There is also the point of getting "the gaming api" out of MS hands .. the fact that carmack thinks it could be a big win for something like steamos says something... year of the linux desktop? Year of the linux gaming box .
http://www.alexstjohn.com/WP/2013/07/22/the-evolution-of-direct3d/
That's a post from one of the DX creators. Just look at what Microsoft wanted to get with the API model and look what AMD is proposing with Mantle. Go with Mantle and you will be stuck with AMD GNC for the rest of your life. Try Intel or Nvidia or Mali and you need a new code path. Heck, even a new AMD architecture that is not compatible with GNC will need a new code path.
That said, Mantle should be good for the Mcports we see today, but for the PC ecosystem it is not good at all.
There is already a crossplatform API between Windows and Linux. Its called OpenGL. And Valve is an open backer. And Mantle is not open.
Not to mention SteamOS will most likely have been out for a good year before even the first Mantle SDK and spec arrives.
Given that we don't have a firm release date for either of these things, and the first use of Mantle is shipping in December whereas the first use of SteamOS is shipping in 20XX, that last bit seems like your traditional AMD bashing.
I think, as I mentioned before, the goal would be to provide the spec and SDK publicly.
VRZ: Any timeline to Mantle’s SDK being made public?
It could be as early as sometime next year or maybe the year after.
The pc ecosystem with ms and Intel tax?
Well the death of it cant come soon enough. Or at least some serious compettion instead of this monopoly that stagnated completely 3 years ago.
Its simply not compettive anymore. Mantle is a brick in a much larger tendency of the market towards android and arm in a more and more mobile market and apple still blasting away from x86 as planned. Consoles with own os capturing the daily life and so on. And that bigger tendency is not going to stop. Its the future.
Mantle is the - THE - future api because of all that.
The consequences for gaming is far bigger than even charlie at sa writes. And bigger than amd wants to say because they are small and very fragile.
Instead of msft and Intel on the software side you will be locked to AMD GPU regardless of the OS/platform you choose? Is this supposed to be a good thing?
And when is SteamOS going to be made public?
Valve says "soon". And the first will start beta in november. So most likely Q1 2014. While Mantle is more between Q4 2014 and sometime unknown in 2015.
Instead of msft and Intel on the software side you will be locked to AMD GPU regardless of the OS/platform you choose? Is this supposed to be a good thing?
I'm not even talking about the fragmentation issue this is going to bring to the PC ecosystem, and how hard it will be to smaller developers to get multiple code branches for multiple low levels APIs.
Instead of msft and Intel on the software side you will be locked to AMD GPU regardless of the OS/platform you choose? Is this supposed to be a good thing?
I'm not even talking about the fragmentation issue this is going to bring to the PC ecosystem, and how hard it will be to smaller developers to get multiple code branches for multiple low levels APIs.
That's correct. But as the often cited Amdahl's law shows the influence of serial components for a hypothetical code (or code fragment), we should have a look at this serial component. Since there are multiple tasks to be done per frame in a game, these can more or less easily (depends on dependencies, no pun intended) be distributed to multiple threads. What's more difficult is bringing all that together. And here the only single immediate context available in DX11 (forcing serialism) might be a problem.Mantle may provide increased GPU performance but I doubt it will provide any real performance increases on the CPU side.
The only way to see true multithreading performance increasing is through the developers coding the games for more threads and also address multithread scalling as well. This may become a practical adoption by the developers of cross-platform titles since the PS4 and XBox1 both utilize poor single thread performance multi-core processors but the adaptation will not necessarily be something that is carries over to PC only titles since there would be no intrinsic benefit to such coding.
It doesn't look like Mantle vs. DX11 is about device agnostizism, but it could simply be the result of how multithreading was included into the API+driver architecture design.MultiThreading rules for Direct3D 11
Each ID3D11DeviceContext is singlethreaded, meaning that only one thread can call into a device context at a time. If multiple threads wish to access a single context, they must synchronize outside the API, using locks such as critical sections. This implies that a single command list can only be generated in a singlethreaded fashion (but multiple deferred contexts can be used simultaneously, each by a single thread, to simultaneously create multiple command lists). This also implies that it is illegal to play back two command lists simultaneously on the immediate context.
