Upcoming graphics features "DX12" - what can we expect?

[3DVagabond]

It's not an insult to call it a niche. AMD only has 35% or so of the discrete GPU market. Of course it is an important update, especially for AMD, but DICE has bigger issues to deal with atm thanks to EA.

It's just a hunch I have because there has been no news of a release. You would think by now that DICE or AMD would have made an announcement. Better late than never though. BF4 can't afford anymore rushed releases.

Nonetheless, this is off topic for this thread.

Market share is irrelevant. AMD payed them a lot of money and they have an obligation. They can't simply decide it's not important enough.

 

[desprado]

Market share is irrelevant. AMD payed them a lot of money and they have an obligation. They can't simply decide it's not important enough.

Agree but if what mantle provide more bugs.Normally EA cant even solve main bugs and patches are waste btw last patch introduced Flickering for multi gpu user that is EA.U cannot trust them as well COD franchise.

 

[futurefields]

Pretty much nothing. I don't see how they can substantially improve DX at this point.

Even today games barely use most of DX11 features, so even if there are some new features in DX12, its going to be likely end of 2014 when its released or 2015 and then additional year till we see one or two games take advantage of it and then additional year or two before most games take advantage of any features it may have.

So, don't expect too much. DX11 for example didn't really bring much innovation, even though it was a decent leap from dx10.

They can certainly improve DX. It's whether or not it will actually be utilized in games. But they can certainly use it to sell new hardware and quasi-force people to upgrade to the next OS. Whatever the case may be, we can be sure that in due time they are going to need some more $$$ from us in order to survive.

 

[blackened23]

Market share is irrelevant. AMD payed them a lot of money and they have an obligation. They can't simply decide it's not important enough.

Erm, wait, so is Mantle BF4 being developed solely due to money? I thought it was because Johann Anderson really wanted a new API? I don't understand this viewpoint. The needs of the many outnumber the needs of the few. Hopefully Mantle BF4 can happen in December but when countless players are still crashing incessantly (on multiple platforms I might add), making the game stable takes priority, period. EA seems to have rushed BF4 out of the door for holiday sales and it has come to bite them in the rear.

 

[3DVagabond]

Erm, wait, so is Mantle BF4 being developed solely due to money? I thought it was because Johann Anderson really wanted a new API? I don't understand this viewpoint. The needs of the many outnumber the needs of the few. Hopefully Mantle BF4 can happen in December but when countless players are still crashing incessantly (on multiple platforms I might add), making the game stable takes priority, period. EA seems to have rushed BF4 out of the door for holiday sales and it has come to bite them in the rear.

Anyway, this is increasingly becoming a tangent so last post of mine about this.

Not even a good try. I never said that.

I don't know what to say though if you can't grasp the fact there's a business deal here and both sides are obligated to each other and EA/Dice can't simply decide they aren't going to honor their commitments.

 

[PrincessFrosty]

I think a better question at this stage is what specific requirements do we have that DX11.1 doesn't address?

It seems to me that the last generation of console games put mainstream PC game development in the ground, we literally have zero AAA PC games/engines which are designed for PC from the ground up and target the upper performance bracket, all thanks to the popularity of console causing lowest common denominator development.

The real issue is what do the consoles support hardware wise, that will be what decides what the next 8 years of gaming looks like, just like it did with the last 8, it's now too late to change it has become the norm. Even if we did have DX12 and loads of new features, which developers would even use them, maybe a handful would use it in the last few years of the console lifecycle.

The only way we'll get a DX12 that matters is if it's a DX12 that the console hardware is already capable of running.

 

[zlatan]

This is my wishlist:

- complete redesign with better abstraction layer
- lot faster and asynchronous dispatching
- very low draw call overhead
- re-usable command buffer that can be created on multiple threads
- precompile support for the shaders
- unlimited resources
- bindless textures
- some kind of ability to get direct/user mode access to GPU memory
- usable occlusion query

Also, I don't want to see driver threads and any hidden performance limitations. I want to see a modern API that basically don't do more than just executing shader programs.

If Microsoft provides a more or less direct copy of Xbox One’s API, than I will extremely happy.

 

[Pottuvoi]

The only way we'll get a DX12 that matters is if it's a DX12 that the console hardware is already capable of running.

We need a lot more than that.
What we do not need is DX to stagnate with a GCN level of programmability for the next 4-5 years. (GCN is already ~2 years old..)

I would certainly want to see proper upgrade to API so we might actually see an improvement on things during coming years.
If we do not see improvement on ROPs, Tex units etc.. or worse on API so developers cannot use those improvements we will be seeing some disappointed developers. (who have wanted some changes for over 10 years..)

 

[blackened23]

Not even a good try. I never said that.

I don't know what to say though if you can't grasp the fact there's a business deal here and both sides are obligated to each other and EA/Dice can't simply decide they aren't going to honor their commitments.

You appear to have two conflicting viewpoints which you did state. You've stated that Mantle was done solely because Andersonn wanted a new API, and now you're stating that there's a business deal forcing DICE to a Mantle obligation. That doesn't change the fact that Mantle takes a back seat to the bigger picture of ensuring the game runs well for everyone without crashes.

Also: nobody knows that there's a business deal concerning Mantle, at least in terms of a required release date. The 8 million $ AMD paid could be solely for the bundle provided with the 290X. Or to make the game AMD GE. It could be for GCN optimization. As far as releasing a Mantle version of the game? How do you know there's a business deal there dictating a release date?? Do you know something we don't? I haven't read anywhere that AMD dictated a release date for Mantle. Of course, it is entirely plausible that Mantle was done for the money and AMD did dictate a release date. But how would anyone know that - I haven't seen concrete news about that anywhere.