running MS these days believe their future is shitty smartphone apps and touch screens and hybrid laptop/tablets.

Hopefully Microsoft continues to lack the situational awareness or care to respond to these shifts happening like Mantle and SteamOS.
 
Hopefully Microsoft continues to lack the situational awareness or care to respond to these shifts happening like Mantle and SteamOS.

While SteamOS is a response to MS, I fail to see what Mantle has to do with it. If anything, it's a long overdue response to the righteous crying of developers who demanded low level access to the iron so they could actually optimize without having to deal with shitty drivers.

Good thing nvidia seems to react accordingly to everything that has been announced for the past few months. They have to.
 
While SteamOS is a response to MS, I fail to see what Mantle has to do with it. If anything, it's a long overdue response to the righteous crying of developers who demanded low level access to the iron so they could actually optimize without having to deal with shitty drivers.

Good thing nvidia seems to react accordingly to everything that has been announced for the past few months. They have to.

I agree, AMD has been in a constant battle to appease developers the past few years. They have tried, won a few and lost quite a ton. This is their blanket cover to appease developers, not current hardware contracts. If anything this should piss off MS a bit, but I doubt MS feels threatened as Mantle doesn't cover all of DirectX, just the draw/3d portion of it. Even then, MS might want that too as it will drop enough overhead from their own development.
 
I agree, AMD has been in a constant battle to appease developers the past few years. They have tried, won a few and lost quite a ton. This is their blanket cover to appease developers, not current hardware contracts. If anything this should piss off MS a bit, but I doubt MS feels threatened as Mantle doesn't cover all of DirectX, just the draw/3d portion of it. Even then, MS might want that too as it will drop enough overhead from their own development.

Isn't the Direct3D part of DirectX the main meat of it though and where most of the changes occur between versions. I mean the rest is DirectSound, etc. Given MS's current corporate policy of placing successive versions of DX behind a paywall (forced windows upgrade, i.e. DX9 XP, DX10 Vista, DX11 Win7, DX11.1 Win8, DX11.2 Win8.1, etc), then it would seem to me Mantle undermines that and would enable older versions of Windows to theoretically be able to run a latest/greatest renderer.

Its not like there isn't precedent for MS becoming irate when anything has challenged DirectX.
 
The only actively-developed portion of DirectX, besides the rendering math libraries (which pretty much just receive compatibility-breaking name changes), is Direct3D. The other APIs have either been deprecated or removed, save for the Xbox-style XInput and X3DAudio2 APIs. To my knowledge, those are no longer under active development.

One could say that DirectX at this point is pretty much just Direct3D. DirectInput is still fairly widely used, despite having been deprecated, however.
 
With EA fully on board with Mantle the other shoe had to eventually drop. Industry wide adoption of Mantle is inevitable, and I love it.

ghthM0i.jpg
 
With EA fully on board with Mantle the other shoe had to eventually drop. Industry wide adoption of Mantle is inevitable, and I love it.

ghthM0i.jpg

Sorry, anytime EA jumps in something with two feet like this it just means more money for them, shitter products for us.
 
Well, having a hard time using quotes for some reason today, but anyways...

I would suspect coding directly to hardware, variances between 5xxx, 6xxx, and 7xxx would be essentially coding 3 different paths. With scheduler tweaks alone between these cards, I can see the GPU could be waiting a lot of the time on older cards. The less HALs you have, the more you have to be precise in what you do.

It all reminds me of a time not-so-long-ago where programmers needed tricks to stretch hardware performance...

Some games which don't necessarily need maximum performance could stick with high level APIs. I can also see some developers making good conversion tools from mantle to OpenCL (for example) eventually knowing pretty precisely the performance losses in doing so.

Ie: A game with100 fps with Mantle code could equal to 30 fps in OpenCL code.

The big difference is OpenCL/DX are not any worst than before, in fact they are just the same. Mantle looks to be just THAT much more efficient, making the others potentially look quite bad due to so much system resources being "wasted" at the expense of easiness of coding.

If anything Mantle may push OpenCL/DX to streamline their own libraries for the sake of performance. To venture a guess, this may be a path competitors may push for?
 
The problem that will keep Mantle from becoming popular is that it won't work on Intel/NVIDIA systems which are the majority of the PC market. So that would mean a developer would have to code 2 versions of the game. Which unless AMD wants to pay for that, they won't do for free.
 
The problem that will keep Mantle from becoming popular is that it won't work on Intel/NVIDIA systems which are the majority of the PC market. So that would mean a developer would have to code 2 versions of the game. Which unless AMD wants to pay for that, they won't do for free.

