Friday, September 16th 2016

AMD Actively Promoting Vulkan Beyond GPUOpen

Vulkan, the new-generation cross-platform 3D graphics API governed by the people behind OpenGL, the Khronos Group, is gaining in relevance, with Google making it the primary 3D graphics API for Android. AMD said that it's actively promoting the API. Responding to a question by TechPowerUp in its recent Radeon Technology Group (RTG) first anniversary presser, its chief Raja Koduri agreed that the company is actively working with developers to add Vulkan to their productions, and optimize them for Radeon GPUs. This, we believe, could be due to one of many strategic reasons.

First, Vulkan works inherently better on AMD Graphics CoreNext GPU architecture because it's been largely derived from Mantle, a now defunct 3D graphics API by AMD that brings a lot of "close-to-metal" API features that make game consoles more performance-efficient, over to the PC ecosystem. The proof of this pudding is the AAA title and 2016 reboot of the iconic first-person shooter "Doom," in which Radeon GPUs get significant performance boosts switching from the default OpenGL renderer to Vulkan. These boosts aren't as pronounced on NVIDIA GPUs.
Second, and this could be a long shot, but the growing popularity of Vulkan could give AMD leverage over Microsoft to steer Direct3D development in areas that AMD GPUs are inherently good at - these include asynchronous compute, and tiled-resources (AMD GPUs benefit due to higher memory bandwidths). AMD has been engaging aggressively with game studios working on AAA games that use DirectX 12, and thus far AMD GPUs have been either gaining or sustaining performance better than NVIDIA GPUs, when switching from DirectX 11 fallbacks to DirectX 12 renderers.

AMD has already "opened" up much of its GPU IP to game developers through its GPUOpen initiative. Here, developers will find detailed technical resources on how to take advantage of not just AMD-specific GPU IP, but also some industry standards. Vulkan is among the richly differentiated resources AMD is giving away through the initiative.

Vulkan still has a long way to go before it becomes the primary API in AAA releases. To most gamers who don't tinker with advanced graphics settings, "Doom" still works on OpenGL. and "Talos Prinicple," works on Direct3D 11 by default, for example. It could be a while before a game runs on Vulkan out of the box, and the way its special interest group Khronos, and more importantly AMD, promote its use, not just during game development, but also long-term support, will have a lot to do with it. A lot will also depend on NVIDIA, which holds about 70% in PC discrete GPU market share, to support the API. Over-customizing Vulkan would send it the way of OpenGL. Too many vendor-specific extensions to keep up drove game developers to Direct3D in the first place.
Add your own comment

111 Comments on AMD Actively Promoting Vulkan Beyond GPUOpen

#1
RejZoR
There is a bit of a falacy there saying it inherently works better on AMD because it was largely derived from Mantle. Mostly because that's not true. It works better on AMD because AMD has been pursuing closer to the metal API for years. They built GCN architecture (hardware) around it and have been enhancing it for years. They made architecture that is inherently better at operating via such API's in general, be it Mantle, Vulkan or DX12. DirectX 12 has nothing to do with Mantle apart from core idea and behold, AMD again way better at it than NVIDIA.
Posted on Reply
#2
FordGT90Concept
"I go fast!1!11!1!"
I want a push to Vulkan for the simple reason it works on other operating systems too (Linux/Mac). That said, what I've read from developers is that making Vulkan games is much more work than DirectX 12 games because Vulkan does less for you. Developers have to do more of their own optimizations.


GCN cards get worse framerates on Vulkan in Tales Principle compared to D3D11.
Posted on Reply
#3
P4-630
For now I'm happily playing my games in DX11 windows 8.1.:)
Let's see what we have in 2 years from now....
Posted on Reply
#4
$ReaPeR$
well, they should! the important part here i think is the fact that google made it their default API for android.
Posted on Reply
#5
the54thvoid
I wrote a huge post and thought, "why bother? I'll just be seen as a fanboy." :roll:

Now no-one will know my mind mastery.
Posted on Reply
#6
The Quim Reaper
Extra performance in Vulcan for AMD says more about their very poor DX11 drivers than it says about Vulcan.
Posted on Reply
#7
LucidStrike
The Quim Reaper
Extra performance in Vulcan for AMD says more about their very poor DX11 drivers than it says about Vulcan.
What about OpenGL to Vulkan as with Doom, where the RX 480 pulls ahead of the GTX 1060 significantly?
Posted on Reply
#8
uuuaaaaaa
LucidStrike
What about OpenGL to Vulkan as with Doom, where the RX 480 pulls ahead of the GTX 1060 significantly?
It is the same, the problem with AMD's drivers as always been overhead, which prevents them from taking the best out of their gpu's. Amd usually needs much more Tflops per fps than nvidia. In low overhead api's like doom vulkan the cards fps rank roughly the same according to their compute power afaik.
Posted on Reply
#9
Ungari
P4-630
For now I'm happily playing my games in DX11 windows 8.1.:)
Let's see what we have in 2 years from now....
What Nvidia has done is ignore the new APIs until they become an issue for their customers.
Despite the planned road map for Volta in 2017 which will probably scale with DX12 and Vulkan, they released an unscheduled "new" architecture in Pascal, which is really Maxwell 3.0 that doesn't improve with these APIs.
Nvidia's philosophy is simply sell their customers a whole new architecture when the deficiencies become too problematic, making the previous generation obsolete in a very short time.
But as long as their loyal fans slavishly buy their product at their command, they will continue to be short-sighted about building their hardware for up-comming technical developments.
Posted on Reply
#10
qubit
Overclocked quantum bit
Let's hope AMD can give NVIDIA a good kicking with this and open up competition once more. I'm sick of paying inflated top of the range prices for NVIDIA cards and they're not even the top of the range.

