Shake up the whole industry. Give an ability to the devs to use GCN much better than what is possible with D3D11 and OGL. GCN is incredibly limited with the legacy APIs, but with the new ones, you will see (this week) what is possible with a currently available hardware.
ShintaiDK
Lifer
- Apr 22, 2012
- 20,378
- 146
- 106
Money, quite simple. You cant pay for something you cant afford. Their DX12 drivers are already behind and need more focus. So its obvious to move the resources.
One of many reasons:
Try to interface with and/or bugfix your own code vs. code that you didn't write. And, more importantly, give hotline support.
It's a different way to improve developer relationships.
So, you need a new low level api for it instead of OpenGL extentions?
nVidia is even able to make their new hardware features available through NVAPI.
BTW: What would happen to AMD if they are much slower than nVidia when somebody is using a low level api for pushing the boundries - like the Star Swarm demo? 50% performance advantage for Kepler and Maxwell is much more than AMD gets with Mantle over nVidia with DX11... :|
AnandThenMan
Diamond Member
- Nov 11, 2004
- 3,991
- 627
- 126
destrekor
Lifer
- Nov 18, 2005
- 28,799
- 359
- 126
Mantle is dead. Long live Vulkan.
Mantle was destined to die from the beginning. Giving it away to become incorporated in Vulkan is a big advertisement and public-image booster, and as others alluded, it helps AMD's flailing CPU division somewhat.
Make no mistake, I could care less about beating the wardrums of AMD or Nvidia. But it's good for all of us to have a healthy AMD. Heck, perhaps it was good to have an AMD cornered by ravenous beasts, because the gaming community at large is getting Vulkan out of the deal.
As was said, obviously Mantle cannot be 100% Mantle on all platforms and architectures. Mantle was coded for GCN. If it is to work on Nvidia, Intel, PowerVR, etc etc etc, then certain features of Mantle had to be abstracted or outright removed.
It's clearly the only way you can tackle that objective. But the meat of what mattered in Mantle appears to be fully featured in Vulkan.
As others have clarified, this makes Vulkan a fork of Mantle as opposed to a clone. As I explained, an outright clone was impossible, and AMD basically gave it away with the full intent and assumption it would be "forked." AMD knew it could be improved upon even if it were only intended for GCN, that was why it was still in active development. Now they start from the point where it is a universal API, and will continue to be improved upon.
I don't care if AMD gets any promotion out of this. Vulkan is good for us. Plain and simple. Mantle ran well, and so to will Vulkan. And Vulkan will be everywhere unlike Mantle by the very nature of its core.
This is a good thing, all around.
It's a DX and OpenGL market, round 2.
I wonder if this iteration of "OpenGL", now known as Vulkan, will have a stronger PC marketshare than past versions of OpenGL. It was never a very successful API on the Windows platform, because DirectX tended to perform better and have more features. One might say DirectX 12 might still be a touch better, but that will take time to truly determine. But if Vulkan takes over for Linux, OS X, Android, perhaps even iOS even though Metal is there to compete, that definitely means there will be at least SOME Vulkan-specific games available on Windows.
I doubt developers like DICE will continue a dual-API approach, so I wonder if they will dedicate resources to Vulkan as a cross-platform development goal, or if they will focus on DX12. I may be wrong and they will do both, but that seems silly. Adding Mantle right now is a bonus because it is lower-level than DX11, but of course it is not universal so they cannot ever scrap DX11. But now if Vulkan is universal, I'll be curious to see where this goes.
For sure, a few EA studios (DICE and Visceral) now have Mantle experience, as does Crytek. Frostbite may not even be used outside of EA, but CryEngine does get used outside of Crytek. It will never be as popular as Unreal Engine, but I wonder if any of this experience will translate towards a desire to focus development with only Vulkan.
Mantle was destined to die from the beginning. Giving it away to become incorporated in Vulkan is a big advertisement and public-image booster, and as others alluded, it helps AMD's flailing CPU division somewhat.
