DirectX and OpenGL to get low level access according to leaked statements (GDC 2014)

[hungtran]

I think the DirectX changes will also help Mantle improve faster due to the competition. For all we know, if there're enough similarities with XBox One's DX due to its using GCN chips, porting to Mantle from consoles will be that much more easier and widespread.

 

[VulgarDisplay]

Yea there is no way to say (scientifically, not saying you dont know what runs better on your system) that the API caused the difference and quantify it.

Trying to keep 120FPS going with all of the eye-candy off seems counter-intuitive to me personally.

There absolutely is a scientific way to do it. That 74fps is constant at those settings staring at that part of the map. More telling is that on dx11 I cannot keep 120fps pretty much anywhere on siege of shanghai in that area of the map. With mantle it is always locked to my frame cap.

Dropping IQ settings is something nearly all 120/144 hz monitor users and competitive players do. This gives them faster input response and better visibility. I do this for a competitive advantage over people. Mantle is a godsend for competitive players because they are far more likely to be cpu limited.

 

[PPB]

The settings that give the more competitive advantage in BF4, as it was also on BF3, are coincidentally the most demanding on the CPU (mesh quality on ultra) and the least taxing on the GPU (basically the rest of the settings either low or off).

Even if you cant stand the jagginess in the shadowmap resolution that lighting quality on low gives you (I know I cant) and you turn that setting to high/ultra, the game still is massively CPU bound in 64p multiplayer maps. With DX there wasnt really a way to go around that, unless you wanted to shell an expensive 6C/12T intel cpu and OC the heck out of it. Now with mantle you can literally underclock your +i5 grade cpu and still be GPU bound at those settings.

 

[omeds]

This is why I like Mantle too, it's of great benefit to 120/144hz users trying to keep frame rate at or above refresh rate. DX11.2 has a similar impact in BF4 on Nv hardware, though I believe Mantle surpasses it. Point is 120/144hz is often CPU limited on high end rigs and these DX/OGL updates, and Mantle, are a god send. It's been a long time coming.

 

[sontin]

AMD is going to try to get mantle into the OpenGL specification, or implement them as extensions to OpenGL in their driver. Microsoft will spend money on DirectX.

Why would anybody use those AMD specific extensions?

The biggest threat to Mantle is OpenGL. With a software and hardware independent API you can port your game nearly to every plattform on the market.

And dont forget: AMD hasn't released any SDK and further information about Mantle. On the other hand you can read a lot more about the new OpenGL 4.4 extensions.

 

[3DVagabond]

I can tell you what it isn't: spending millions on Mantle development with the intention of encouraging DX/OGL to make it even more irrelevant than it is now.

If their intention was to simply reduce CPU overhead to encourage Microsoft, they're now out of pocket for millions while their competitors benefit for free. You'd have to be an absolute moron to run a business like that.

Who do you think will adopt Mantle if these optimizations come to fruition?

You are only seeing the small picture. AMD pushes for open standards as their standard business practice. I think AMD would prefer being strictly a hardware company. They would rather have others pick up and continue the work.

 

[mindbomb]

If you think this is definitely going to end mantle, I think you don't appreciate the complexity of the situation. Firstly, opengl isn't relevant, so let's get that out of the way. With DX12, MS will historically force you to buy a whole new OS for the privilege of using a new API. You'll also have to buy a new gpu most likely, whereas mantle can work with existing hardware. DX12 doesn't seem like it would work on maxwell, so you would have to wait for a new nvidia architecture to be released before you could use it, but kaveri might retroactively support it due to similarities between it and the xbox one.

So you effectively end up with the same sort of dynamic, the low level api that only works on amd cards vs dx11 nvidia cards.

 

[omeds]

I would expect win8 support from any DX updates tbh, DX11.2 under 8.1 already exposes some benefits similar to Mantle, its quite possible an incremental update (11.3) could provide the extensions under existing OS. I'm unsure if it would require a hardware update though, but I would think not since its related to CPU overhead.

I can understand that many still use win7 so its no consolation, though.

 

[MutantGith]

If you think this is definitely going to end mantle, I think you don't appreciate the complexity of the situation. Firstly, opengl isn't relevant, so let's get that out of the way. With DX12, MS will historically force you to buy a whole new OS for the privilege of using a new API. You'll also have to buy a new gpu most likely, whereas mantle can work with existing hardware. DX12 doesn't seem like it would work on maxwell, so you would have to wait for a new nvidia architecture to be released before you could use it, but kaveri might retroactively support it due to similarities between it and the xbox one.

