[PCPER] 3DMark API Overhead Feature Test

[jpiniero]

Because it's incredibly risky/difficult. Only a limited number of people can write low-level code like that correctly and not shoot themselves in the foot in some manner. You basically need to be a guru-level programmer to properly handle D3D12.

Which is a huge problem, given that most developers want to spend the absolute minimum on the PC version. Plus, it requires Windows 10 and there's no way to forsee how well that will be adopted. OTOH, it could go the other way if it's easier to do DX12 only titles by porting the XBox One version.

 

[t1gran]

Because it's incredibly risky/difficult. Only a limited number of people can write low-level code like that correctly and not shoot themselves in the foot in some manner. You basically need to be a guru-level programmer to properly handle D3D12.

Which is a huge problem, given that most developers want to spend the absolute minimum on the PC version. Plus, it requires Windows 10 and there's no way to forsee how well that will be adopted. OTOH, it could go the other way if it's easier to do DX12 only titles by porting the XBox One version.

So it's not about more effective API - DX12/Mantle/Vulkan/Metal just give developer tools for more effective (low level) but hard programming? Than opposite question: why should developer use these incredibly risky/difficult tools for PC and mobile games? We know their prices are lower, so it doesn't seem reasonable from commercial point of view.

The only exception might be DX 12. If Windows 10 is a single OS for all platforms (PC, console, mobile), then there is no need for porting, is it? It also should be easier porting on Android, as Vulkan is going to be single API for desktop and mobile platforms, right?

Or perhaps it's just impossible to develop optimal game for specific OS - it can be done only for specific model of PC/gadget (when developer knows which CPU and GPU is used in it)? But it seems to be wrong guess - 3DMark API Overhead + DX 12 is effective on any top dGPU and iGPU.

 

[ShintaiDK]

This is why DX11.3 exist.

 

[ViRGE]

Which is a huge problem, given that most developers want to spend the absolute minimum on the PC version. Plus, it requires Windows 10 and there's no way to forsee how well that will be adopted. OTOH, it could go the other way if it's easier to do DX12 only titles by porting the XBox One version.

And that's the thing. A lot of these programmers are already writing low-level code for the consoles, so they know exactly what they're doing. D3D12 is not substantially different in that regard (the API calls may be different, but the conventions 0are the same), so you don't even need to go as far as port XB1 games.

So it's not about more effective API - DX12/Mantle/Vulkan/Metal just give developer tools for more effective (low level) but hard programming? Than opposite question: why should developer use these incredibly risky/difficult tools for PC and mobile games? We know their prices are lower, so it doesn't seem reasonable from commercial point of view.

Why use them? Because the performance is much faster and much more consistent. DX11 holds your hand; it abstracts you from the hardware. For lesser programmers this is a good thing. For guru programmers, they could be doing better if they had low-level access, and this is what DX12 provides.

Think of DX11 as safety scissors, and DX12 as really sharp large scissors. You can do a lot more with the latter, but if you run and aren't careful you're going to poke an eye out.

 

[Noctifer616]

So it's not about more effective API - DX12/Mantle/Vulkan/Metal just give developer tools for more effective (low level) but hard programming? Than opposite question: why should developer use these incredibly risky/difficult tools for PC and mobile games? We know their prices are lower, so it doesn't seem reasonable from commercial point of view.

They already to this with consoles, Mantle/DX12/Vulkan don't really change anything, they just bring it to the PC.

Also, those low level optimizations are not done on a game basis but engine basis, so not every dev out there has to actually work with those new API's, just the guys making the engines.

 

[t1gran]

Why use them? Because the performance is much faster and much more consistent. DX11 holds your hand; it abstracts you from the hardware. For lesser programmers this is a good thing. For guru programmers, they could be doing better if they had low-level access, and this is what DX12 provides.

Think of DX11 as safety scissors, and DX12 as really sharp large scissors. You can do a lot more with the latter, but if you run and aren't careful you're going to poke an eye out.

They already to this with consoles, Mantle/DX12/Vulkan don't really change anything, they just bring it to the PC.

As I said, developers may not be interested in making many efforts in respect of low-cost PC and mobile games.

Also, those low level optimizations are not done on a game basis but engine basis, so not every dev out there has to actually work with those new API's, just the guys making the engines.

You mean game developers get source code of engines and make some modifications in it?

So did I get it right:
1) There will be no porting need for games developed on Windows 10/DX 12 (as it will be single OS/API for both PC, console and mobile)?
2) Developers do not need to make low level optimizations (if they wish) for specific hardware (CPU and GPU) - they will need it for specific OS/API/engine only?

 

[loccothan]

I've got ~12.500.000 in Mantle 😀 Thats Great 11x faster than DX11_0
4.07GHz + 2.8NB + 2.55HT on 232FSB
XFX Black on 1060/1600 1.193v +12POW

But in BF4 , NFS Rivals, Alien Isolation + More (only AMD Game Evolved) we Have DX11_1 -> Working only on GCN.
Still DX11_1 is better than Old DX11_0