Make no mistake, I could care less about beating the wardrums of AMD or Nvidia. But it's good for all of us to have a healthy AMD. Heck, perhaps it was good to have an AMD cornered by ravenous beasts, because the gaming community at large is getting Vulkan out of the deal.
As was said, obviously Mantle cannot be 100% Mantle on all platforms and architectures. Mantle was coded for GCN. If it is to work on Nvidia, Intel, PowerVR, etc etc etc, then certain features of Mantle had to be abstracted or outright removed.
It's clearly the only way you can tackle that objective. But the meat of what mattered in Mantle appears to be fully featured in Vulkan.
As others have clarified, this makes Vulkan a fork of Mantle as opposed to a clone. As I explained, an outright clone was impossible, and AMD basically gave it away with the full intent and assumption it would be "forked." AMD knew it could be improved upon even if it were only intended for GCN, that was why it was still in active development. Now they start from the point where it is a universal API, and will continue to be improved upon.
I don't care if AMD gets any promotion out of this. Vulkan is good for us. Plain and simple. Mantle ran well, and so to will Vulkan. And Vulkan will be everywhere unlike Mantle by the very nature of its core.
This is a good thing, all around.
It's a DX and OpenGL market, round 2.
I wonder if this iteration of "OpenGL", now known as Vulkan, will have a stronger PC marketshare than past versions of OpenGL. It was never a very successful API on the Windows platform, because DirectX tended to perform better and have more features. One might say DirectX 12 might still be a touch better, but that will take time to truly determine. But if Vulkan takes over for Linux, OS X, Android, perhaps even iOS even though Metal is there to compete, that definitely means there will be at least SOME Vulkan-specific games available on Windows.
I doubt developers like DICE will continue a dual-API approach, so I wonder if they will dedicate resources to Vulkan as a cross-platform development goal, or if they will focus on DX12. I may be wrong and they will do both, but that seems silly. Adding Mantle right now is a bonus because it is lower-level than DX11, but of course it is not universal so they cannot ever scrap DX11. But now if Vulkan is universal, I'll be curious to see where this goes.
For sure, a few EA studios (DICE and Visceral) now have Mantle experience, as does Crytek. Frostbite may not even be used outside of EA, but CryEngine does get used outside of Crytek. It will never be as popular as Unreal Engine, but I wonder if any of this experience will translate towards a desire to focus development with only Vulkan.
Crytek has Mantle experience? AFAIK they havent announced the full integration of Mantle in the Cryengine...
AMD has a lot of libraries (R-SDK, ADL,AGL, ...) for accessing special hardware features. But this has nothing to do with the graphics API.
NTMBK
Lifer
- Nov 14, 2011
- 10,525
- 6,050
- 136
They're using it as a tool to grow the total GPU market, whereas NVidia uses its extensions as a weapon to claim more share of that market.
AnandThenMan
Diamond Member
- Nov 11, 2004
- 3,991
- 627
- 126
And I think we know which approach works out better from a financial point of view. Don't get me wrong I think it's incredible what Mantle has done...for the industry, but for AMD that remains to be seen.
Not sure why anyone is complaining. Mantle gave gamers a preview of what can be done with a more focused API. DX12 looks to use some of Mantle's features. Everyone will benefit in DX12 in part because of Mantle. I haven't seen so much interest in API's on this forum until Mantle came along. It would have been nice to see it mature, but once DX12 looked to share much of the benefits Mantle brought, I think it was likely to turn out this way.
Phynaz
Lifer
- Mar 13, 2006
- 10,140
- 819
- 126
Actually it's just the opposite, they are adding a HLSL to OpenGL
From Ars:
http://arstechnica.com/gadgets/2015/03/khronos-unveils-vulkan-opengl-built-for-modern-systems/
Last edited:
http://community.amd.com/community/...blog/2015/03/03/one-of-mantles-futures-vulkan
Official statement from AMD. Vulkan is the future of Mantle.
Official statement from AMD. Vulkan is the future of Mantle.
You mean more the end, huh?
How can Vulkan be the future of Mantle when the evolvement of Vulkan has nothing to do with the future of Mantle...