So you effectively end up with the same sort of dynamic, the low level api that only works on amd cards vs dx11 nvidia cards.

Wait...So now, DX 12 (not released in any way yet) won't work with Maxwell cards (again, not really released in any substantive way yet)?

And we're supposed to infer this how?

Microsoft is suddenly going to release a version of DX, whose entire point for existence was supposed to allow for as broad a range of hardware to be addressed as uniformly as possible, that will suddenly only work for AMD cards...but only because those cards are locked into the architecture of the APU in the XBOX - which will at that unspecified point in the future be several years old by then...Gah. The levels of cyclic logic and self-fulfilling marketing prediction is baffling.


How does that make any sense, for anyone involved? Prognostication is great and all, but sometimes, people announcing developmental talks and laying out initial frameworks for future software API optimizations are just that...there doesn't have to be winners and losers in everything.

 

[OCGuy]

There are no DX 11.2 games. Amazing how there can be so many misconceptions.

 

[omeds]

I was under the impression BF4 uses some DX11.2 features which are only exposed under win8.1 - if not then DX11.1.

 

[Silverforce11]

Intel stands to lose the most if bottleneck for gaming moves from the CPU to GPU. It's interesting they would agree here.

No that is just a small view, on the mobile space, this will be massive.

Intel dominates laptops, ultrabooks, pushing into tablets as well. These are where low powered CPU/APUs reside and its practically their own market.

Also, whatever makes a weak CPU faster will also make a strong CPU faster still. They have nothing to lose and everything to gain, let others do the handy work and they profit!

 

[Carfax83]

With DX12, MS will historically force you to buy a whole new OS for the privilege of using a new API.

I can't believe that after how many years, people still repeat this fallacy..

The reason why APIs sometimes require an OS upgrade, isn't because Microsoft is just trying to squeeze more money out of consumers..

It's because the new API comes with a driver model update as well. Vista had a completely new driver model that added a number of new features and enhancements that increased the stability, performance and capability of the display drivers over that of the XP version..

Could they have back ported it? Possibly, but it wouldn't have been worth the effort..

Same thing with Windows 7 and Windows 8/8.1. Windows 8 came with a new driver model

 

[TerryMathews]

I can't believe that after how many years, people still repeat this fallacy..

[snip]

Could they have back ported it? Possibly, but it wouldn't have been worth the effort..

You mean financially wouldn't have been worth the effort?

 

[Will Robinson]

There's no way(that I see)that MS can possibly build a new API with Mantle's granularity because they do not design or build the hardware.
Even if they do release a lower level DX edition it will still have to use some kind of abstraction layer to interface with the video card(and CPU if AMD)and therefore have less ability to fully exploit the architectures of either,compared to Mantle.
NV will release their own(Mantle) which will serve a similar purpose but still require abstraction.
Think NV is going to send Microsoft its schematics for their next gen hardware so they can write a really close to the metal DX?
How about Intel...think they'll hand over the latest Core design so MS can optimize "DX13" for it?
No way.

 

[cytg111]

I can't believe that after how many years, people still repeat this fallacy..

The reason why APIs sometimes require an OS upgrade, isn't because Microsoft is just trying to squeeze more money out of consumers..

It's because the new API comes with a driver model update as well. Vista had a completely new driver model that added a number of new features and enhancements that increased the stability, performance and capability of the display drivers over that of the XP version..

Could they have back ported it? Possibly, but it wouldn't have been worth the effort..

Same thing with Windows 7 and Windows 8/8.1. Windows 8 came with a new driver model

I can't believe that after how many years, people still repeat nonsense like this.
There is no where in hell that any directx version should/could not be running on windows-xp. It is just friggin software from one level to the next to the metal - and xp was pretty damn mature in the end. Fallacy? Repeat it a couple more times and act like its a fact.. common, you know you want to.

 

[Jodell88]

Why would anybody use those AMD specific extensions?

The same reason anyone will use a Nvidia specific extension? It offers something that they want.

The biggest threat to Mantle is OpenGL. With a software and hardware independent API you can port your game nearly to every plattform on the market.

OpenGL isn't a threat to mantle since AMD are going to and already have been developing performance orientated OpenGL extensions with the end goal of them being incorporated into the core specification. AMD has had at least one such extension incorporated into the OpenGL specification. I linked to it earlier in this thread.

 

[VulgarDisplay]

I can't believe that after how many years, people still repeat this fallacy..

The reason why APIs sometimes require an OS upgrade, isn't because Microsoft is just trying to squeeze more money out of consumers..

It's because the new API comes with a driver model update as well. Vista had a completely new driver model that added a number of new features and enhancements that increased the stability, performance and capability of the display drivers over that of the XP version..

Could they have back ported it? Possibly, but it wouldn't have been worth the effort..

Same thing with Windows 7 and Windows 8/8.1. Windows 8 came with a new driver model

Since you are obviously a software engineer with extensive Operating System coding experience can you please explain why different driver models can not run on variations of the same OS Family.

I mean there really must be a good reason since Windows 2000, XP, Vista, 7, and 8 are all part of the same Windows NT family of operating systems.

It's obviously not because MS wants to force upgrades is it?

 

[OCGuy]

Until someone can directly answer how something with less far less than 1/3rd market penetration (not all AMD cards support mantle) is going to "take out" D3D, which has 100% market penetration and is used as the API in Xbone, then this conversation is useless.

 

[Gloomy]

Until someone can directly answer how something with less far less than 1/3rd market penetration (not all AMD cards support mantle) is going to "take out" D3D, which has 100% market penetration and is used as the API in Xbone, then this conversation is useless.

Mantle has more market penetration than DX12 does

 

[Cloudfire777]

It was extremely obvious that Microsoft would update DX to include similar optimizations. I have said that many times, and was one of the reasons why I said Mantle would fail.

There is NO WAY a new API would take over for DirectX which have the gaming industry in their hand. And any sane person know that developers also prefer to have 1 API to create their games with because they want to ship out the games to the market the quickest.

Mantle was not an attempt to move the APIs forward although some of you think they made it in the kindness of their heart. DICE wanted a new API, AMD came onboard because they saw opportunity to make money on it because they get to market Mantle with GPUs. Not to mention it would help their lacky APUs foward to remove some of the CPU bottleneck where they falled behind Intel.

AMD is nothing but depressed now that Microsoft acted so quick on the matter. And they should be, not just because they don`t have Mantle to lure people in with, but because DirectX is universal while Mantle only works on GCN GPUs. DirectX is not brand specific. It does not try to favour any system.

Like it should be.

Mantle was not an attempt to move the software development forward. It was an attempt to make money on APIs. And Microsoft will call their bet. While the rest of the gaming developers and majority of system builders will embrace Microsoft in support that Microsoft beats AMD out of the API tournament.

 

[Carfax83]

You mean financially wouldn't have been worth the effort?

Both financially, and practically. Changing driver models, especially to the degree that was done between XP and Vista, is not trivial I'm sure, and would have required a major rewrite of the OS's code.

In instances like that, I think it's better to start from scratch and do it properly for the greatest impact. Microsoft has shown that it's willing to break compatibility between DirectX versions to move the technology forward, and DirectX sometimes requires a new driver model to add or expose new features.

DX9 to DX10 was such a paradigm shift..

 

[TerryMathews]

Both financially, and practically. Changing driver models, especially to the degree that was done between XP and Vista, is not trivial I'm sure, and would have required a major rewrite of the OS's code.

In instances like that, I think it's better to start from scratch and do it properly for the greatest impact. Microsoft has shown that it's willing to break compatibility between DirectX versions to move the technology forward, and DirectX sometimes requires a new driver model to add or expose new features.

DX9 to DX10 was such a paradigm shift..

Would it be more work than say building in a system to disable a computer if the serial matches a black list? And then delivering that system via the System Update process to live installations?

 

[SiliconWars]

There is no way that the current DX with extensions will ever be able to match Mantle's true CPU multithreading. It will always fall down in certain scenarios.

There is no way that DX will be able to fully support both Nvidia's and AMD's entire capability, especially going forward. To get the full support of your hardware you need to write your own API - that's the way it's going to be. Do you really think Microsoft will continue to update DX every time one of the vendors adds new stuff?

Anything that makes DX and OpenGL better for CPU multithreading simply helps AMD against Intel - that's why they are happy to back these initiatives now.

Mantle is a long term play for AMD. It will continue to be updated alongside the hardware - and more importantly it allows AMD to make new hardware improvements that won't be pointless under DX.

You're forgetting something very important also - Mantle is AMD's guarantee against Nvidia's meddling and any dodgy future GameWorks initiatives. They are not dropping Mantle - far from it.

 

[OCGuy]

Mantle has more market penetration than DX12 does