A graphics API that works reliably and with high performance across operating systems, especially Windows and Linux would be fantastic, but I suspect that vested interests won't let that happen.
Posted on Reply
#11
phanbuey
Ungari
What Nvidia has done is ignore the new APIs until they become an issue for their customers.
Despite the planned road map for Volta in 2017 which will probably scale with DX12 and Vulkan, they released an unscheduled "new" architecture in Pascal, which is really Maxwell 3.0 that doesn't improve with these APIs.
Nvidia's philosophy is simply sell their customers a whole new architecture when the deficiencies become too problematic, making the previous generation obsolete in a very short time.
But as long as their loyal fans slavishly buy their product at their command, they will continue to be short-sighted about building their hardware for up-comming technical developments.
Nvidia's philosophy is "let's make the fastest card on the most popular APIs right now". And that's what they do. AMD simply can't do that so their PR department harps on whatever advantage they can spin and they do that over and over again.

They did the same thing with dx 11.1 and mantle and honestly it's just PR and false hope for their customers. By the time these architecture advantages become material we will already have two more arch releases.
Posted on Reply
#12
RejZoR
False hope. Gets 30fps more than DX11/OGL on same hardware... What kind of strong weed are you smoking?
Posted on Reply
#13
Nergal
Reason to buy 480 over 1060 increased. ;)

Thou, isn´t Android based on LINUX?
So would this choice not be the only logical one for google?
Posted on Reply
#14
I No
RejZoR
False hope. Gets 30fps more than crap support for DX11/OGL on same hardware... What kind of strong weed are you smoking?
There I fixed that for you. If AMD would get their drivers on par with DX11/OGL you wouldn't see that huge of "leap" in performance. Nobody said AMD's cards are bad the software that's behind them up is utter trash. If memory servers right this is just typical nVidia business DX11 was horrible for them while ATI had the crown. Within 2 gens tables turned. I'm all for rooting for the underdog but when you put jack into RnD you get jack back and this is what AMD is doing (because they can't support it financially). Furthermore why wouldn't they promote it? It's less work for them trying to patch drivers up.
Posted on Reply
#15
Ungari
This is where the fruits of indiscipline come into play.
As long as gamers are willing to get Win 10 in order to use DX 12, and as long as the majority of gamers are willing to pay the obscene prices of these Paxwell cards, nothing will change.
Posted on Reply
#16
Aquinus
Resident Wat-man
I No
There I fixed that for you. If AMD would get their drivers on par with DX11/OGL you wouldn't see that huge of "leap" in performance. Nobody said AMD's cards are bad the software that's behind them up is utter trash. If memory servers right this is just typical nVidia business DX11 was horrible for them while ATI had the crown. Within 2 gens tables turned. I'm all for rooting for the underdog but when you put jack into RnD you get jack back and this is what AMD is doing (because they can't support it financially). Furthermore why wouldn't they promote it? It's less work for them trying to patch drivers up.
nVidia does a lot of work in their driver to accelerate API calls in things like OpenGL and DX11 because the API worked in a way that was predictable enough to essentially queue up calls and execute them asyncronously on the CPU, transparently to the application. DX12 and Vulkan are sufficently complex and give enough control where doing this isn't likely to be feasible going forward because this very optimization technique that nVidia has been doing is now going to be left on the game developers to utilize by either the game engine or mechanisms exposed by Vulkan or DX12.

Simply put, AMD didn't make the same assumptions by how applications would be using these APIs (DX11 + OpenGL) and conformed to a more strict interpretation of these API specifications whereas nVidia implemented it in a way there the API conforms to the spec but, what is expected to happen when theses APIs are called might not be actually what's happening under the hood. The benefit of this is that nVidia's driver a lot of times can turn around and say, "Yup, I'm done doing what you told me to do," but, in reality, it might have just been put on a queue to be processed later. With DX12 and Vulkan, that kind of work is expected to be done by the game devs, not driver devs.
Posted on Reply
#17
bug
The Quim Reaper
Extra performance in Vulcan for AMD says more about their very poor DX11 drivers than it says about Vulcan.
Well, it's not really about poor drivers. It's more about Pentium4-style architecture: if the code isn't written in a certain way, it can't put all the hardware to good use (i.e. they can't put all those shaders to work using traditional APIs).
The real reason (well, one of them anyway) they're pushing Vulkan, is their drivers still suck at OpenGL.
But since Vulkan is cross-platform and assuming it's actually good, I have no problem with any company pushing for support. Because developers will need incentives if they're required to do more work now.
Posted on Reply
#18
Caring1
bug
The real reason (well, one of them anyway) they're pushing Vulkan, is their drivers still suck at OpenGL.
More like they realised that horse is dead and pushing it wont help.
Posted on Reply
#19
Bansaku
RejZoR
False hope. Gets 30fps more than DX11/OGL on same hardware... What kind of strong weed are you smoking?
30fps is roughly the difference for Doom 3 between DX11 and Vulcan with my HD7950 3GB. Max detail 1200P @ 88fps average. I'll take it!!