When the driver does not support multithreaded creates, the API will use synchronization internally. When the driver does not support command lists, the runtime provides software command lists. That is, you can use the multithreaded features of D3D11 without ever checking whether or not the driver supports the features.
DXGI shares the same thread resources as the D3D immediate context. It is illegal to use DXGI methods (other than AddRef, Release, QueryInterface) at the same time as the immediate context.
For applications with separate threads for processing command lists and displaying frames, there is the question of how to keep the GPU busy while still displaying frames in a timely fashion, since synchronized access to the immediate context is required for both tasks, and processing in general could take arbitrarily long. One possible solution is to use a separate D3D11 device for each thread, and to share textures to be displayed by creating them with the D3D11_RESOURCE_MISC_SHARED flag. In this scenario, ID3D11DeviceContext::Flush should be called on the processing thread to complete the execution of the command list prior to displaying the final texture in the display thread.
All architectures will be taken into consideration when hardware decsions for consoles are taken. There is no dismissals because of ideological reasons or otherwise.And i am pretty sure ARM is lined up as candidate for the next consoles. We will see what armv9 will bring.
All architectures will be taken into consideration when hardware decsions for consoles are taken. There is no dismissals because of ideological reasons or otherwise.
Simply when decisions for next gen consoles will be made all architecures that met 'borderline' requirements will be taken under consideration. If i.e. ARM, Power or any other architecture will be evaluated as better overall choice then x86 will be dumped, if x86 will be considered overally better then it will be used again. There is no sentiments there.
There is already a crossplatform API between Windows and Linux. Its called OpenGL. And Valve is an open backer. And Mantle is not open.
Not to mention SteamOS will most likely have been out for a good year before even the first Mantle SDK and spec arrives.
We've been told at the GPU14 Tech Day event that the Mantle API is open, so theoretically Nvidia could purpose the technology in their GPUs. It should also make cross-development between PC and console games a lot easier, and also more incredible for those with a high-performance AMD GPU.
Techspot confirmed that Mantle is open.
http://www.techspot.com/news/54134-...y-mantle-api-to-optimize-gpu-performance.html
There has been a lot of speculation regarding AMD's new Mantle low-level API, specifically regarding if it's open or not, meaning can Nvidia & Intel add support to Mantle if they wish ? the official answer from AMD's Robert Hallock, Enthusiast Graphics Marketing Manager is a definitive NO.
Tech Fan@ic . ‏
@Thracks is MANTLE open source ?
Details
Robert Hallock ‏
@GnrlKhalid No. It is an API for the industry-standard GCN Architecture and its specific ISA, done at the request of game developers.
So not only is MANTLE AMD specific, it's also architecture specific, so it's only compatible with Graphics Core Next (GCN) based products, that's the 7000 series and up for discrete graphics, Kaveri & later for APUs.
And AMD confirmed its not.
But isn't it kind of irrelevant anyway whether the code is "open" or not? In either case it locks you into GCN hardware.
Yes its a good thing. There is a natural monopoly tendency in the market. It cant be different. The nonsense about fragmentation is from a technical perspective. From a business perspective this is excact the oposite.
We then change one monopoly like situation to another monopoly like situation. The change is abrupt. But so is the solution. Mantle is simply a far better solution to the new situation that i describe above.
So with Mantle everything is locked to AMD GPU and even worse, locked to just one architecture of them. Everything different will need a new code path. Where's the good here?
The only case I see any good is in reducing work to port games to PC with AMD hardware, otherwise it's back to the 90's wild west, where every video card and sound card did have a specific code branch.
Maybe it is good for AMD as a company, as they get leverage over developers, but how does the consumer win with Mantle?
On the plus side, it makes me happy I got a 7770 instead of something last-gen like the 6850
Your second paragraph hits the nail on the head. It lets PC gamers (on GCN) benefit from low-level optimizations made for the consoles, it lets game devs squeeze higher performance onto the same PCs (and get a bucket o' money from AMD), and AMD sell more GPUs. It's a win-win situation, so long as the effort of porting from console to Mantle is pretty low.
On the plus side, it makes me happy I got a 7770 instead of something last-gen like the 6850
pretty sure the Xi3 piston people are less happy about it. they are stuck with trinity/richland for their 1st gen. the prospect of mantle level performance boost on steamOS later in the lifetime of the console would have really encouraged adoption. now they are stuck with vliw.