Considering that Mantle is from PS4 and Xbone, the code is literally free. That doesn't necessarily mean it's not a problem for Intel and Nvidia or developers. Nvidia will have to create their own low level api, but I believe Nvidia will be crafty about it. Like maybe encapsulate it around OpenGL, which means a lot of it could be OpenGL but with some custom extensions that are low level for Nvidia hardware. Intel will likely stick with OpenGL and Direct 3D, at least until Larrabee.

There are 3 different developers that have different views of Mantle.

#1 PC Exclusive

These developers are the least willing to go for Mantle. Mainly because their games already work fine on Direct 3D/OpenGL, and see Mantle as extra work. Games like World of Warcraft, SWTOR, and League of Legends will probably never see Mantle support.

#2 PC and Console developers

Since these developers will have to work on Mantle anyway, they see Mantle as an added bonus for PC gamers. Especially since it can lower system requirements, which means it can increase their audience.

#3 Bad console ports


While these developers do make games for PC, it's an afterthought. The kind of games that run badly on modern PC hardware, and ask you to hit the 'B' button on your Xbox 360 controller, when there's no controller plugged in. They are literally direct ports, and you would think Mantle wouldn't be an issue, but Mantle won't be included because of the possibility it can cause technical support problems. Even though the code is there and essential free, they don't want to add more things that can go wrong to their game.
 
that is very interesting.
Im sick of MS/DX holding games/graphics hostage so to speak.
same here, i hope nvida does something. but something tells me they are pushing more towards mobile platforms.

Yeah, DirectX is a big barrier as an API and I'm happy to see a lot of Greenlight games and other applications use OpenGL or middleware APIs (SDL) with OpenGL to help to assist in being cross platform compatible as possible.

Last time I checked the next version of SDL had wrapper and implementation stuff for iOS, so... :)
 
I finally got around to reading AnandTech's article on what's been announced about Mantle so far. It's interesting for both the obvious reasons, lower overhead, and possible reasons for existing in the first place, taking advantage of console ports better than a port to (full) DX.

Carmack's comment brings up a good point too: MS and Sony may get frowny faced about helping console developers make better PC ports.

...

Yeah, but what about all the armchair industry analysts saying "PC is dead" and "Post-PC era" ? Shouldn't be a threat then, should it (assuming this is true)? :D :p
 
Isn't the Direct3D part of DirectX the main meat of it though and where most of the changes occur between versions. I mean the rest is DirectSound, etc. Given MS's current corporate policy of placing successive versions of DX behind a paywall (forced windows upgrade, i.e. DX9 XP, DX10 Vista, DX11 Win7, DX11.1 Win8, DX11.2 Win8.1, etc), then it would seem to me Mantle undermines that and would enable older versions of Windows to theoretically be able to run a latest/greatest renderer.

Its not like there isn't precedent for MS becoming irate when anything has challenged DirectX.

DX is hardly a paywall for MS. They supported Win98 with DX9. The only reason they didn't release it for Vista is because 1) outdated driver model and 2) it would cost too much money to even be worth it. 11.2 is Win8 for various reasons why DX10 was Vista. Vista is still fully compatible up to 11.1.

That's a lot of Direct3D support spread around their OS's.

The others provide easy venue's to access the device/driver/whatever, still mostly there, but like I said, I don't see MS waving DX around like a reason for upgrades. If anything, it would relieve a lot of stress from platform upgrades if they did decide to let it go.
 
It would do nothing other than shift the paywall over to AMD.
 
DX is hardly a paywall for MS. They supported Win98 with DX9. The only reason they didn't release it for Vista is because 1) outdated driver model and 2) it would cost too much money to even be worth it. 11.2 is Win8 for various reasons why DX10 was Vista. Vista is still fully compatible up to 11.1.

That's a lot of Direct3D support spread around their OS's.

The others provide easy venue's to access the device/driver/whatever, still mostly there, but like I said, I don't see MS waving DX around like a reason for upgrades. If anything, it would relieve a lot of stress from platform upgrades if they did decide to let it go.

So you're saying DX11.1 can install on Vista? That's news to me. In any case nobody said MS waves it around as a reason to upgrade. Its one of those subtle gotchas where backward compatibility is an afterthought -- obviously. Happy to derail further and get deeper into the history but would get pointless.

Suffice to say, tying renderer improvements to paid windows upgrades whether by accident or design is a model it would be nice to get away from.
 
I think you have a weird definition of backwards compatible DPI.

Most people consider things like USB, HDMI, etc... Backwards compatible, but that doesn't mean someone comes to your house and updates your TV everytime a new HDMI standard releases.

Backwards compatible means that the newer version can support the older (backwards compatible) Not the older versions support newer standards. I can certainly run DX 9 games on windows 8.1 and that is what backwards compatibility is.

I think you are looking for the magical forward compatible.
 
Back
Top