:toast:
Posted on Reply
#21
the54thvoid
Ungari
This is where the fruits of indiscipline come into play.
As long as gamers are willing to get Win 10 in order to use DX 12, and as long as the majority of gamers are willing to pay the obscene prices of these Paxwell cards, nothing will change.
Meanwhile, with even the latest and greatest AAA title shipping as DX11 and with DX12 support patched in, there is no rush to buy a DX12 card right now. Given the programmed visuals on DX11 are the same as DX12 in Deus Ex MD, why would anyone need to move to DX12? Nvidia isn't the bad guy here - they're simply profiteering because it has no peer right now.

I was looking to buy a Fury X a while back as a sidegrade project (with a 6700k) to replace my existing set up. Problem is that in the UK at least the Fury X is still retailing at or above £500. So I still have to pay that premium for a 'last gen' card when AMD have already launched their new gen Polaris.

You can always tell the balanced posters because they bitch about both sides being culpable for the current state of affairs. AMD are just as much the reason Nvidia do what they do as AMD haven't stepped up to the plate. Their long game is so long that people get bored waiting and move on to something new. But then there is the fact that AMD is so focused on providing for the new consoles that they neglect the PC side. It seems fairly obvious now that Polaris was co-designed on the PC side as it was being produced to go into the PS4 Pro.

So we have to wait for AMD to deliver Vega, likely in the same time frame the 1080ti will be out, perhaps even followed closely by Volta. And even if Vega is 50% better than Fury X (AMD's usual generational perf improvement) that won't really be enough to 'swing' the boat around. Sure, in selected titles but we really need AMD to produce something special. And then there is the stinger...

The Fury X is generally only about 20% faster than the 390X (therefore only 30% faster than the 290X).

The GCN changes haven't yielded significant generational improvements so Vega needs to really ramp it up. If Vega doesn't deliver, your pain will only continue.
Posted on Reply
#22
rtwjunkie
PC Gaming Enthusiast
Ungari
What Nvidia has done is ignore the new APIs until they become an issue for their customers.
No. What NVIDIA have been doing is because they can afford to, and that is make the best card for the current most popular API. Nothing wrong wiith that. They saw before Maxwell, and rightly predicted, that DX12 adoption would be super slow. And it has been. Why waste resources putting out products now that will do DX12 and perform worse on 11 if they don't have to?

As to Vulkan, apparently this situation doesn't apply as much to NVIDIA, since it works pretty well on the current gen.
Posted on Reply
#23
RejZoR
I No
There I fixed that for you. If AMD would get their drivers on par with DX11/OGL you wouldn't see that huge of "leap" in performance. Nobody said AMD's cards are bad the software that's behind them up is utter trash. If memory servers right this is just typical nVidia business DX11 was horrible for them while ATI had the crown. Within 2 gens tables turned. I'm all for rooting for the underdog but when you put jack into RnD you get jack back and this is what AMD is doing (because they can't support it financially). Furthermore why wouldn't they promote it? It's less work for them trying to patch drivers up.
I've had like 6 generations of Radeon cards and I don't know where people find all this crap support and awful performance on DX11 and OpenGL. I've never had performance issues with any Radeon and I've played UT99 via enhanced OGL renderer as well as all id Software games. Which were all OGL as well.
Posted on Reply
#24
Estaric
I havent played with vulcan yet, but im glad AMD are trying to adopt it from what i hear its simply amazing in performance. People who are hating on nvidia should stop because they are doing simply amazing at coming out with some of the FASTEST cards so what they dont get as big of a bump from Vulcan they still kick butt
Posted on Reply
#25
wiak
FordGT90Concept
I want a push to Vulkan for the simple reason it works on other operating systems too (Linux/Mac). That said, what I've read from developers is that making Vulkan games is much more work than DirectX 12 games because Vulkan does less for you. Developers have to do more of their own optimizations.


GCN cards get worse framerates on Vulkan in Tales Principle compared to D3D11.
ive followed vulkan a little bit and there is more ways to convert a dx11/opengl titles

* use opengl code and simply run it with a vulkan wrapper (this is probably what was done with talos)
* create native vulkan code and run it (this is probably what doom uses)

aka

* Talos runs like crap because it it uses opengl code behind it
* Doom runs like butter because its close to 100% vulkan code
Posted on Reply
Add your own comment