/edit: Reading more and more it gets obvious that DX12 has nothing to do with Mantle and Khronos took Mantle to speed up the process. Otherwise they would be more than a year beyond Microsoft.
How can Vulkan be the future of Mantle when the evolvement of Vulkan has nothing to do with the future of Mantle...
/edit: Reading more and more it gets obvious that DX12 has nothing to do with Mantle and Khronos took Mantle to speed up the process. Otherwise they would be more than a year beyond Microsoft.
Last edited:
Vulkan is a fork of Mantle. They use the same code base in some parts. It's just like how Windows was the future of DOS...
http://www.pcworld.com/article/2891672/amds-mantle-10-is-dead-long-live-directx.html
Gamers should be thanking AMD instead of trying to stoke a flame war. You're getting a shiny new and more advanced API that will benefit all gamers with current cards. To me this isn't Mantle dying so much as it succeeded in getting a more efficient API in the works, which seemed to always be the goal. Just because it'll be DX12 instead of Mantle doesn't take away from the fact that we're all benefiting from AMD and MS's work here (as well as Intel and Nvidia, I'm sure).
Honestly, it's a good thing for all PC gamers as a whole, but probably not a great thing for AMD going forward. There's a part of me that's a bit sad they didn't develop Mantle further specifically as an AMD only feature. It was one that would have set them apart at least from their competitor. That being said, I do find it a bit troubling that you'll see posters come in here cheering the death of Mantle rocking 980's in their sigs. Will they be so happy when those same mid-range/high end cards start selling for $100-200 more with no price drops in sight if AMD disappears.
Dx12 and new opengl does the same for amd as mantle does except without the cost. For pc and mobile platforms. It just couldn't be better from amd perspective. Such a gigantic change in api tech fitted to gcn and more cores, within one and a half year - for such a small investment !
Its only doable because EA is the real driving force and ofcource the consoles as a platform.
Its only doable because EA is the real driving force and ofcource the consoles as a platform.
Last edited:
destrekor
Lifer
- Nov 18, 2005
- 28,799
- 359
- 126
Well AMD had to of known in the long-term Mantle could not be sustained as an AMD-only initiative. They learned a lot working on the Xbox One with Microsoft and found a way to bring back a low-level vendor-specific API. This has, really, always been possible, and used to be the norm. It wasn't on anybody's radar to develop such because vendor-specific APIs are a bad thing for the market in the long run.
I fully believe AMD had no intention of keeping Mantle to themselves. It removes a little bit of a unique performance lead for their products only, but they also recognize they had a small marketshare and couldn't afford to be pushy with developers the same way Nvidia can. When previous manufacturers had such a lead, it was the same way.
A Mantle that remained indefinitely as a vendor-specific API would not be popular for long these days, because while it brought huge benefits to the market, it helped splinter developer resources and if it was truly successful as a vendor-specific API, it could splinter game compatibility and risk going back to where huge performance advantages (and disadvantages) with major IQ differences would be the norm, and determined based solely on which product you chose. You get that in a very very limited sense these days, with a library of additional features for one camp or the other, but not at the API level.
If Mantle was very successful, and failed to spur Microsoft to develop a similar vendor-neutral version, I don't doubt that in time we would have seen Nvidia try to push something of their own.
Of course, all of what happened is very much the natural result of Mantle coming onto the scene. Perhaps the glNext initiative would not have occurred in some reality, but one way or the other, somebody, be it the Khronos Group or Microsoft, would invest time and resources to develop a similar Mantle-like vendor neutral API. And, as we see, that is exactly what happened.
Except in this reality, AMD saw how beneficial it would be to actively help get high-quality vendor-neutral APIs on to the scene, two of them, in fact. AMD contributed to Direct3D 12, as did Nvidia and a host of other parties. Likewise with Vulkan, all of the major players participated and helped shape the API.
I'm not arguing that all of the Nvidia fanboys (again, count me as one if I must be labeled, I'm still worried about not having Nvidia for GameWorks games and wanted the 980 in SLI had it not cost so much) should bow down and praise AMD like it was a god, but we humans should be mature enough to recognize good things when and where they happen, no matter who provides.
Nvidia is just as much to thank for some DX12 features, I'm sure. And it is likely not using Mantle code whatsoever but there is assuredly AMD touches that produce the same effect.
Whatever one believes Vulkan is, and regardless if one recognizes it as a proper successor to Mantle, it doesn't matter. The fact of the matter is Vulkan and DX12 are almost here, and AMD directly contributed and pushed these results. This is good for everyone.
AnandThenMan
Diamond Member
- Nov 11, 2004
- 3,991
- 627
- 126
Given that DX12 is also essentially the best parts of Mantle, the wide reach Mantle ended up having on the industry is absolutely amazing. I sincerely hope it helps AMD sell hardware they need all the help they can get, we need a 50/50 GPU market.
Gloomy
Golden Member
- Oct 12, 2010
- 1,469
- 21
- 81
Here's how Nvidia's Vice President tells the story:
Just in case Sontin, Shintai, BFG10K et al weren't salty enough over being wrong. :awe:
Seems like they tore Mantle down to the command buffer and explicit memory management concept (which to be fair, is probably most of what makes Mantle, Mantle... I mean, that's the whole idea right?) and then stacked their own shader language on top. The new shading language seems pretty awesome too.
Here's a blog post from Christopher Riccio about the SPIR-V: http://www.g-truc.net/post-0714.html#menu
It seems to have advantages over everything else on the market right now! Cool stuff. :thumbsup:
Hopefully they didn't chop Mantle down so far GCN can't do all that cool GCN stuff zlatan was talking about.
http://www.pcper.com/news/General-Tech/GDC-15-Khronos-Acknowledges-Mantles-Start-Vulkan
Just in case Sontin, Shintai, BFG10K et al weren't salty enough over being wrong. :awe:
Seems like they tore Mantle down to the command buffer and explicit memory management concept (which to be fair, is probably most of what makes Mantle, Mantle... I mean, that's the whole idea right?) and then stacked their own shader language on top. The new shading language seems pretty awesome too.
Here's a blog post from Christopher Riccio about the SPIR-V: http://www.g-truc.net/post-0714.html#menu
It seems to have advantages over everything else on the market right now! Cool stuff. :thumbsup:
Hopefully they didn't chop Mantle down so far GCN can't do all that cool GCN stuff zlatan was talking about.
Indeed, people are missing the point. This is a win for AMD. They have managed to get the two industry standard to build their APIs using Mantle as a foundation. That puts them in a good position on the CPU side and the GPU side, they won't have to spend resources developing Mantle further, because MS & Kronos will take care of that!
But the good thing is, this is a win for gamers, NV users too. Smooth frame times, great SLI scaling (that's smooth), very high minimum fps (the most important out of the min/avg/max metric), great CPU scaling.
There's zero logical reason to be a hater.
Red Hawk
Diamond Member
- Jan 1, 2011
- 3,266
- 169
- 106
I think people should keep in mind that Mantle has, all along, been AMD giving developers what they want. I remember an interview with someone at AMD (I believe it was Richard Huddy) back in 2010 or 2009. He said that one of the requests they get most often from developers is "Make the API go away!" Developers wanted more direct control of a computer's resources, and they felt that DirectX was getting in the way. I'm pretty sure that that pressure from developers is what motivated AMD to see the appeal of closer-to-hardware development. Mantle and DirectX 12 are still APIs, but, as we've seen, there's much closer control possible through them than DirectX 11.
Mantle was not about coming up with a bullet point to sell more AMD graphics cards. It was about working with developers to find a way to make technically better and more efficient games on PC. Better and more efficient games on PC itself generates more interest in PC graphics cards. They did get Mantle to a working state with support in several games, and with the advent of DirectX 12 and Vulkan, I say, mission accomplished.
Mantle was not about coming up with a bullet point to sell more AMD graphics cards. It was about working with developers to find a way to make technically better and more efficient games on PC. Better and more efficient games on PC itself generates more interest in PC graphics cards. They did get Mantle to a working state with support in several games, and with the advent of DirectX 12 and Vulkan, I say, mission accomplished.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 24K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter